An Evaluation of the Significance of Permeability Measurements

Abstract
A technique for comparing core-analysis permeabilities to those determined from pressure build-up tests has been developed. This comparison has been used to obtain a qualitative indication of the spatial permeability distribution within the producing formation. Data from 14 wells, equally divided between the Bromide sands and the McKnight dolomite, have been analyzed; results have indicated that the permeabilities in those particular reservoirs are randomly arranged so that they appear to be homogeneously heterogeneous. It has been emphasized that more accurate core data, improved pressure build-up testing and independent determinations of anisotropy are needed before a more quantitative description of the reservoir is feasible. Introduction: The purpose of this paper is to compare permeabilities determined by core analysis with those calculated from pressure build-up data. Since a single value for the effective permeability of the drainage area is obtained from a build-up test, it is necessary that the core analysis values be averaged in some manner to allow the comparison to be made. If an averaging method that reconciles the core and build-up permeabilities can be found, it is possible to infer, in a qualitative sense, the spatial distribution of permeabilities that exists in the reservoir. Since the suggested approach gives a gross indication of the type of reservoir heterogeneity that is based on a comparison, the inherent errors in the raw data must be carefully considered. Recent experimental and theoretical research on the nature of recovery mechanisms, developments in the field of numerical analysis and the evolution of truly high-speed digital computers have made it possible to predict the performance of virtually any reservoir that can be adequately described. Unfortunately, the necessary condition that demands an adequate description of the formation cannot be satisfied in general. As a result, the usefulness of the available numerical models is seriously impaired.

This publication has 0 references indexed in Scilit: