Prediction of Shale Plugs Between Wells in the Heavy Oil Sands at Long Lake Using Seismic Attributes

F. David Gray, Paul Anderson, and Jay Gunderson
Veritas, Calgary, AB

Bitumen sands of the Long Lake Project are produced using the SAGD (Steam-Assisted Gravity Drainage) process. SAGD consists of two parallel horizontal wells, one situated a few metres above the other. The upper well injects steam into the reservoir, which reduces the viscosity of the bitumen. The heated bitumen then flows down the outside of the steam chamber to the lower horizontal well, where it is collected and pumped to the surface. Difficulties arise if either of these horizontal wells encounter or are in proximity to “shale plugs”. These shale plugs may be abandoned channels or overbank deposits in the McMurray reservoir at Long Lake. If they are sufficiently thick and wide, then they can interfere with the efficiency of the SAGD process. A method that uses seismic AVO attributes input to a neural network to predict shale proportion (Vsh) is shown to be effective at predicting the presence of these shale plugs at Long Lake by comparison to core studies along cross-sections. The seismically-derived Vsh prediction suggests that these shale plugs may have a smaller lateral extent than suggested by the core studies and therefore that there is a higher net-to-gross than anticipated based on core analysis alone.

AAPG Search and Discovery Article #90039©2005 AAPG Calgary, Alberta, June 16-19, 2005