Abstract

The prediction of reservoir permeability has historically been based on the empirical relationship of total porosity to the logarithm of permeability. Regression analysis determines a ‘best fit’ line through the data and the equation of the line is used to transform log derived porosity to permeability. In many sandstones this practice can yield reasonable results; however, the complex nature of the pore systems in carbonate rocks typically produces less than desirable results using this method. Improvements to the correlation can often be achieved when the carbonate interval is divided into geologically similar units such as depositional facies or textures. Diagenetic processes common to dolomite rocks can further complicate the relation.

Although a rock with no porosity has no permeability, there is no theoretical relationship between total porosity and permeability. Permeability is primarily controlled by the length of the flow path through the rock, diameter of the pore throats, and the number of pore throats connecting each pore. The methodology proposed here uses a combination of standard and nonstandard analysis to determine the inputs to the theoretical Carman Kozeny equation from log data. By utilizing data from the Rxo, bulk density (combined with a photoelectric or acoustic), and gamma ray curves, inputs for the equation can be estimated and a close approximation of the permeability generated on a continuous basis. The advantage of the method does not tie the response to a visual facies but acts as an objective characterization of the rock.

As a portion of a Class II DOE project (DE-FC22-93BC14990), a reservoir model was constructed for numerical simulation. To maximize available reservoir data, this methodology was developed to approximate the reservoir permeability from well logs and has proven to provide reasonable results. Portions of this work have been presented at the 1996 SPE Oil & Gas Recovery Conference, Midland Texas.