Inversion Procedure Of Meso-parameters For Geotechnical Materials Based On Macroscopic Experimental Data

Geotechnical materials differ from general continuum at macro-scope which approach assemblies of particles at micro-scale. Cracks, holes and structural planes usually exist in geotechnical materials. The discrete element method(DEM) was introduced in 1971 which simulates the force, deformation, interaction and movement of circular particles at micro-scale, DEM is widely used and further studied in many fields. PFC2D(Particle Flow Code in 2 Dimensions) is developed based on DEM and has been applicable widely in many fields of geotechnical engineering such as tunnel excavation, foundation pit support, slope stability, geotechnical experiment simulation, rock fragmentation, seepage phenomenon, dynamic load problem. The accuracy of particle flow simulation mainly depends on meso-parameters. Meso-parameters are usually calibrated by a "try-error-try" method which bears high occasionality and seems lacked in theory. Calibration of meso-parameters remains difficult.Taking account of the incompleteness of the general calibration method mentioned above, this article combined experimental design, response surface method(RSM) with BFGS(Broyden Fletcher Goldfarb Shanno) algorithm and put forward an inverse procedure of meso-parameters for geotechnical materials. The rockfill materials and conditioned soils are simulated by using linear contact model and parallel bond model, respectively. According to triaxial compression experimental data of rockfill materials and conditioned soils, meso-parameters of geotechnical materials were calibrated firstly and a number of particle flow simulations were designed. The nonlinear relationship between macroscopic deviatoric stress and meso-parameters of geotechnical materials was created by using RSM and inversion of meso-parameters was conducted based on BFGS algorithm. Fragmentation process of rock under indenter was simulated by using the estimated meso-parameters. The simulated results were in qualitative agreement with the published literatures. The investigation shows that meso-parameters inversion of geotechnical materials which combines RSM with BFGS algorithm based on macroscopic experimental data has better predicted precision and it’s appropriate to simulate rock by using parallel bond model.