Simulating secondary waterflooding in heterogeneous rocks with variable wettability using an image‐based, multiscale pore network model

The two‐phase flow properties of natural rocks depend strongly on their pore structure and wettability, both of which are often heterogeneous throughout the rock. To better understand and predict these properties, image‐based models are being developed. Resulting simulations are however problematic in several important classes of rocks with broad pore‐size distributions. We present a new multiscale pore network model to simulate secondary waterflooding in these rocks, which may undergo wettability alteration after primary drainage. This novel approach permits to include the effect of microporosity on the imbibition sequence without the need to describe each individual micropore. Instead, we show that fluid transport through unresolved pores can be taken into account in an upscaled fashion, by the inclusion of symbolic links between macropores, resulting in strongly decreased computational demands. Rules to describe the behavior of these links in the quasistatic invasion sequence are derived from percolation theory. The model is validated by comparison to a fully detailed network representation, which takes each separate micropore into account. Strongly and weakly water‐and oil‐wet simulations show good results, as do mixed‐wettability scenarios with different pore‐scale wettability distributions. We also show simulations on a network extracted from a micro‐CT scan of Estaillades limestone, which yields good agreement with water‐wet and mixed‐wet experimental results.