Print this pageNew Approach, New Play, Same Area: Utilization of Neural Network Approach from Well Logs and 3D Seismic Data, a Case Study in Middle Baong Sand, North Sumatra Basin, Indonesia
Middle Baong Sand or also known as MBS sand is the most prolific reservoir in North Sumatra Basin, yet to characterize this particular reservoir is quite a challenge when using conventional methods to define its facies, geometry, and distribution. Deposited as a submarine sand, lateral discontinuity of MBS is one of the problems that freeze the exploration and development strategies in the area. This study tries to re-determine the MBS by applying the Artificial Neural Network (ANN) approach in electric-facies analysis and seismic multi-attribute analysis.
The ANN approach in reservoir characterization and geometry analysis in MBS sand was conducted on two different types of data set, Well logs and 3D seismic data. The reference well, Besitang-1 exploration well, was used as the training well. A complete data set that provided from the well, support the electric-facies analysis that resulted in four different type of sand that developed in Besitang-1, from base to the top are (1) Sand-4: medium to coarse grain sand with coarsening upward character, (2) Sand-3, medium to coarse grain sand, (3) Sand-2: interbedded medium-fine sand and silt, and (4) Sand-1: fine sand to silt. The facies definition in Besitang-1 well then applied to Ruby-1 well to predict the facies development in the study area. Training and testing processes on both wells successfully identified and recognized the four different types of sand in Ruby-1. Lateral facies distribution was then predicted through out the study area using the 3D seismic data, where synthetic seismogram on Ruby-1, was generated to recognize each of facies type on seismic.
Seismic analysis and interpretation showed that all of the facies are recognize able, and 14 seismic attributes for each facies was selected to define facies’s geometry and mapped the lateral distribution. Combination and cross-plot were conducted to determine the best seismic attributes that unique to differentiate one facies with the others, with the help of Artificial Neural Network.
This approach and its result, give a new meaning and life for the study area. After years of no activities, a new prospect was effectively generated with significant amount of potential recourses.
AAPG Search and Discover Article #90100©2009 AAPG International Conference and Exhibition 15-18 November 2009, Rio de Janeiro, Brazil