Facies Mapping by Intelligent Algorithms

Facies mapping is a cross-section representation of the lateral variation of a petrophysical property based on stratigraphic correlation criterions and involving all logged boreholes in an oil field. The procedure to produce the facies mapping can be divided in three steps. In the first one, in a cored borehole is performed the calibration among facies and log readings that establish a log zonation for non-cored boreholes. Second step calculates the petrophysical property to be mapped in all boreholes. The last step performs the well correlation based on the log zonation and stratigraphic sequence. The influence of clay occurrence in the L-K plot interpretation may be attenuated adding a third axis scaled in clay volume, as the GR-L-K plot. Admitting the existence of angular patterns in the points distributed in the GR-L-K plot, we introduce a new competitive neural network, nominated as generalized angular competitive neural network, specialized in the search of angular patterns present in ndimensional data. This characteristic allows the classification of the points in the GR-L-K plot in terms of sedimentary facies previously identified. Well correlation determines the correlation lines by a fuzzy inference system able to identify the stratigraphic sequence in a cored borehole and promote the sequence match in others boreholes in oil field. Thus, facies mapping is constructed interpolating the petrophysical property guided by the correlation lines. This method is presented with synthetic and evaluated with porosity logs and core analysis from two boreholes in the Namorado oil field, Campos' Basin, Brazil.