Artificial Neural Networks Optimization for Reservoir Gas Zones Detection Using Petrophysical and Geological Data
Artificial Neural Networks Optimization for Reservoir Gas Zones Detection Using Petrophysical and Geological Data
Mohammad Reza Masoudian1 Mosayyeb Kamari2 Masoumeh Bashiri3 Ehsan Hossaini4
1) Islamic Aazad University of Tehran,
2) Petrophysical Evaluation Department, National Iranian South Oil Company
3) Petroleum University of
4) MI Services Ltd. Kish,
IOI Code
:
XYGD-BCKBK
Publication :
The 5th International conference of Sciences and Engineering(5icesconf.com)
Abstract :
Reservoir rock lithology and fluids contacts can be determined by petrophysical logs. This determination uncertainty leds to uncertainty in hydrocarbon inplace estimation. The objective of this paper is introducing a new approaches for this purpose by artificial neural networks as one of the artificial intelligence technique. Estimation of hydrocarbon saturation and detection of these zone was done by this technique.
For this purpose, after data preprocessing and applying bad hole models, the investigations be done by four different types of artificial neural networks (MLP, RBF, LSSVM, CMIS) and petrophysical fullset logs (Rt, DT, RHOB, PEF, NPHI, GR, CALI). The results correlation coefficient for this type of ANN were 0.995, 0.995, 0.992 and 0.997 respectively. Compare with similar research by Discriminant Analysis with correlation coefficient of 92.2, this networks give the better outputs.
Keywords : gas zones gas saturation percent petrophysical logs artificial neural networks (ann)