A Fast Convolution‐Based Peridynamics for Rock Fracture Under Compressive Load

ABSTRACT The fast convolution‐based method for peridynamics (FCBM‐PD) is an efficient approach for solving fracture propagation problems. However, current FCBM‐PD method fails to distinguish between tensile and compressive strains at material points, limiting its capability to accurately simulate fracture propagation under compressive loading. To address this issue, the spectral decomposition method is employed to obtain strain invariants, and tensile strains are extracted by using strain decomposition. By utilizing the separated tensile strains, a bond failure criterion is reconstructed, resulting in a damage model capable of capturing the tension‐compression asymmetry of geomaterials. Additionally, an initial integrity factor is introduced to correct unrealistic damage values near the initial fracture faces, which arise even in the absence of fracture propagation. A modulus reduction technique borrowed from traditional damage mechanics is applied to mitigate the influence of surface effects on fracture propagation. The FCBM‐PD method is extended for the first time to address compressive loading scenarios and validated by experimental data from a central‐initial fractured sandstone specimen under uniaxial loading. A comparison of computation time and memory requirements between FCBM‐PD and traditional peridynamics (PD) methods demonstrates that the proposed method significantly reduces computational cost. To further demonstrate the performance of the proposed method, five additional numerical examples on rock fracture under compressive loading are presented, confirming that the proposed FCBM‐PD method can effectively simulate rock fracture initiation and propagation under compression, and therefore, showing potential for large‐scale geotechnical engineering problems.