Understanding unstable immiscible displacement in porous media

Our global heavy and viscous oil reserves are immense. 70% of our current global oils reserves are viscous or heavy. For an energy secure future, exploitation of heavy oil reserves is necessary to mitigate the impact of steadily declining conventional reserves. Though most viscous and heavy oils are produced by thermal stimulation, several cases do exist where thermal methods are neither technically feasible nor economically profitable. In such cases, non-thermal EOR methods have to be applied. Any displacement process at such high viscosity ratio will be influenced by viscous fingering. Polymers are typically added to the water to stabilize the displacement but for oils above a couple of 100 cp viscosity a stable displacement is not feasible. As unstable displacements are not very well understood, visualization along with experimentation is critical for understanding and modeling the process. In this study, multi-scale experimental strategy was employed; experiments were conducted in cores at lab-scale to generate quantifiable data and were repeated in small micro-fluidic cells for visualization of the mechanism. Polymer flood as an alternative non-thermal process in a structurally complex carbonate formation was tested. In carbonates formations, thermal methods are not preferred as mineral dissolution and precipitation lead to formation damage. Effect of timing of polymer flood was studied in great details. Result from both the micromodels and core-floods indicate that for heavy oils, unlike light oils, timing of polymer injection is not critical and a tertiary polymer flood at the completion of waterflood can also produce significant incremental oil. In some cases, tertiary polymer flood even out-performs a secondary polymer flood. A major problem with modeling and predicting the performance of an unstable flood is largely due to our inability to accurately capture viscous fingering or its effects. Viscous fingering is a complex phenomenon and is dependent on several parameters such as injection rate, viscosity ratios, heterogeneity and dimensions. The micromodels were used to visualize the variation in flow pattern at different viscosity ratio and injection rates while core floods provided essential modeling data. Based on the results two new models were developed: a simplified network model that could accurately predict the viscous fingers for all viscosity ratios and a lumped model that capture the effect of viscous fingers at larger scales through pseudo-relative permeability functions. A dimensionless scaling parameter similar to the instability parameter of Peters and Flock (1981) was also developed that is useful in predicting the recoveries of all unstable displacement at various viscosity ratios, injection rate, permeability and width. The scaling parameter showed excellent fit with experimental data of over 60 experiments.