Mohammad Reza Asef - Kharazmi University, Tehran, Iran
Hesam Zarehparvar Ghoochaninejad - Kharazmi University, Tehran, Iran
Seyed Ali Moallemi - National Iranian Oil Company, Exploration Directorate, Tehran, Iran

A full waveform acoustic logging can provides stoneley wave velocity in conjunction with the shear and compressional wave velocities. The stoneley wave is usually used to estimate reservoir permeability and detecting probable open fractures. In subsurface, stoneley wave velocity can be obtained by running the Dipole Sonic Imager (DSI) tool. However, in most of the drilled wells this measurement is not available, due to high costs of running it. In recent years, artificial intelligence techniques are successfully employed to predict reservoir parameters, from conventional well logs. In this study, by using the Adaptive Neuro-Fuzzy Inference System (ANFIS), the stoneley wave velocity is estimated from the bulk density (RHOB), compressional wave velocity (DTCO), and deep resistivity (Rd) well logs, which are usually available in every drilled well. ANFIS combines the adaption capabilities of neural networks with the approximate reasoning of fuzzy systems to formulating between the input and output spaces. Used data for training and testing the model acquired from Louisville Sea Mount, located in the Central South Pacific Ocean. Results have shown that the developed model can provides reliable estimations of Stoneley wave velocity, wherever the full waveform acoustic acquisition is not available.

ANFIS, Petrophysical well logs, Sonic logging, Stoneley wave velocity, Subtractive clustering