Tight gas reservoirs show challenges to geologists to characterize because of their tendency to begenerally heterogeneous due to depositional and diagenetic processes. The value petrophysical properties are very valuable in static and dynamic reservoir modeling. This paper presents a prediction of Klinkenberg permeability by using artificial neural network, composite logs and core data in basin in western USA. The klinkenberg model approximates a linear relation between the measured gas permeability and the reciprocal absolute mean core pressure. This model has been a consistent basis for the development of methods computing the absolute liquid permeability of a core sample based on a single data point.

In tight gas reservoir with increasing in gas slippag ( klinkenberg effect) cause to decrease in pore-throat size and permeability parameters. In advanced there were some method to determine Klinkenberg permeability in situ which can be obtained by measuring just routine air permeability and Klinkenberg parameter such as byrnes in 1997 & 2003 but these ways intensively depend on core permeability, so it needs some core plugs and inevitably we have to spend much time and money.

The goal in this study was to research about relationship between core Klinkenberg permeability and composite logs (gamma ray, density, neutron and formation resistivity and so on) by using MLP & Back Propagation methods (Artificial neural network) to characterize the Klinkenberg permeabilityin situ in 3 different wells in 3 stages (training, validation and application) with suitable core calibration. For two wells there is very good core calibration and the R2 is more than 0.7 in training and application processes.

The importance of evaluation of tight gas reservoir with high heterogeneity by using artificial neural network and conventional logs is spending less capital or time and finally obtaining reliable Klinkenberg permeability in situ.

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