Quantification of dynamic contact angles and interfacial curvatures for liquid/supercritical CO2 displacement of water in a micromodel

In an effort to mitigate the effect of greenhouse gases and decrease the overall global carbon footprint, the exploration of CO2 sequestration in underground reservoirs has shown to be a plausible solution to reduce CO2 emissions. The major challenge present in this solution is the unpredictable behavior of CO2 after injection. Better understanding of flow physics at the pore-scale can provide the data needed to develop and improve the numerical models used in field-scale simulations. In this study, a homogeneous micromodel was employed to study the fluid flow of scCO2 as it was injected and as it radially displaced the resident water in the system. A high-speed camera captured a series of images and an in-house MATLAB code was used to measure the scCO2-H2O contact angles. Utilizing the consecutive image frames, the growth speeds and areas were also gathered, as well as a categorization of "advancing," "static," or "receding" for each meniscus in relation to where menisci were located previously. Notably, no receding meniscus was detected, in contrast to previous studies conducted in rectangular micromodels. Furthermore, capillary pressure (P_c), curvature (κ), capillary number (Ca), and CO2 saturation were also calculated and compared among the three movement categories. It was found that the average contact angle for static menisci was roughly 40° and the average for advancing was a slightly smaller value (around 30°). The growth speed and growth area of the leading finger did not have a linear relationship with time nor CO2 saturation, but instead displayed periodic bursts that sometimes surpassed the constant injection rate of scCO2 in the micromodel. The radial distance of the leading finger did increase with time and as CO2 saturation increased, but some distance plateaus suggests other fingers grew instead and it was only after a certain threshold that the leading finger started to move again. Lastly, the P_c's calculated programmatically matched the behavior of the values measured by a transducer, but were about 400 Pa higher.