Toc Determination of Gadvan Formation in South Pars Gas Field, Using Artificial Neural Network Technique

Abstract

TOC Content in a source rock is potentially affecting logging data (density, sonic, neutron and resistivity logs). Hence analyses on these logs assist to a reliable assessment of a source rock which is quick and economically cheap method rather than direct geochemical analysis. A source rock interval poses less density, low velocity, higher sonic porosity, high GR values and increase in resistivity. In this research Gadvan Formation was studied in two boreholes as potential source rocks. The log data of two wells were used to construct intelligent models in a source rock of the south Pars Gas field, southwest of Iran. A suite of geophysical logs (neutron, density, sonic and resistivity Logs) and cutting chips samples were used to determine TOC Content of this Formation. Rock- Eval pyrolysis data reveal that Formation is poor source rock (less than 0.5%), whereas logging data and intelligent methods calculations suggest the Gadvan Formation as poor source rock. Hence we attempt to correlate between geophysical data and direct TOC content measurements using ΔLogR, Rock- Eval and neural network techniques. The results showed that intelligent models were successful for prediction of TOC content from conventional well log data. In the meanwhile, similar responses from different other intelligent methods indicated their validity for solving complex problems.