Permeability prediction is an important step in reservoir characterization for effective modeling of water-flood performance and subsequent EOR processes. Permeability should be determined accurately versus depth in all cored and logged wells in a field. This is normally achieved by the use of cross-plots of core permeability versus core porosity from log-derived porosity in un-cored wells. The main aim of this research is to use X-ray CT images of reservoir cores to identify geological textures and predict permeability based on well-established models in carbonates.
Full-diameter reservoir cores were imaged using dual energy X-ray CT. Porosity logs were derived along the entire cores from the dual energy data. Textural variations were identified from the appearance of the CT images. Statistical sample selection was designed to honor the distribution of porosity and CT-textures. Porosity-permeability correlations were established in the texture domain using plug-scale poroperm data and thin-section photomicrograph analysis.
Plug-scale geological analysis confirmed the CT textures in the images, which allowed the derivation of texture logs along the cores. The porosity and permeability data were fitted into unique trends that were derived from the detailed textural analysis. This process provided the link between the poroperm trends and the different textures in the core enabling permeability to be predicted along the entire whole core intervals.
This paper describes a novel approach of combining textures with porosity to model permeability along the reservoir column. A unique dual energy CT technique was used to ensure that all the core property variations were well represented in the plug-scale core analysis measurements. This analysis will open the way for predicting permeability in un-cored wells using texture information from down-hole image logs. In the future, machine-learning algorithms will be employed following the same workflow for effective permeability prediction in the field.