عنوان مقاله [English]
One of the major geochemical parameters is total organic carbon (TOC) which is used to evaluate hydrocarbon generation potential of source rocks. Measurement of such important parameter requires performing tests on small-scale drill cuttings which is too expensive and they are measured on a limited number of samples. However, petrophysical data are measured for all drilled wells in a hydrocarbon field. In this paper, the artificial neural network technology was used to estimate TOC from petrophysical logs. The correlation coefficient between the estimated TOC from neural network and measured data from Rock-Eval pyrolysis is 71% which is an acceptable value. Then, the estimated total organic carbon log is used to identify the organic facies with a maximum amount of TOC. The methodology used in this paper is cluster analysis that includes MRGC and AHC methods. The results of these two methods are compared and evaluated based on cluster validity test and the best method of data clustering was used to cluster petrophysical data into certain facies. The results showed that the MRGC clustering provides better results with higher accuracy. Moreover, using this method has advantages in comparison to AHC for determination of organic facies and has capabilities to provide high resolution clusters. The presented methodology was explained by using a case study from one well of Azadegan field, Abadan plain.