Simulation of flow and hydrodynamic dispersion in porous media.

Abstract

The goal of this work is to provide a fundamental understanding and quantitative description of transient and asymptotic hydrodynamic dispersion in con fined random porous media represented by random-close packed beds of spherical particles in containers (confinements) exhibiting a variety of cross-section al shapes. A thorough experimental study of the general dispersion phenomenon is impeded by a number of instrumental limitations, like the inability to pre pare sphere packings with identical properties or measure dispersion over very different (and discrete) time and length scales. Analytical models operating with averaged values are not able to predict transient and asymptotic dispersion properly. Because of the very different time and length scales involved i n the problem of dispersion in confined sphere packings (from the interparticle pore-scale to the column-diameter scale) the adoption of numerical analysis requires huge computational resources even nowadays. The application of alternative numerical approaches (lattice-Boltzmann method; random-walk particle-t racking technique) allowed us to achieve a close-to-linear performance scaling in terascale computing and, therefore, to use effectively thousands of proce ssor cores and hundreds of gigabytes of memory in the pore-scale simulations of the relevant transport processes and in the detailed analysis of their dyna mic scaling (diffusion and convection in sphere packings; velocity-dependent hydrodynamic dispersion). However, most of the real-world systems are still no t accessible for this study even with such large computational resources. In this respect, our cooperation with the DEISA consortium promotes large-scale s imulations of transport processes in porous media to a new and exciting level where it allows us to study flow and dispersion for the first time in suffici ently large systems which are stringently required for gaining fundamental insight into the scaling, processing, and optimization of technical and analytic al processes in science and engineering that rely decidedly on packed-bed operation and a detailed understanding of the hydrodynamics, in particular.