Modélisation de l'écoulement diphasique dans les injecteurs Diesel

Heterogeneous cavitation at the entry cornerCavitation developmentPocket breakingPocket advectionFigure 5: Chronological pocket breaking. Flow fromleft to right.that this area depends on the needle position duringthe injection. This leads to a further problem,i.e. the impact of the upstream conditions on thedevelopment of the flow in the nozzle. As a matterof fact, real nozzles are multi-holed and asymmetrical.Soteriou et al. [12] have shown that upstreamconditions are really important consideringthe development of cavitation, and some factors canperturb the flow in the nozzle. In [12] it is shownthat the form which the cavitation takes dependson Reynolds number of the flow. Arcoumanis etal. [10] have visualized the flow in ×20 scaled 6-holed injectors. They have seen string cavitation inthe sac, connecting several holes together (see figure6). This phenomenon takes place in an extremelytransient and intermittent way. It seems obviousFigure 6: String cavitation in the sac [1 0].that the bubble foam due to string cavitation influencesthe flow in the nozzle. Sac cavitation hasbeen also seen by K im et al. [17] in ×10 scaled injectors,and by Eifler [8] in real-sized injectors. Itis certainly due to the high velocities present nearthe needle, and the needle vibration. Badock [6]has confirmed this by seeing sac cavitation duringthe needle opening and closing. In the same way,Soteriou et al. [12] have remarked that turbulencecaused by the flow in the sac improves the reattachmentof the boundary layer in the nozzle. Date et al.[18] have seen a secondary flow in the nozzle, due tothe boundary layer reattachment in the nozzle’s 3Dgeometry, which explains the spray visualizationsof Soteriou et al. [12], showing the spray directeddownwards at the nozzle exit. Furthermore, cavitationsignificantly increases turbulence in the nozzle[12], but this turbulence seems to be qualitativelyand quantitatively different from that generated aftera liquid-filled recirculation bubble, as Ruiz andHe [3] have shown in their experiments under quasicavitatingconditions. This tends to prove that furtherstudies have to be performed in order to modelturbulent internal cavitating flows, as conventionalturbulence models do not seem adapted to this configuration.In the case of Diesel injectors, the velocities encounteredare so great that the compressibility ofthe fluid can be conceivable : Some experimentalresults have shown that flows in venturis can bechocked [19]. By the same way, flows in injectorscan be chocked [12]. A one-dimensional analyticalstudy of the flow [20] can be used to understand thechocking condition of the flow. The sound velocityin the flow, and the vapour fraction value can leadto Mach number values about 1 [21, 22]. In thiscase, compressibility of the liquid and vapour is tobe considered. In the same way, such catastrophicphenomena like collapse can cause huge fluid velocities[4, 23, 24, 25].All those experimental results lead to a conclusionwith regards to cavitation modelling in the injector:• Cavitating flows in injectors are characterizedby a huge shear stress.• Fluid velocities in the nozzle are close to thetwo-phase flow sound velocity.• Cavitation develops as pocket, and not as bubblesas usually seen in large-scale flows.• Cavitation is very unstable, transient and intermittent.• Further experimental studies have to be performedin order to characterize the turbulencein the nozzle.• Cavitation is influenced by the upstream flowin the sac.• The flow is tridimensional in the sac and theorifice.Impact of nozzle flow on atomizationUp to now, spray has been considered as a resultof aerodynamical interation between the liquid jetissued from the nozzle and the gas contained in thecombustion chamber. But several studies tend to