PDF modeling of turbulent flows on unstructured grids

摘要: In probability density function (PDF) methods of turbulent flows, the joint PDF several flow variables is computed by numerically integrating a system stochastic differential equations for Lagrangian particles. A mathematically exact treatment advection, viscous effects and arbitrarily complex chemical reactions possible; these processes are treated without closure assumptions. set algorithms proposed to provide an efficient solution transport equation modeling velocity, frequency concentration passive scalar in geometrically configurations. An unstructured Eulerian grid employed extract statistics, solve quantities represented at fixed locations domain track All three aspects regarding make use finite element method. Compared hybrid methods, current methodology stand-alone, therefore it consistent both level turbulence consistency conditions. Several newly developed described that facilitate numerical geometries, including stabilized mean-pressure projection scheme, estimation conditional unconditional statistics their derivatives from particle fields, tracking through grids, redistribution procedure techniques related random number generation. The solver has been parallelized optimized shared memory multi-core architectures using OpenMP standard. Relevant performance parallelism on cache-based machines discussed presented detail. shows great promise simulation high-Reynolds-number incompressible inert or reactive flows realistic