Development of a Multidimensional Residual Distribution Solver for Large Eddy Simulation of Industrial Turbulent Flows.

摘要: This thesis presents the results of author's research activity to develop a time-space accurate flow-solver for Large Eddy Simulations (LES), based on Multidimensional Residual Distribution approach. The aim this work is attempt compact high-order algorithm which includes multidimensional flow physics, in hope creating better LES turbulent compressible flows. author proposes natural extension residual distribution (RD) schemes from steady-state computations, where these have shown increased accuracy with reduced stencil and ease parallelization, unsteady computations by using Jameson's dual time steps Thus unstationary problem finding pseudo solution at each real-time step an efficient marching pseudo-time. Second order both space can be obtained Firstly, we study one dimensional Burgers equation, two linear wave equation circular advection cone. Secondly, discretization different terms Navier-Stokes equations described. Proper boundary conditions are discussed. A series test-cases presented assess algorithm, as it has been implemented our solver, called NAS3D. new convective part proposed. scheme proved third while having same second-order schemes. Details three Sub-grid Scale (SGS) models code given. parallelized code, uses either MPI or PVM message passing, proves scale extremely well number processors. Finally, some simulations, simple test-case, e.g. channel flow, industrial applications presented.