Le développement actuel des moteurs doit répondre à une volonté de réduction de la consommation et à des normes de plus en plus sévères. Les moteurs Diesel, bien que performants, présentent des émissions trop importantes d'oxydes d'azote (NOx) et de particules de suies. Ces émissions polluantes proviennent essentiellement de l'existence de zones riches en carburant ne permettant pas une combustion optimale. La préparation du mélange air / carburant, fortement conditionné par l'entraînement d'air par le spray est donc essentielle. La Vélocimétrie par Images de Particules sur traceurs Fluorescents, associée à un traitement spécifique des champs de vitesses instantanés de la phase gazeuse externe sont utilisés afin d'obtenir des mesures en proche frontière du spray. Dans la zone quasi statique du spray, un effet important de la densité ambiante sur le taux de mélange est mis en évidence. D'autre part, la diminution du diamètre de trou d'injecteur améliore sensiblement le taux de mélange local. Pour compléter cette analyse, un gain du taux de mélange entre les deux injecteurs est calculé pour chaque densité ambiante. Pour les fortes densités, le gain issu de la réduction du diamètre de trou peut at! teindre une valeur importante (> 60%). La phase gazeuse latérale dans la zone instationnaire du spray est ensuite investiguée et la méthodologie de la F-PIV est transposée à cette zone. Un lien entre la longueur de pénétration du spray et des longueurs d'entraînement est défini. La densité ambiante n'a pas d'effet sur ces longueurs d'entraînement. La diminution du diamètre de trou engendre une augmentation du gaz entraîné à une distance plus courte du nez de l'injecteur. Cependant, cette étude montre que les deux sprays n'engendrent pas un entraînement de gaz similaire. / The actual development of the engine must reply to a will of fuel consumption reduction and to stricter norms concerning the pollutant emissions. Although the Diesel engines are competitive, the NOx and soot particle emissions mainly come from the existence of wealthy fuel zone preventing an optimal combustion. Therefore, the air / fuel mixing preparation, highly controlled by the air entrainment in spray, is essential. Particle Image Velocimetry on fluorescent tracers, associated with a specific processing of the instantaneous velocity fields have been applied to obtain measurements in the near vicinity of the spray edge. In the "quasi- teady" region of the spray, the important effect of the ambient density on the mixing rate has been pointed out. On the other hand, an orifice diameter decrease significantly improves the local air / fuel ratio. To complete this analyse, a gain in mixing rate between two different injectors has! been calculated for each gas density. For high densities, the gain due to a reduction of the hole diameter can reach important value (> 60%). Then, lateral gaseous phase in no stationary zone of spray is studying and F-PIV method is transposed to this zone. A link between the penetration length and entrainment lengths is defined. Ambient density has not effect on entrainment lengths. The hole diameter decrease generates entrained gas increase at shorter distance of injector nozzle. However, this study shows two sprays do not generated the same gas entrainment

Undesired air entrainment in bottom outlet conduits of dams may cause pressure transients, leading to conduit vibrations, blowback, discharge pulsation and even cavitation, and jeopardize the operational safety. Due to design limitations or construction costs, it is impossible to create an air free environment in a pressurized pipe. Therefore, it is essential to understand the air transport in enclosed pipes in order to provide guidance in bottom outlet design and operation. The commonly used criterion of the air-pocket movement in pipe flow is the water flow velocity for starting moving an air pocket, the so-called critical velocity. In this thesis, the classical Volume of Fluid (VOF) model combined with the k-ε turbulence model is adopted for the computation of the critical velocity of a 150-mm pipe. The computed critical velocities are compared with the experimental results. The governing parameters investigated in this study include pipe slope and diameter, wall shear stress and air-pocket volume. Meanwhile, the carrying capacity (air-pocket velocity/ flow velocity) at all pipe slopes are analyzed. The simulation results of air pockets with different volumes in the bottom outlet conduit of Letten Dam in Sweden are presented in this study. Moreover, experimental study was conducted to measure the critical velocity for a 240-mm Plexiglas pipe. The results are in agreement with the experiments performed by HR Wallingford (HRW) in 2003 in terms of the effects of pipe slope and air-pocket volume; however, the critical Froude pipe number is slightly smaller in this study. In rough pipes, a larger critical velocity is required compared with that in the smooth pipe. The removal mechanism in the rough pipe involves the successive loss of air caused by turbulence. This explains that the air-pocket size, with the dimensionless air-pocket volume n < 0.015, has little impact on the critical velocity for the rough pipe. In addition, roughness has little impact on the air-pocket velocity when it moves upstream in the downward inclined pipe. The trapped air bubbles most likely remain permanently in the rough pipe. / QC 20120110