A Hosseini - Faculty of mining and petroleum engineering, Shahrood University of Technology
A Kamkar Rouhani - Faculty of mining and petroleum engineering, Shahrood University of Technology
A Roshandel - Faculty of mining and petroleum engineering, Shahrood University of Technology
J Hanachi - Iranian Offshore Oil Company

Water saturation (Sw) is a significant petrophysical parameter usually used for reservoir estimation and production. This parameter is one of the mostdifficult petrophysical properties to determine and predict. The conventional methods for water saturation determination are core analysis and well testdata. These methods are, however, very expensive and time-consuming. One of the comparatively inexpensive and readily available sources ofinferring Sw is from well logs. In recent decades, artificial Intelligent (AI) has many applications in the petroleum engineering as well as other areas ofresearch. The aim of this paper is to use two diverse machine learning technology named back-propagation neural network (BPNN) and generalregression neural network (GRNN) for predicting the water saturation of four wells in Burgan reservoir, south of Iran. Comparing the obtainedresults of these two methodologies has shown that BPNN is a faster and precious method than GRNN in prediction of water saturation.

porosity, well log data, petrophysics, general regression neural network, back-propagation neural network