Author(s)

M.A. loannidis & E. Lange

Abstract

A systematic study of the geometry and topology of statistically homogenous porous media is undertaken. Pore and throat size distributions and pore-to-pore connectivity measures (genus and coordination number) are determined by applying 3D reconstruction algorithms to simulated porous microstructures. These data are subsequently correlated with the statistical properties of the simulated media (porosity and 2-point correlation function) using neural networks. It is shown that the mean and variance of pore and throat size distributions are related to and can be estimated empirically from knowledge of the porosity and autocorrelation function. On the contrary, the variability of the average coordination number appears to be unrelated to these paramet

Keywords