A refined radial-basis-function (RBF) method with a forward selection algorithm to improve the stability of the prediction of pore throat sizes was recently reported by the authors. Subsequently, from the pore throat size distribution data, permeability and pore typing models were developed. These models were developed with the core samples from the high-to-medium quality reservoir sections of several Middle East carbonate wells.

Because the RBF is an interpolation method, the validity of RBF based petrophysical models is enveloped by the petrophysical parameter range that the core samples represent. For economic reasons, it is common practice that core analysis be conducted on high reservoir quality rock samples because they are most important to production. To apply RBF-based models for interpreting well logging data, it is important that such models be developed with a broad range of rock qualities to help prevent misinterpreting the lower-quality formation rocks.

To expand the application envelop of the RBF based nuclear magnetic resonance (NMR) permeability models, a new set of core measurements from different reservoir quality sections, as well as non-reservoir quality sections of several carbonate wells, are added to retrain the RBF-based NMR permeability models. Standard statistical validation methods are used to demonstrate the necessity and improvements of newly retrained RBF-based models. The new models are applied to well logging data with varying reservoir quality sections, proving that the new models are adequate for better permeability prediction of all rock quality formations.

You can access this article if you purchase or spend a download.