Investigation Into Calibration Method Of Discrete Element Model Parameters For Bulk Material And Its Application

As a kind of discontinuous medium,the granular material is widely found in many fields such as coal mining,bulk material processing,rock and soil mechanics,powder engineering.The discrete element method(DEM)provides a convenient way to study the complex behavior of bulk materials,and has been widely used.The macroscopic behavior characteristics of bulk materials are closely related to their micro properties.However,it’s the tedious to calibrate parameters of particles in the discrete element numerical simulation,and the calibration methods are varied.None of relative calibration standard is proposed at present.It directly affects the simulation efficiency and the reliability of simulation results.In this paper,the discrete element software EDEM was used as the numerical simulation platform,and based on experiments,a new method to realize the rapid and accurate calibration of discrete element parameters of bulk materials was proposed.At first,samples space was constructed by adopting the optimal latin hypercube design Method(Opt LHD).And then,based on the back propagation(BPNN)neural network model,the mapping relation between the two was constructed.The next,combined with feasible field search algorithm developed autonomously and cluster analysis,discrete element parameters that reflect a special bulk material macro behavior can be obtained.Finally,an experiment is conducted to verify the calibration results.The main research works are as follows:(1)Based on the literature and engineering practice,the relevant criterion of calibration experiments was worked out,according to which,the calibration experiment scheme was determined.Bulk density,static angle of repose and dynamic angle of repose were used to characterize the macroscopic behavior of bulk material,which provided necessary data for the calibration.The effects of the period boundary condition and discrete element parameters for the macroscopic behavior of bulk material were researched,and results showed as follows.It is time saving to simulate the calibration experiments by using the period boundary condition,but the changes of the simulation results are little.Particle density,rolling friction coefficient and static friction coefficient are the main factors to effect the macroscopic behavior of bulk material,and all of them are the discrete element parameter calibrated.The calibration experiments and discrete element parameters were studied,which laid the foundation for the calibration.(2)The optimal latin hypercube design Method(Opt LHD)was used to construct the samples space.Based on these simulation data,a back propagation(BPNN)neural network model was trained,studied and optimized,and the mapping relation between the two was constructed.For calibrating parameters,a feasible field search algorithm was developed,and its effectiveness was verified.Combined with cluster analysis,the parameters calibration was realized.And a calibrating software was developed based on flow of calibration.Compared with the generic EDEM material model(GEMM),the software can meet the engineering demand better.However,owing to the limited bulk material information provided by calibration experiments,the results show varied and limited,which indicates the necessity of verification.(3)The calibration method and the numerical simulation were combined to study the motion law of bulk material in the vertical screw conveying.And it is watched that the rotation velocity of particle is relative to both radius and height,and a phenomenon of stratified flow similar to fluid is observed.According to which,a new motion model was proposed.The fitting results show that the new model can reflect the motion law of most particles more accurately,and the relative error is under 5%.Furthermore,the coefficients of the model are approximately linear with the rotation speed of the screw shaft.