Research on the influence of rock micro-parameter in particle flow code (PFC) on the rock fracture morphology

Abstract

Particle Flow Code (PFC) has been widely used in simulation of geotechnical engineering. In the simulation of laboratory test of rock and soil, the generated fracture appears different distribution, which is significantly affected by the micro-parameters. The generated fracture morphology is a suitable constraint to suppress the non-uniqueness of the micro-parameters. In this research, 1517 rock mass models with different micro-parameters are established, and the corresponding uniaxial compression test, Brazilian splitting test, and direct shear test are simulated. The simulation results are classified into multiple groups according to the distribution of the generated fractures. The relationship between the generated fracture morphology and the micro-parameters are researched. In addition, a method for evaluating the fracture morphology based on K-nearest neighbors is proposed, and it is verified by 40 rock numerical model with random micro-parameters. The evaluating accuracy of the uniaxial compression test, the Brazilian splitting test, and the direct shear test is 82.5%, 82.5%, and 85%. The proposed method is helpful for evaluating the generated fracture morphology to save the time cost of micro-parameters calibration, and have potential on suppressing the non-uniqueness of the micro-parameters in the calibration process, and establishing accurate numerical rock model.