Proper permeability distribution in reservoir models is very vital for reliable full field development and optimization in terms of well placement and completion strategy, managing wells’ and field’s production and injection rates in addition to estimating the recovery factor. Despite its importance, however, it is still one of the most difficult petrophysical properties to model or predict accurately.

Reliable permeability data are those derived from core measurements conditioned to well-test permeability. In most cases, however due to cost and other issues, no more than 10% of the reservoir wells are cored. To fill the gap, many approaches have been undertaken to predict permeability in the non-cored wells from log data.

Besides conventional empirical techniques, recent studies show that Artificial Intelligence (AI) technologies can predict permeability from conventional well logs with better accuracy. This study employs hybrid of both approaches to improve the prediction of permeability of a very heterogeneous carbonate reservoir.

This study was conducted in three main phases. In the initial phase, reservoir layers were grouped considering data availability and layers’ lithology, and petrophysical data. Then each group was divided into at least three clusters depending on porosity and core permeability cross-plots. In the second phase, Adaptive Neuro Fuzzy Inference System (ANFIS) has been designed to predict the index of the rock clusters based on wells’ GR, Rxo, and Porosity logs. The final phase encompassed the utilization of conventional curve fitting techniques to predict the permeability. This prediction was further improved by introducing corrections depending on the core versus predicted permeability cross-plots. Considering the heterogeneous carbonate reservoir under study, this hybrid approach proved to be superior over applying either the conventional approach or the Artificial intelligence method. Plots of depth versus core and predicted permeability shows very good match for all wells considered in this study.

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