Evaluation of gas potential in low permeability reservoirs (< 0.1 md) generally referred to as Tight Gas reservoirs is not very straight forward as in conventional reservoirs. This study is focused on modeling permeability in the Travis Peak Formation where there are many challenges. One may encounter low resistivity pay due to clay coated grains and alternating thin laminae of finer and coarser sandstones with distinctly different pore geometries, natural fractures, bituminous zones and multilateral fluvial channel sandstones in broad lenses. Core data is sparsely available. Most importantly, there are no structural features that may construe trapping mechanisms. In view of these challenges, a permeability model was developed primarily for the Travis Peak Formation of Robertson and Leon counties where it has produced 96 BCF of gas and 0.54 MMbbl of oil.

A permeability model was developed by integrating core and log data using the Adaptive Neuro Fuzzy Logic Inference System (ANFIS) that combines the functionality of neural network and fuzzy logic techniques. A combination of conventional logs such as lithology (GR, SP), porosity (RHOB, NPHI), deep and shallow resistivity logs were integrated with core data (porosity, permeability).

The results showed excellent agreement with measured core permeability values. The results of the conventional porosity-permeability transform are also presented for comparison purposes. Since the data has been used from the Travis Peak Formation of East Texas Basin comprising several counties, it is expected that the permeability model