A series of multivariate statistical techniques (cluster, principal component and discriminant analysis) was tested and applied to well log data from the Camorim field (offshore Sergipe State, Brazil) in order to identify facies previously defined through core description. The second step in the process of facies determination was supported by auxiliary methods (compositional and sequence facies analysis), which produced better results in the calibration of rock-log data, when combined with the multivariate techniques. The facies determination, once established, permits the refinement of the process of formation evaluation, enabling the examination of each facies separately. This procedure made it possible to choose, for each lithology, the parameters used in log interpretation. In parallel, this process allowed the summation of thickness, porosity, fluid saturation and the adoption of different cut-offs for each group, separately. Other applications included: enhancement in the estimation of porosity and permeability, the adaptation of algorithms designed for fast porosity estimation, the mapping of variables useful in the characterization of the vertical variability of the reservoir rocks and the automatic generation of stratigraphic sections. Finally, the possibility of integration of the work from this study with statistical systems of reservoir description, other facies determination techniques currently being developed and the utilization of multivariate statistical methods in well log data, as an auxiliary exploratory tool, were illustrated.
|
