Increase of oil production is an important challenge for oil industry and deeply dependent of a realistic knowledge of reservoir petrophysical properties, which vary with the change of geological facies crossed by the borehole. The facies description helps to reduce the misinterpretation in the petrophysical properties and in the oil reserves calculations. The coring techniques are made only in a few wells in an oil field causing a sparse detailed facies descriptions and consequently not allowing a more realistic reservoir characterization. In this work is presented an intelligent algorithm that is able to make the transport of facies information generated by core analysis for all the logged wells in an oil field, by the project of an artificial neural network, which is trained to map the geological information in terms of physical properties registered in the wireline logs. The applicability of this methodology is verified using three cored, offshore boreholes, drilled in Namorado oil field, Basin of Campos, Brazil. For the cases appraised here, the neural network exhibits results compatible with core analysis and in a way completely independent of the dimension of the training set.
|
