No abstract

No abstract

Two different implementation techniques of wall functions for cell-vertex based numerical methods are described and evaluated. The underlying wall model is based on the classical theory of the turbulent boundary layer. The present work focuses on the integration of this wall-model in a cell-vertex solver for large eddy simulations and its implications when applied to complex geometries, in particular domains with sudden expansions (more generally in presence of sharp edges). At corner nodes, the conjugation of law of the wall models using slip velocities on walls and of the cell-vertex approach leads to difficulties. Therefore, an alternative F. Jaegle · O. Cabrit · S. Mendez CERFACS, 42 Av. Gaspard Coriolis, 246 Flow Turbulence Combust (2010) 85:245-272implementation of wall functions is introduced, which uses a no-slip condition at the wall. Both implementation methods are compared in a turbulent periodic channel flow, representing a typical validation case. The case of an injector for aero-engines is presented as an example for an industrial-scale application with a complex geometry.

Abstract. Discontinuous Galerkin methods have become a powerful tool for approximating the solution of compressible flow problems. Their direct use for two-phase flow problems with phase transformation is not straightforward because this type of flows requires a detailed tracking of the phase front. We consider the fronts in this contribution as sharp interfaces and propose a novel multiscale approach. It combines an efficient high-order Discontinuous Galerkin solver for the computation in the bulk phases on the macro-scale with the use of a generalized Riemann solver on the micro-scale. The Riemann solver takes into account the effects of moderate surface tension via the curvature of the sharp interface as well as phase transformation. First numerical experiments in three space dimensions underline the overall performance of the method.

Large-Eddy Simulations (LES) of an evaporating two-phase flow in an experimental burner are performed using two different solvers, CDP from CTR-Stanford and AVBP from CERFACS, on the same grid and for the same operating conditions. Results are evaluated by comparison with experimental data. The CDP code uses a Lagrangian particle tracking method (EL) while the code AVBP can be coupled either with a mesoscopic Eulerian approach (EE) or with a Lagrangian method (EL). After a validation of the purely gaseous flow in the burner, liquid-phase dynamics, droplet dispersion and fuel evaporation are qualitatively and quantitatively evaluated for three twophase flow simulations. They are respectively referred as: CDP-EL, AVBP-EE and AVBP-EL. The results of the three simulations show reasonable agreement with experiments for the two-phase flow case. RésuméSimulations eulériennes et lagrangiennes aux grandeséchelles d'unécoulement diphasiqueévaporant Les simulations aux grandeséchelles (SGE) de l'écoulement diphasiqueévaporant dans un brûleur expérimental sont réalisées avec deux codes numériques différents, CDP du CTR-Stanford et AVBP du CERFACS, sur le même maillage et pour les mêmes points de fonctionnement. Les résultats obtenus sont validés par comparaison avec des données expérimentales. Le code CDP peutêtre coupléà une méthode de suivi Lagrangien de la phase liquide (EL). Le code AVBP peut soitêtre coupléà une méthode mésoscopique Eulérienne (EE), soità une méthode de suivi Lagrangien (EL). Après validation de l'écoulement purement gazeux dans le brûleur, la dynamique de la phase liquide, la dispersion et l'évaporation du carburant sontévaluées qualitativement et quantitativement pour trois simulations diphasiques dénotées respectivement : CDP-EL, AVBP-EE et AVBP-EL. Les résultats obtenus par les trois simulations sont en accord raisonable avec l'expérience pour l'écoulement diphasique.