Discrete Element Modelling Of The Wheat Particles And Short Straw In Cleaning Devices

Cleaning is a key component in mechanized harvesting of cereals,and therefore the working performances of cleaning devices have a significant influence on loss rate and impurities rate during harvesting processes.With the development of computer technologies,numerical simulation techniques have been widely used in the research of cleaning process.Especially in recent years,more and more researchers used the CFD,DEM and CFD-DEM to study the distribution of airflow in cleaning devices and the cleaning-separation process of cereals with and without airflow,which have greatly boosted the developments of separation theory and relevant equipment.Although the numerical simulation techniques are simple and fast,they are highly dependent on the established models,the accuracy of which affects the simulation results greatly.The present research on modelling of threshed mixture was still insufficient,more work was needed in the study of simulation parameters,contact models and particle models.In view of the shortcomings mentioned above,explorative studies were performed based on:(1)the direct measurements and virtual calibrations of DEM simulation parameters of the wheat particles and short straw in a cleaning device;(2)the construction of visco-elasto-plasticity contact model of wheat particles;(3)flexible straw model.The main research contents and conclusions of this study are listed as follow:The direct measurement method was applied to measure the physical properties(which required in DEM simulation)of the wheat particles and short straw which collected from the4LZ-6B1 self-propelled combine.Statistical analysis methods,such as data distribution test,singular value check and variable coefficient,were used to analyze the measured data,in order to provide a guidance for the modelling of the wheat threshed mixtures.Results show:(1)the three axis dimensions of the wheat particles obey normal distribution,thus the particles could be generated in a normal distribution;(2)the length of the straw weighed significantly,thus the straw length must be considered in simulations;while the difference of external diameter was small,thus a single external diameter could be used in order to simplify the modelling processes;(3)the pycnometer method(Standard:ASTM D854-10)could well measure the true density of the wheat particles;(4)the elastic/shear modulus of wheat particles can be well measured by the standard of American Society of Agricultural Engineers(S368.4,DEC2000,ASAE,R2012),while the measured values of the short straw had a large variable coefficient which may comes from the difference of wall thickness;(5)the wheat-wheat restitution coefficient measured from double pendulum test had large variable coefficients within and between the groups,which showed that this method had a bad repeatability when used to measured this parameter;(6)the three-point slipping test could well measure the wheat/straw-straw static friction coefficient;(7)the wheat/straw-straw static friction coefficient measured from particles slipping on inclined plate(straw sticks together)had large variable coefficients,thus this method had a bad repeatability;(8)the rolling friction coefficient can hardly be directly measured,thus parameter analysis and virtual calibration was required.(2)The Plackett-Burman design was used to screen the significant simulation parameters of the 5-sphere wheat model in two typical angles of repose tests(cylinder lifting and rotating drum).The Box-Behnken design was used to study the effect of calibration tests on calibrated parameters qualitatively and quantificationally.The results show that the two calibration tests had significant influence on calibrated parameters.Based on this,the rotating drum test was selected and 19-sphere wheat model was created considering the kinestate of mixture in cleaning devices.Based on the Box-Behnken design,a quadratic regression model between the dynamic angle of repose and the screened significant parameters was developed.Tthe optimal combination of wheat-wheat static friction coefficient,wheat-steel static friction coefficient and wheat-wheat rolling friction coefficient were 0.18,0.23 and 0.043.The calibrated parameters were validated by the rotating drum test and vibrating screen test,which shows that these parameters can well simulate the dynamic angle of repose and distribution of wheat under screen.Moreover,the Plackett-Burman design was performed to screen significant simulation parameters of the two wheat models(5-sphere and 19-sphere)in vibration screening processes.The results show that the wheat-wheat static friction coefficient,wheat-wheat restitution coefficient and wheat-steel static friction coefficient had significant influence on mean velocities of the both models;while other parameters exhibited little influence.Furthermore,the influence of wheat-wheat static friction coefficient was much greater than the other two parameters,thus special attention must be paid to determine this parameter.(3)Based on the existing contact model,a visco-elasto-plasticity contact model of wheat was created by the API of EDEM software and C++language.The simulation parameters required in this contact model was calibrated and then validated by the distribution of wheat under screen.Based on this contact model and calibrated parameters,the influence of vibrating parameters on the motion of wheat particles was studied under two feeding rates.The results show that the increase of feeding rate would decrease the mean velocities of wheat particles and therefore the feeding rate must be considered in the study of cleaning processes.Meanwhile,with the increase of amplitude and frequency,the mean velocities of wheat particles increase almost linearly,while the difference between the two feeding rates decreased;the vibrating direction angle had little influence on particle velocities under 10000/s,while the particle velocities increased and can be well fitted by a quadratic polynomial equation under 20000/s;compared with amplitude and frequency,screen slope had small influence,while the particle velocities with screen slope showed an opposite trend.(4)Based on Hertz-Mindlin with bonding contact model,a segment flexible straw model was developed by the API of EDEM software and C++programming language.The three-point bending test was in sensitivity analysis of the bonding parameters,which shows that the elastic modulus of straw was increased with the bonding disk radius,normal and shear bonding stiffness.Based on Box-Behnken design,a regression between the three bonding parameters and the elastic modulus was developed and optimized.The optimal combination of these parameters was 2.57 mm,2.35×10~9 N/m~3 and 7.91×10~9 N/m~3.Based on EDEM-Fluent coupling method,the motion of rigid and flexible straw models with different lengths and initial postures in horizontal air channel were simulated and compared with experimental results.The results show that the flexible straw model developed in this work can better simulate the X-direction displacement and would not be influenced by the straw length(the experimental results show that the straw length had insignificant on its X-direction displacement).The flexible straw model can also simulate the posture changes caused by the airflow.Furthermore,the flexible straw model solved the mesh size limitation(the CFD-DEM coupling method required the mesh size not smaller than the particle size).