Estimating the permeability of reservoir sandstones using image analysis of pore structure

Abstract

In this thesis, a method is developed for predicting the permeabilities of a core using only a small number of SEM images, without resorting to computationally intensive procedures. The pore structure is idealised as consisting of a cubic network of pore tubes having an arbitrary distribution of cross-sectional areas and shapes. The areas and perimeters of the individual pores are estimated from image analysis of scanning electron micrographs of thin sections, with appropriate stereological corrections introduced to infer the true cross sections of the pores. Effective medium theory is used to find the effective single-tube conductance, based on the measured distribution of individual conductances, thereby allowing a prediction of the permeability. The methodology has been applied to several reservoir sandstones from the North Sea, and also an outcrop sample from Cumbria, UK, yielding predictions that fall within a factor of two of the laboratory measurements in most cases. The procedure, although based on Kirkpatrick's intrinsically isotropic effectivemedium approximation, is not only capable of yielding reasonably accurate estimates of the permeabilities, but also gives a qualitatively correct indication of the anisotropy ratio. It also found that the use of an Bernasconi's anisotropic effective-medium approximation does not lead to a systematic improvement in the results, perhaps because the samples used in this study were insufficiently anisotropic for the approaches to yield different results. The validity of the effective medium approximation was also tested against exact pore network calculations. For the rocks examined in this study, with pore conductance distributions having log-variances less than 3, the effective medium approximation was found to be accurate to within a few percent.