Pore structure characterization of solvent extracted shale containing kerogen type III during artificial maturation: Experiments and tree-based machine learning modeling

Abstract

Shale samples with type III kerogen from the Damoguaihe formation were exposed to hydrous and anhydrous pyrolysis (HP and AHP) in the temperature range of 300–450 °C. Next, soluble organic matter (OM) and the liquid yield were removed from the pyrolyzates after each step of thermal maturation. N2 adsorption tests were performed to assess the pore structures of the samples. Furthermore, deconvolution and fractal dimension analyses were executed to unveil pore clusters and discern the complexity of the pore network within each sample. Finally, two tree-based machine learning algorithms, random forest (RF) and extra trees (ET) were applied to model the N2 adsorption/desorption data of pyrolyzates to predict pore structure variations. Based on the results, HP pyrolyzates had higher bitumen reflectance (BRo%) values than AHP ones at all temperature sequences, which confirms water controls thermal maturity. AHP pyrolyzates exhibited larger average pore diameters across all temperatures, while HP pyrolyzates displayed higher BET surface areas in contrast to AHP pyrolyzates. Also, the average pore diameter of HP pyrolyzates with soluble OM and liquid yield extracted did not change significantly during thermal maturation, while the average pore diameter of AHP ones increased, peaking at 400 °C in post-mature stage. The largest total pore volume of HP samples was observed at the end of the wet gas window, while this was observed for AHP samples in the dry gas window. Three mesopore and four macropore clusters were identified in the original sample. Moreover, pore clusters of the pyrolyzates underwent diverse changes during thermal maturation, without following a specific trend, influenced by both temperature and pyrolysis conditions. Abundance of micropores, finer mesopores (2–10 nm), and a smaller average pore diameter in HP pyrolyzates compared to the AHP is the main reason for the higher fractal dimensions showing a more complex pore structure and/or rougher pore surface. Intelligence modeling exhibited that RF outperformed ET in estimating N2 adsorbed/desorbed volumes for the samples. Ultimately, sensitivity analysis revealed that water exerted a more significant influence on pore development in shales during thermal progression, outweighing the impact of temperature.