Abstract: |
Reservoirs in fluvial sequences are generally challenging to model for reservoir properties. This is due to the large contrasts between the different facies associations, the distinct architecture, and the different heterogeneity within each facies association. Therefore a combination of different geostatistical methods is necessary to reflect the multitude of variations. This combination is illustrated with a case study. A target reservoir for studying a pilot CO2 injection occurs at Ketzin in the Stuttgart Formation of Middle Keuper (Upper Triassic) age. The Stuttgart Formation is lithologically very heterogeneous: sandy facies of assumed fluvial origin with good reservoir properties alternate with flood-plain-facies with poor reservoir quality. The thickness of the sandstone intervals can reach several tens of meters where channel deposits are stacked. The width of a fluvial channel belt system is known from analogue studies to vary between several tens of meters to several hundreds of meters. The challenge in the geological modelling of such a sequence is to incorporate a diversity of information of both local and regional origin, and combine it for a model describing both the geometry of the facies distribution and describing the reservoir property variation internally in the facies. For this modelling study has been chosen an object modelling tool generating stochastic models of sinuous channel bodies within a back-ground matrix. This pattern is reflecting channel belts in a floodplain setting, where each channel belt is the cumulative result of a river channel migrating during some period of time. The resulting heterogeneity within each channel belt is produced by using a Sequential Gaussian Simulation scheme to populate with petrophysical properties of porosity and permeability. The SGS is used to generate a variability with a geometry/size mimicking the patchy occurrence of pointbar and channel fill porous sands within the channel-belts.
|
