n fouladi - Shiraz University, Shiraz, Iran,
a rabiei - University Petroleum of Technology, Ahwaz, Iran
m riazi - Shiraz University, Shiraz, Iran,

Prediction of permeability, as one of the most important parameters of a reservoir, has been one of the fundamental challenges to petroleum engineers. Exact knowledge of permeability value is indispensable for evaluating the hydrocarbon reservoirs, managements and development of a reservoir, forecasting of future production and design of production facilities. In this regard, different methods have been employed to estimate this important parameter. Direct measurements of permeability in the core analysis laboratories, applying correlations, well tests and well log data calibrated with core data are the most common practices of permeability measurement in petroleum industry. Each method has its own drawbacks and due to the reservoir heterogeneity, they may not be applicable in most of situations. Therefore, presenting a method which make it feasible to predict the permeability at different heterogeneity conditions is crucial. In the recent years Artificial Neural Networks (ANN) have been increasingly applied to solve various problems in the petroleum industry due to their intrinsic abilities to capture the complex heterogeneity in reservoirs. In this study, an artificial neural networks is applied to predict permeability from well logs data. To achieve more reliable results, the model was optimized by genetic algorithm (GA) as a revolutionary technique. The model was developed by 675 data gathered from 4 wells in south of Iran. MSE and R2 of 0.0016 and 0.99072 respectively, confirmed the accuracy and capability of this developed model in predicting of permeability

Artificial neural network, genetic algorithm, permeability, well log