Improvement in Facies Discrimination Using Multiple Seismic Attributes for Permeability Modeling of the Athabasca Oil Sands, Canada.

Abstract

The objective of this reservoir modeling study was to predict the permeability distribution which has impact on the performance of SAGD (Steam Assisted Gravity Drainage), the in-situ bitumen recovery technique. Lithologic facies of a fluvial-estuary channel system observed in the study area are classified into three groups having different characteristics of permeability. Discrimination of the three lithologic facies is a key step in permeability modeling, because different facies use different formulas to estimate facies permeabilities. Seismic data contribute to the lithologic facies prediction by improving facies probability to be used in geostatistical facies modeling. In particular, this study employs a probabilistic neural network utilizing multiple seismic attributes for further improvement of the facies probability. Improvement in facies prediction due to using multiple seismic attributes was demonstrated by comparison with using only a single seismic attribute. Adding P-wave velocity to the group of multiple seismic attributes is a key to enhanced facies discrimination. This paper also discusses a possible cause of the different P-wave velocity of different facies, where sand matrix porosity is uniquely evaluated using a cross-plot of log-derived porosity and photographically predicted mudstone volumes.