Integrating experimental study and intelligent modeling of pore evolution in the Bakken during simulated thermal progression for CO2 storage goals

Pore characteristics of the formation exert significant control over both the development and enrichment of shale plays, as well as CO2 storage capacity of shale reservoirs. This research delved into the alteration of the Bakken shale's pore structure - a major target of underground CO2 storage under the PCOR partnership - when exposed to anhydrous and hydrous pyrolysis (AHP and HP) across a broad temperature spectrum (300–450 °C), aiming to uncover the influence of water in this process. First, N2 adsorption analysis was carried out, followed by the delineation and characterization of distinct pore families within the pore size distribution (PSD) curves of the samples. Subsequently, fractal dimension analysis was employed to gauge the intricacy of the pore structure. Next, generalized regression neural network (GRNN) and radial basis function (RBF) neural network were employed to model the N2 adsorbed/desorbed volume of pyrolyzates obtained from pyrolysis tests. Throughout the process of thermal maturity, AHP and HP pyrolyzates exhibited an overall rise in N2 adsorption capacity, BET surface area, meso-, macro-, and total pore volume. Conversely, the average pore diameter decreased. Also, HP pyrolyzates displayed notably higher N2 adsorption capacity compared to AHP pyrolyzates. Differences in total pore volume and surface area, attributed to mesopores and macropores, were evident between HP and AHP pyrolyzates across all temperatures, with HP pyrolyzates consistently displaying higher values for these pore characteristics. Five pore families akin to those in the original Bakken shale were found in AHP and HP pyrolyzates, displaying similar mean pore sizes. Although there were significant differences in elevation and magnitude, the pyrolyzates showed heightened peaks and increased volumetric representation compared to unheated shale. Overall, AHP and HP pyrolyzates exhibited a general decrease in pore surface complexity and roughness, while concurrently displaying heightened complexity within the pore network during thermal maturation. In the modeling section, the GRNN model demonstrated supremacy in estimating N2 volume adsorbed or desorbed, with an average absolute percent relative error (AAPRE) of 3.61% across the entire dataset. The subsequent sensitivity analysis underscored the significant impact of relative pressure and pyrolysis method on the output, emphasizing the significant contribution of water in shaping pore development throughout shale thermal evolution. This study can serve as a guideline for identifying areas in the Bakken Formation that are better suited for CO2 storage, considering factors such as thermal maturity and water saturation, given extensive availability of such data.