Predictive Pore-Scale Modelling of Multiphase Flow

Flow properties for a variety of porous media canbe predicted using pore-scale modelling withgeologically realistic networks. A state of the artpore-scale network model is developed for thispurpose. It combines topologically disorderednetworks that represent real systems with detaileddisplacement mechanisms for any sequence of water andoil flooding and any wettability. The model isverified by successfully predicting experimentalrelative permeability from a well described water-wetBerea sandstone. More complex experimental data aresubsequently investigated. By developing amethodology combining realistic network topology withnetwork properties tuned to easily obtainableexperimental data such as mercury injection capillarypressures the flow properties for several porousmedia can successfully be predicted. This isinitially applied to a water-wet sand pack. The aimof this work is not simply to match experiments butto use easily acquired data to predictdifficult-to-measure properties. Furthermore thevariation of these properties in the field due towettability trends and different pore structures cannow be predicted reliably.