Impact of Compositions on the Fracture Behavior of Shale: A Numerical Simulation Study Using an Improved Finite Discrete Element Method

Abstract

Hydraulic fracturing is a key technology for the efficient development of shale oil and gas resources, while mineral composition and its spatial organization can substantially affect crack initiation, propagation, and coalescence within shale. This study proposes an improved finite-discrete element method (FDEM) modeling approach based on statistical reconstruction of mineral-mapping data. Phase fractions, correlation length, anisotropy, and local clustering characteristics are extracted from mineralogical images to generate millimeter-scale, statistically equivalent structures comprising clay, felsic minerals, carbonates, kerogen, and natural pores. OpenFDEM is then used to simulate their fracture behavior under uniaxial compression. Statistically constrained models are first constructed from mineral-mapping data for the Gulong and Ordos shales, followed by a series of idealized models designed to investigate the effects of phase fractions, spatial clustering, and pore distribution. The results show that independent structures generated using the same statistical parameters exhibit similar macroscopic load-bearing responses while retaining distinct local fracture paths. Carbonate enrichment promotes earlier main-fracture formation and greater accumulated fracture, whereas increasing the felsic fraction generally suppresses interface failure. Kerogen-related boundaries promote local branching, while natural pores connect discrete microcrack clusters and facilitate their coalescence. This approach links shale composition, spatial structure, and fracture-network evolution at the mesoscale, providing a numerical basis for understanding the fracture mechanisms of shales with different compositions.