Using facies, data mining and artificial intelligence concepts in the evaluation of a carbonate reservoir in Campos basin, Southeastern Brazil

Abstract

Three different types of facies of a carbonate reservoir in Oilfield A, Campos Basin - Southeastern Brazil, were considered in this work. Initially, Petrobras geologists performed the lithofacies classification of the Wells A03 and A10, which was treated as a starting point. The electrofacies, on the other hand, were determined from geophysical well logs, whose clusters contributed to the understanding of the formations in a simplified way in the whole reservoir, since several lithofacies were reduced to a few electrofacies. The petrofacies, differently, were estimated from laboratory tests on core samples and, and there are three groups resulting from this process. The Winland approach was used to do this, which utilizes mercury injection capillary pressure measurements and, a multiple linear regression, which used the porosity and permeability laboratory evaluations. Together, they showed that the close relationship between pore throat, porosity and permeability controls the flow of fluids in the porous medium. This procedure was developed in these two reference wells mentioned above and extrapolated to the other seventeen wells in a blind test, using seismic data as reference. To do this, the petrofacies, supervised by lithofacies and electrofacies, were estimated by a multiple linear regression and the artificial neural network approaches throughout the complete oilfield. Finally, the maps of the petrofacies suggest that a reinterpretation, in reservoirs of this type, should be performed using the before mentioned approach. This would allow its application in reservoir simulation studies or in optimizing the location of new wells within the development area.