Micro-parameter calibration of layered rock samples with a dual model in PFC3D based on a combined optimization method

A combined optimization method termed PB-RSM-NSGA2 was proposed to investigate micro-parameter calibration of layered rock samples with a dual model in Particle Flow Code (PFC). Firstly, the numerical uniaxial compression test with the layered rock samples was conducted, in which the parallel bonding model and the smooth joint model were used to simulate the rock matrix and the interlayer cementation, respectively. The micro-parameters with significant influence on macro response were screened out by the Plackett-Burman (PB) test design. Subsequently, non-linear relationships between macro-parameters and their main controlling micro-parameters were established via the response surface method (RSM) and model verification. Thirdly, the non-linear multi-objective mathematical optimization problem was solved by the Non-dominated Sorting Genetic Algorithm-II (NSGA2), and the micro-parameters were determined. Finally, the uniaxial compression stress-strain curves were modified with a compaction coefficient and compared with the physical test results. The research results revealed that the dual model was highly suitable for simulating the mechanical properties of layered rocks. In the smooth joint model, normal stiffness, tangential stiffness, tangential strength, and bedding thickness significantly impacted layered rock samples' interlayer strength and deformation. The parallel bond model's micro elastic modulus and normal bonding strength significantly influenced the uniaxial compressive strength and peak strain. The corrected simulation curves agreed with the physical test results. The study can provide valuable insights and methods for micro-parameters calibration in PFC 3D numerical simulations for layered rock mass engineering stability.