--> Tombua Reservoir Modeling: Using MPS to Facilitate Geologically Robust Connectivity

Datapages, Inc.Print this page

Tombua Reservoir Modeling: Using MPS to Facilitate Geologically Robust Connectivity

 

Beeson, Dale1, Ricardo Van-Deste2, Sebastien Strebelle3 (1) ChevronTexaco, Bellaire, TX (2) Sonangol, Luanda, Angola (3) ChevronTexaco, San Ramon, CA

 

The Tombua field was discovered with the Tombua-1 well bore in 2001. It is part of the Tombua-Landana project which is a major CVX-lead deepwater development encompassing 470 square kilometers in Angola’s prolific Block 14. There are over 20 identified reservoirs in the project area containing oil in place of approximately 1.2 billion barrels. Reservoir sands were deposited as part of an extensive Miocene age lower slope turbidite channel system.

Tombua reservoir modeling relies heavily upon seismic imaging to help predict and spa­tially distribute the reservoir properties. A principal component (PCA) workflow is used to predict volume shale which is then transformed into reservoir PKS values. Tombua model­ing is heavily dependant upon the seismic imaging for conditioning the spatial distribution of reservoir properties. However, where reservoir sands are poorly imaged by the seismic data, especially the thinner channel sands at depth, difficulty arises in modeling reservoir body continuity/connectivity using conventional variogram-based geostatistical simulation methods.

Model variograms represent “nearest neighbor” measures for guiding channel simula­tions and are dependant upon the seismic imaging. When the channel continuity is poorly imaged, the geostatistical simulation continuity suffers. Variogram-based models for the thinner channel sands are often poorly connected even though our geologic analog data as well as much of our seismic imaging suggest more connected solutions. Since reserves are largely a function of producibility via reservoir connectivity, this is a significant concern. Our solution has been to incorporate geologically robust channel training images using Multi-Point Statistics (MPS) to develop much improved model-based connectivity.