Time-lapse density prediction for reservoir characterization using probabilistic neural networks at Postle Field, Texas County, Oklahoma

Abstract

At Postle Field the main challenge has been to map the Morrow A sandstone. By using a non-linear approach to predict density values from p-wave seismic. I was able to discriminate the reservoir sandstones and identify areas of high quality reservoir. My work confirms the connectivity of the two northern sandstone bodies, increasing the potential resource volume in the study area. The baseline density prediction showed the dry wells were drilled in areas of poor reservoir quality. With the results of this work future drilling locations can be located with less uncertainty. The application of time-lapse neural network prediction successfully predicted changes in the reservoir. Analysis of reservoir modeling and simulation suggest that the time-lapse density changes shown in the neural network prediction at Postle Field are related to pressure changes in the reservoir. Two areas of high quality reservoir have been identified and proposed for future drilling programs. Further engineering evaluation is recommended.