ESTIMATION OF PERMEABILITY FROM IMAGE-ANALYSIS OF RESERVOIR SANDSTONES

A simple correlative model is proposed for permeability estimation of sandstones using data derived from image analysis of thin sections. A total of 54 sandstone samples with a wide range of porosity and permeability values from a North American oil reservoir were used in this study. The model is empirical, and based on only two-dimensional image data. Image-derived porosity and the diameter distribution of the largest inscribed circles within two-dimensional pore areas are shown to be sufficient for permeability prediction. Tortuosity, pore shape and pore connectivity appear to have only small effects on permeability unless these properties are themselves correlated with porosity or pore size. The regression analysis between image-derived data and core-measured permeability shows that 90% of the variance in permeability can be explained by the two-dimensional data within a factor of 1.68. In this study it is shown that there are strong associations between two-dimensional pore sizes and three-dimensional throat sizes, and that synthetic mercury injection curves can be constructed from image data. These strong correlations are thought to be the main reason for successful estimation of permeability from two-dimensional image data. The application of the results is in estimating permeability and in generating synthetic mercury injection capillary pressure curves from samples too small for conventional testing. This provides a cost-effective means of formation evaluation and contributes to better reservoir description.