Derivation Of A Core Permeability Log And Extrapolation To Uncored Intervals

Conventional methods to estimate permeability from wireline log responses often consist of building models between regularly spaced core plug measurements and logs without paying much attention to the various investigated scales or to core plug sampling. In order to quantify the influence of these elements, a detailed method of permeability prediction based on a core permeability log was developed and applied to the case of a mixed lithology oil reservoir. The method consists of three stages. Firstly, a continuous high resolution core permeability log representative of in-situ conditions is derived in cored intervals using ONLY core data (conventional and special plug measurements, probe measurements and core lithological information). At this stage conventional permeability measurements need to be validated and corrected, core lithological descriptions are used to restore permeability values in cored intervals for which laboratory measurements could not be performed. Secondly, the core permeability log is normalized in order to obtain a vertical resolution close to that of conventional logs - commonly used for permeability prediction. Thirdly, two different prediction models are constructed in the cored intervals using the normalized core log as the target parameter and conventional logs as prediction parameters. The models, one based on multiple regression techniques and the other on discriminant analysis, were found to be more predictive than the conventional models which use regularly spaced core plug measurements, The second modeling method was found to be more predictive than the multiple regression method.