Neural networks and their applications in lithostratigraphic interpretation of seismic data for reservoir characterization

摘要

Paradigm shift in hydrocarbon exploration and development strategies has increasedutilization of 3D seismic attributes many fold for reservoir characterization. The high datasampling provided by 3D seismic enables a better description of reservoir heterogeneities andmore realistic assessment of original hydrocarbon in-place. But to establish the complicatedand nonlinear relationship between seismic attributes and reservoir properties has been amajor challenge for working geoscientists in synergistic interpretation. In recent past,supervised neural networks are utilized for predicting reservoir properties away from theboreholes in inter-well regions after establishing the relation between multi-seismic attributesand well log data. The effectiveness of these neural network techniques in 3D seismicinterpretation has been demonstrated through a real data example from the Cambay Basin ofIndia.This study shows that systematic interpretation approach along with artificial neural networktechniques has helped in understanding the subsurface image, allowed precise mapping ofreservoir sand geometry and internal reservoir properties. The ability of artificial neuralnetworks to extract information from data and present it in a way that accentuates thefeatures of interest allows us to make more intelligent interpretations. Predicted effectiveporosity through supervised neural networks has provided very high degree of confidence inanalyzing the porous and non-porous zones of the reservoir. This study has been very helpfulin providing more meaningful geological information about the extent, shape and laterallithology variation of reservoirs. Further, it has explained hydrocarbon entrapment in the 3Dcovered area in a more realistic way which has reduced the cost of exploration andproduction, improved the recovery through more efficient production, better reservoirmanagement and has opened up new adjoining areas for further exploration.