Concerns regarding climate change and global warming have intensified over the past decade. One of the main strategies to mitigate the rise in global temperatures is CO2 sequestration in geological formations. Reservoir simulators are the tools to evaluate the behavior of CO2 while being sequestered in the aquifers. However, the reliability of the simulation runs relies on the accuracy of the geological model, which is often associated with various uncertainties. Developing multiple realizations of the geological model is a common practice in the industry, which is typically adopted to incorporate uncertainties. There is a practical limitation to conduct detailed simulation studies for all the geological models. Ranking geological models using certain indicators is key to perform detailed studies on selective cases that capture the range of subsurface uncertainties. Dykstra-Parsons coefficient of variation (VDP) and static Lorenz coefficient (Lc) are common static indicators that are used in the ranking process. The main advantage of using VDP and Lc is that they can be easily and quickly calculated. However, studies have shown that dynamic Lorenz coefficient (DLc) is a superior tool to rank geological models and quantify heterogeneity. Numerous studies have examined the impact of aquifer heterogeneity on CO2 trapping mechanisms. However, existing research typically employs static indicators to assess aquifer heterogeneity, such as the coarse grain to fine grain ratio, sand to shale ratio, Dykstra-Parsons coefficient, and coefficient of variation. These static measures inadequately capture the spatial connectivity between reservoir grids and layers. Additionally, other studies compare homogeneous models with a single heterogeneous model without adequately quantifying the level of heterogeneity. In this study, the Sequential Gaussian Simulation method was utilized to generate multiple simulation models with varying levels of heterogeneity. The heterogeneity of these models was quantified using dynamic data from rapid streamline simulation runs and assessed with the dynamic Lorenz coefficient. Alongside a homogeneous model, four models with increasing heterogeneity (DLc values of 0.2, 0.4, 0.6, and 0.8) were analyzed. Sensitivity analyses were conducted on factors such as water salinity, wettability, injection rate/volume, and completion interval. Depending on the completion interval, the increased heterogeneity can enhance both solubility and residual trapping while reducing the amount of mobile CO2. This study uses dynamic data obtained from fast streamline simulations to quantify heterogeneity through the dynamic Lorenz coefficient, providing a more accurate measure of spatial connectivity of the reservoirs/aquifer and utilize that information to assess the impact of heterogeneity on different trapping mechanisms of CO2 in saline aquifers.
