Automatic Identification Of Mesoscopic Parameters Of Discrete Element For Sand Material

Sand material is widely used in civil engineering.Traditional theoretical research and engineering design methods are based on the theory of continuous medium to study sand.They do not consider the size,shape of particles,the relationship between contact forces between particles and the arrangement of particles,etc.Therefore,the study of sand based on the theory of continuous medium mechanics is inconsistent with the reality.Discrete element method is a discontinuous numerical simulation method.Compared with traditional numerical methods,it can more reasonably describe the dispersion of particle size,physical properties and shape,as well as the transient,fluctuation,collision,particle cohesion,and the rupture and fragmentation behavior of aggregates.It is especially suitable for the study of discontinuous deformation of soils.In the discrete element theory,accurate meso-parameters are the premise of reasonable simulation using discrete elements.However,the precise theory of describing the relationship between meso-parameters and macro-parameters of discrete element method is not perfect,and the functional relationship between macro-parameters and Micro-Parameters of sandy materials has not been established.At present,the selection of meso-parameters of discrete element simulation has considerable subjectivity and uncertainty,which seriously affects the reliability of the analysis results.Therefore,accurate determination of meso-parameters has become a key problem to be solved urgently in the application of discrete element method.Based on the optimization theory,an automatic identification method for the meso-parameters of sand materials by discrete element method is proposed in this paper.Firstly,PFC discrete element software is used to simulate the biaxial compression test of sand,and the sensitivity of the meso-parameters is analyzed.It is concluded that the friction coefficient of sand particles and the contact stiffness of sand particles are more sensitive to the macro-parameters.Then,on the basis of the biaxial compression numerical test,the error between the stress-strain curve of PFC numerical simulation and the biaxial compression experimental curve of a standard sandy soil is minimized as the optimization objective,and the mode search method is used to realize the automatic calibration of the meso-parameters of the discrete element of sandy soil material.Then,40 sets of Micro-Parameters and their corresponding macro-parameters are automatically identified by pattern search method.The mapping relationship between macro-parameters and Micro-Parameters is found by BP artificial neural network analysis method.Within the error range of 10%,the simulated stress-strain curve and the simulated stress-strain curve have higher fitting degree.Finally,based on the triaxial compression test of a highway slope soil in laboratory,the PFC simulation and stability analysis of the highway slope are carried out by using the identified meso-parameters.