George Mason University
GMU Fairfax Campus
Today @ Mason

College of Science

CAMP Home
   People
    
Alumni
          ├ J. Lindeman, PhD
          ├ J. Bakosi, PhD
          ├ E. Novakovskaia, PhD
          ├ N. Ahmad, PhD
          ├ J. C. Chang, PhD
         
F. Camelli, PhD

   Research
  
Simulation gallery
   Publications
   Resources for students
   Data archive
   Annual conference
  
Computing resources
   Related links
  
About our webpage
   Contact us

József Bakosi, PhD


Year of graduation: 2008

Degree: Doctor of Philosophy, Computational Sciences and Informatics

Research field: Statistical turbulence and dispersion modeling, probability density function methods utilizing finite element and stochastic methods on unstructured grids, code optimization, high-performance computing

Thesis title: PDF modeling of turbulent flows on unstructured grids

Thesis abstract: In probability density function (PDF) methods of turbulent flows, the joint PDF of several flow variables is computed by numerically integrating a system of stochastic differential equations for Lagrangian particles.  Because the technique solves a transport equation for the PDF of the velocity and scalars, a mathematically exact treatment of advection, viscous effects and arbitrarily complex chemical reactions is possible; these processes are treated without closure assumptions. A set of algorithms is proposed to provide an efficient solution of the PDF transport equation modeling the joint PDF of turbulent velocity, frequency and concentration of a passive scalar in geometrically complex configurations. An unstructured Eulerian grid is employed to extract Eulerian statistics, to solve for quantities represented at fixed locations of the domain and to track particles. All three aspects regarding the grid make use of the finite element method. Compared to hybrid methods, the current methodology is stand-alone, therefore it is consistent both numerically and at the level of turbulence closure without the use of consistency conditions.

Several newly developed algorithms are described in detail that facilitate a stable numerical solution in arbitrarily complex flow geometries, including a stabilized mean-pressure projection scheme, the estimation of conditional and unconditional Eulerian statistics and their derivatives from stochastic particle fields employing finite element shapefunctions, particle tracking through unstructured grids, an efficient particle redistribution procedure and techniques related to efficient random number generation.

The solver has been parallelized and optimized for shared memory and multi-core architectures using the OpenMP standard. The methodology shows great promise in the simulation of high-Reynolds-number incompressible inert or reactive turbulent flows in realistic configurations.

Current employment: Los Alamos National Laboratory


See the CAMP contact page for detailed information on contacting us.
Last Modified:

Validate this page with W3C