The Application of Probabilistic Neutral Network Analysis to Determine Sandstone Reservoir Distribution



Azuni Field is located on South Sumatra Basin was discovered in 2004 and since 2006 this field producing oil from Lower Talang Akar Formation (LTAF). This field also has potential gas reservoir in Gumai Formation (GUF). There are two main potential gas reservoirs in this formation, Sand-11 and Sand-12 which are potentially to be developed in the next phase. The Gumai sand formation is relatively thick that represents pro-deltaic environments depositional system. The seismic section over this interval indicate alternating peak and through which may indicate the presence of sand reservoir layers alternating with shale. To generate Gumai reservoir distribution map of Sand-11 and Sand-12, probabilistic neural network (PNN) analysis based on multi-attribute utilizing 3D PSTM seismic data and 14 wells has been applied. Based on sensitivity analysis, sands and shales which representing reservoir and non-reservoir can be differentiated from density log. A statistical approach, including multi-attribute and PNN analysis were used to derive the relationship between attributes of the seismic data and density log. The established relationship was used to estimate pseudo density at each seismic trace on the 3-D. The results of PNN analysis indicated that four attributes show high correlation 0.86 of log density properties in the study area. This method provides the distribution and orientation of reservoir targets consistent with well data. It has been successfully used to define the distribution of Sand-11 and Sand-12 of GUF. In future this study result will advance the petroleum development activities in the area.


Authors : J. Wiyono, M. N. Alamsyah, K. W. Nugroho, Muhtar, I. Buldani