A Simulation Study Of The Phenomenon Of Misclassification Of Walnuts In A Fence-type Drum Classifier

With the production of walnuts increasing year by year,the traditional way of selling walnuts as a commodity has brought about disadvantages such as increased storage and transportation costs and low utilisation rate of the commodity,while primary and deep processing of walnuts can improve this problem,of which shelling is an important part of the primary and deep processing of walnuts.The size of the walnuts falling into the gap and the degree of matching of the gap will directly affect the shelling effect and production efficiency.The grading method is divided into physical grading and photoelectric grading,of which physical grading has the advantages of low cost and high productivity,and is the main grading method in the walnut processing.The grating roller classifier is one of the most important grading methods of physical grading,and is widely used for grading and sorting agricultural materials because it is convenient and quick,and has fewer consumables.However,there is a problem of "mixed grades" in the grading of walnuts in the grading process,and the traditional "black box" analysis method has limitations that restrict the development of the walnut industry to a certain extent.The discrete element method has the advantages of low cost and short cycle time,visualisation and real-time detection of microscopic information,and has been widely used in the analysis of agricultural materials and the optimisation of machinery and equipment design.Therefore,it is important to investigate the distribution and separation characteristics of walnuts in the grading process based on the discrete element method,in order to clarify the mixing characteristics of walnuts and to obtain the optimum grading parameters for the grading of the grating drum.In this study,the physical and contact parameters of walnuts were first measured,followed by the development of a discrete element model of walnuts and the calibration of the measured results,and then the verification of the calibration results based on cylinder lifting and grading tests.Then,based on numerical simulations,the influence of drum speed and drum inclination on the circumferential and axial distribution of walnuts in the drum and the influence of feed rate and drum length on the axial separation of particles of different sizes were investigated,taking the grading principle of the grating drum as an entry point.Finally,based on the Box-Behnken test,the effects of drum speed,drum length and drum inclination on classification accuracy and classification efficiency were investigated,and a regression model was established to find the optimum parameters based on the regression model.The main findings of this study are:(1)Firstly,the moisture content of walnut kernels,walnut shells,walnut density,walnut shear modulus and stacking angle were measured by drying test,liquid discharge test,uniaxial compression test and cylinder lifting test respectively.Next,the measurable contact parameters: walnut-walnut collision recovery coefficient,walnut-steel plate collision recovery coefficient and walnut-steel plate rolling friction coefficient were measured by the collision test and the inclined roll-off test.The Plackett-Burman test was then used to analyse the coefficients that significantly affect the magnitude of the stacking angle.The results show that the walnut-walnut static friction coefficient,the walnut-walnut rolling friction coefficient and the walnut-steel plate static friction coefficient significantly affect the walnut stacking angle,while the remaining contact parameters have a minor effect on the walnut stacking angle.Then,the steepest climb test was used to quickly determine the optimal value interval of the three significance parameters,and the walnut-walnut static friction coefficient,walnut-walnut rolling friction coefficient and walnut-steel plate static friction coefficient were calibrated by Box-Behnken test.Finally,three sets of parallel tests were conducted to verify that the relative errors were small and very close to the physical stacking angle size and shape,indicating that the calibration results were accurate and valid.(2)The separation index gradually decreases with increasing drum speed,indicating that the distribution of walnuts is becoming more and more uniform.The separation index decreases slightly and then increases sharply with the increase in drum inclination,i.e.the walnut particles are first evenly distributed and then gradually deviated.The coefficient of variation increases and then decreases as the drum speed increases,indicating that the walnut particles are first aggregated,then dispersed and then aggregated.The coefficient of variation of the axial position of the particles tends to decrease as the drum inclination increases,indicating that the particles are increasingly aggregated in the distribution under this condition.As the feed rate increased,the Hellinger distance showed a tendency to fluctuate and then decrease.As the drum length increases,the Hellinger distance tends to increase sharply and then increase gently.(3)The influence of three factors on the grading effect of the grating drum was analysed.The results showed that,in terms of grading accuracy,the quadratic term of drum speed,the interaction term of drum speed and inclination angle had no significant effect on it,while the remaining parameters had a significant effect on it.In terms of classification efficiency,the interaction terms of drum speed and length,and drum speed and inclination have no significant effect on the classifying efficiency,while the other factors have a significant effect on it.The optimum combination of parameters was obtained by fitting a model and the optimisation results were verified.