Ho Trong Long's Homepage

posted by Ho Trong Long @ 9:29 PM

EAGE Near Surface 2005 Conference & Exhibition, 04 - 07 September 2005, Palermo, Italy

DETECTION OF WEAK ZONES BY INVERSION OF 2-D ELECTRICAL IMAGING DATA USING NEURAL NETWORK

Ho Trong Long1*, Gad El-Qady1, Keisuke Ushijima1
1 Department of Earth Resources Engineering, Graduate School of Engineering,
Kyushu University, Hakozaki 6-10-1, Higashi-ku, Fukuoka 812-8581, Japan

ABSTRACT

There is increasing interest, particularly for environmental and engineering studies of nearsurface areas. Electrical Imaging or electrical tomography is a survey technique developed for the investigation of areas of complex geology where the use of resistivity sounding and other techniques is unsuitable. Inversion techniques of Electrical Imaging data normally done using least-squares method (e.g. Sasaki 1989, 1992, Loke et al. 1996, 2002, 2003) based on finite-difference method (Dey and Morrison 1979) or finite-element method (Silvester and Ferrari 1990). In this study, we used Neural Network to invert the 2-D Electrical Imaging data and applied for real data to detect weak zones of Mekong river banks of south Vietnam. Detection of the weak zones has a great advantage because it can be potential zones of erosion. The erosion of this river takes permanently and causes great damage to infrastructures and human beings in the area.

KEYWORDS: neural network, electrical imaging, weak zones.

posted by Ho Trong Long @ 1:21 AM

Second Annual Petroleum Conference and Exhibition, 15-19 May 2005 in Cairo, Egypt

POROSITY & PERMEABILITY ESTIMATION IN A2-VD OIL PROSPECT,
SOUTHERN OFFSHORE VIETNAM USING ARTIFICIAL NEURAL NETWORKS

Ho Trong Long1*, Bui Thi Thanh Huyen2, Gad El-Qady1, Keisuke Ushijima1
1 Department of Earth Resources Engineering, Graduate School of Engineering, Kyushu University, Hakozaki 6-10-1, Higashi-ku, Fukuoka 812-8581, Japan
2 Department of Civil and Earth Resources Engineering, Graduate School of Engineering,
Kyoto University, Yoshida-Honmachi, Sakyo-ku, Kyoto 606-8501, Japan


ABSTRACT

Porosity and permeability play an important role in evaluation of hydrocarbon potential in a reservoir. Their distribution can be combined with another data to predict major faults or fractured zones relatively to high porosity area. A2-VD prospect, located in Cuu Long basin, southern offshore Vietnam is a main target area for oil and gas exploration. In this study, we apply Artificial Neural Networks (ANNs) to derive porosity and permeability directly from well log data.
To generate training patterns for the ANNs, a combination of well logs, that comprises neutron porosity, bulk density, P-sonic, deep resistivity, shallow resistivity and MSFL were used as input parameters of ANNs. Firstly, ANNs was trained by data obtained from core samples. Several ANNs paradigms have been tried on a basis of trial and error. We implemented batch back-propagation algorithm for training porosity network while found that quick propagation algorithm is more effective in the training case of permeability network. Secondly, trained ANNs was tested and applied for real data set of some wells to calculate and reveal the distribution maps of porosity or permeability. This distribution can be correlated with seismic data interpretation for fault identification in the studied area. The derived ANNs give out quite good results of porosity and permeability distribution with high reliability because of fast, accurate and low cost features. The present ANNs should widely be applied in oil and gas industry in Vietnam.

KEY WORDS: A2-VD prospect, neural network, porosity, permeability, well log, seismic, fault.

posted by Ho Trong Long @ 1:10 AM