Facially inhomogeneous medium: increasing reliability of prediction of hydrocarbon reservoir net productive volume

There is a tendency towards increasing hydrocarbon content in facially inhomogeneous accumulations. To improve reliability of petroleum initial in-place resources (PIIP) it is necessary to expand the set of algorithms used in prediction of net reservoir volume. Currently, the prediction is based on the results of seismic amplitude interpretation of 3D data, and there are some limitations related to resolution of this method. Where the results of 3D seismic data interpretation cannot be used, standard approaches used in the practice of reserves estimation shall be applied. However, in the case of facial inhomogeneity, well spacing usually exceeds the size of the studied sedimentological units, which reduces the accuracy of reservoir occurrence prediction between well locations when the conventional approaches are applied. To improve reliability of prediction of hydrocarbon deposit net reservoir volume, the author propose calculation options based on object-oriented algorithms of facies data geostatistical analysis, they are: object-based clustering and declustering. Object-oriented algorithms of geostatistical analysis are a set of methods of quantitative description of reservoir facies composition on the basis of the results of facies interpretation in well columns taking into account geometry parameters typical of sedimentary bodies. In this case, object-oriented clustering involves combining identical facies penetrated by two or more wells into one sedimentary body; and in the case of object-oriented declustering, identical sediments encountered in wells are associated with separate sedimentary bodies. Compared to conventional approaches, the supposed calculation options allowed improving reliability of net reservoir volume prediction for gas accumulation, narrowing the range of estimation uncertainty for the studied parameter, delineating the zone of the expected presence of shaled-out channel facies, and obtaining a set of equally probable realizations, including the option recommended for preliminary assessment of PIIP