Publications

A project to evaluate the benefits post-stack seismic inversion and the multi-attribute Neural-Net technologies was successfully performed. The main focus of the project was on carbonates of Oligocene age in East Java. The inversion and multi-attribute Neural-Net analysis was performed on 2D seismic data, which passed through a number of wells. The wells were used as calibration for parameter optimization and as blind well tests to aid in the evaluation of the final results.The inversion project produced band-limited and absolute acoustic impedance models that more clearly identified lower and higher acoustic impedance layers in the subsurface. Seismic data is composed of reflected events that identify the top and base of layers, but has limitations on resolving thin layers and amplitude issues due to tuning and wavelet interference effects. Inversion can minimize tuning effects and produces a layer based result, which can have advantages over the seismic reflection images.The Multi-attribute Neural-Net technology has been shown to be useful in predicting reservoir properties in Carbonates (Soroka, et.al, 2008). The multiattribute Neural-Net analysis will be used to predict an acoustic impedance model. In addition a test to predict both a porosity and resistivity model will be conducted as part of the multi-attribute Neural-Net project. The Neural-Net technology has the advantage of being a nonlinear approach that can identify complex relationships between a collection of seismic attributes and a target reservoir property from calibration wells. The results from this project demonstrate both advantages and limitations to the Neural-Net technology that can aid in future studies.This project report describes the inversion and multi-attribute Neural-Net technologies. The work flows, quality control steps and results are described to document the valuable lessons learned throughout the project. The final results are encouraging and show that advanced seismic techniques are capable of extracting additional valuable information about carbonate reservoir properties from seismic. The results also demonstrate that input data quality has a direct impact on final result quality and that good seismic and well information is essential in predicting higher quality rock and reservoir properties.Keywords: Seismic Attribute, Reservoir Characterization, Acoustic Impedance Inversion, Multi-Attribute Neural-Net Technologies.