Effect of Network Structure on Image-based Network Modeling of Real Materials
Abstract
Network modeling is an important intermediate-scale modeling tool; because it is a pore-scale technique, it can be used to study first-principles behavior that Darcy-scale models cannot. At the same time, it is a much coarser approach than direct numerical simulation and therefore is capable of modeling orders-of-magnitude larger characteristic scales than more rigorous techniques such as the lattice Boltzmann method. Additionally, the proliferation of 3D imaging techniques such as x-ray microtomography has spurred the development of image-based network modeling techniques, which implies that the network structure is mapped directly from the pore structure in real materials. Because no definitive rules exist for the division of void space into pores and pore throats, different network generation techniques can lead to vastly different network structures, even if they are created from the same 3D image. Little is known (especially quantitatively) about how these differences in network structure impact results obtained from network modeling. To begin addressing this issue, we have performed image-based network modeling using x-ray microtomography data from three different porous media: a sphere packing, an unconsolidated sand, and a high-porosity bed of fibers. For each material, four independent network structures were created, in which pore densities range over approximately two orders of magnitude for each material. The large variation in pore density impacts other parameters such as pore coordination number, pore size distribution, and pore throat size. In this talk we present the differences in network structure, results from the modeling of single phase flow, and results from modeling quasi-static drainage. We show that despite the dramatically different pore-size distributions, phenomena that depend on pore-throat properties (such as permeability and drainage) remain quite insensitive to the orders-of-magnitude variations in pore density in the networks, although this depends on the methods for computing hydraulic conductance. Finally, we illustrate differences in both network properties and transport properties for the high-porosity fibrous materials versus the unconsolidated granular materials.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2009
- Bibcode:
- 2009AGUFM.H13C0989B
- Keywords:
-
- 1828 HYDROLOGY / Groundwater hydraulics;
- 1832 HYDROLOGY / Groundwater transport