Simulation of flip-flow screening adhesive organic fertilizer particles based on DEM-MBD coupling method

For screening adhesive organic fertilizer particles, a Discrete Element Method (DEM) and Multi-Body Dynamics (MBD) coupling model of screening adhesive organic fertilizer particles using a flip-flow screen is established. Then, the velocity, the distribution and the trajectory of the particles during the screening process are observed. Finally, the effects of the surface energy γ, the rotational speed n, the tensional amount ∆l and the feed rate M are investigated. The results show that the flip-flow screen could provide a high velocity for depolymerization of agglomerated particles and separation of adhesive particles from the screen panels, so adhesive organic fertilizer particles can be successfully screened by using the flip-flow screen and organic fertilizer particles in an easily absorbed range are obtained. With the increase of γ, both the flow rate and the screening efficiency decrease. With the increase of n, both first increase and then slightly decrease. With the increase of ∆l, both increase at a low n, or slightly decrease at a high n. With the increase of M, the screening efficiency decreases, while the total flow rate first increases and then decreases. Through adjusting n, ∆l, M, flip-flow screen can also be used to screen other adhesive particles.


Introduction
As possessing high moisture and high viscosity, adhesive organic fertilizer particles not only easily agglomerate together, but also easily adhere to and block the screen apertures, so they are difficult to be screened using conventional screening technology [1].However, a flip-flow screen can provide agglomerated particles with high velocity for depolymerization, so it has become an effective solution for screening adhesive particles.
Thus, some scholars have begun to study the effect of screening adhesive particles using the experimental method [2].Li et al [3] propose a flip-flow screen with crankshaft-link structure for screening moist coal.The experimental results show that the screening efficiency can exceed 85.82%, and the screening efficiency first increases and then decreases with the increase of the rotational speed.Yu et al [4] propose a vibrating flip-flow screen for deep screening moist minerals.The experimental results show that a high screening efficiency of up to 89.05% could be obtained.Peng et al [5] develop a novel centralized-driving flip-flow screen for separating moist coal and investigate the effects of the rotational speed and the tensional amount.The experimental results show that the flip-flow screen has a high screening efficiency and the rotational speed and the tensional amount have a significant effect on the kinematic characteristic.To avoid screen blockage during screening adhesive materials, Zhao et al [6] design a double-deck vibrating flip-flow screen and study its dynamic characteristics.The experimental results show that the screening efficiency could reach 97.09%.
Although the above experimental method is in favor of analyzing the screening effect, it is not in favor of observing the screening process, such as the trajectory and the velocity of particles, the collisions among particles and the impact between the particles and the screen panels.Therefore, a few scholars have begun to simulate the screening process using the DEM method.Davoodi et al [7,8] used the DEM method to simulate the process of screening dry particles and investigate the effect of the particle density.The simulation results show that a large particle density is favorable to a high screening efficiency.However, as the feed rate increases, a thick particle layer begins to build up, resulting in a decrease in the screening efficiency.Liu et al [9] used the DEM method to simulate the process of screening dry particles.The simulation results show that a high screening efficiency could be obtained when the inclination of discharge is 10°and the increment of the screen panel inclination is 5°.In addition, the screening efficiency would not significantly increase when the total length of screen panels exceeded 430 mm.Harald et al [10] used the DEM method to simulate the process of screening non-spherical particles.The simulation results show that the particles' shapes and sizes have a great influence on the screening efficiency.Zhao et al [11] used the DEM method to simulate the process of screening irregular rock particles.The simulation results show that the frequently occurring screen blockage results in a decrease in the screening efficiency during linear vibrating or circular vibrating screens.Chen et al [12] establish a DEM model of a LIWELL flip-flow screen, observe the trajectory of one particle, and investigate the influence of the screen panels acceleration.Zhang et al [13] establish a dynamic model of a flip-flow screen by DEM and find that the screening efficiency increases nonlinearly with the decrease of the eccentricity, the inclination angle and the size fraction.
However, the simulated screening objects in the above literatures are all dry particles.Only a few literatures research on simulating screening adhesive particles.Yu et al [14] establish a DEM model of a flip-flow screen, simulate screening adhesive particles, and observe the flip-flow screening process.The simulation results show that agglomerated particles are depolymerized and the screen blockage problem is solved.Furthermore, adhesive organic fertilizer particles usually have higher moisture and higher viscosity than other adhesive particles, such as moist coal, so they are more difficult to be screened.However, screening adhesive organic fertilizer particles has rarely been researched until now, either using the experimental method or using the DEM method.In addition, a rigid(non-flexible) screen panel can be established in only a DEM model, complex motion is difficult to be performed and the impact of the particles on the screen panels cannot be realized.
Therefore, for screening adhesive organic fertilizer particles, a Discrete Element Method (DEM) and Multi-Body Dynamics (MBD) coupling model of screening adhesive particles using a flip-flow screen is established.In this model, a complex motion, the expanding and contracting motion, is performed, and both the collisions among particles and the impact between the particles and the screen panels are realized.Then, the velocity, the distribution and the trajectory of the particles during the screening process are observed.Finally, the effects of the surface energy γ, the rotational speed n, the tensional amount Δl and the feed rate M are investigated.

Adhesive organic fertilizer particles
Adhesive organic fertilizer particles are produced by rotary drum granulation.Their diameter ranges from 0 to 14 mm, and their gradation is acquired after classification, as shown in figure 1 and table 1 [15][16][17].In it, organic fertilizer particles in the range of 0-3 mm are most easily absorbed by plants, so it is essential to screen them and acquire the easily absorbed particles.
Their moisture content usually ranges from 20% to 30%, as shown in figure 2. As the moisture content increases, both their viscosity and their surface energy γ increase, leading to a increase of agglomeration particles both in number and in volume.Thus, adhesive organic fertilizer particles are difficult to be screened using conventional screening technology, so the flip-flow screen is employed as it can provide agglomerated particles with high velocity for depolymerization.

Johnson-kendall-roberts(JKR) contact model
For simplicity, all the adhesive organic fertilizer particles are assumed to be homogeneous and spherical, no matter what diameter they are, as shown in figure 3.
The contact among adhesive particles is defined by using the JKR model.As shown in figure 4, a particle of radius R 1 adheres to a particle of radius R 2 .Under the action of the adhesive force, their contact radius increases from a to a 0 .The new contact radius a 0 , the overlap amount δ n and the equivalent contact radius R * can be calculated by equations (1)-(3): * Where γ represents the surface energy of particles, E * represents the equivalent elastic modulus (N m −2 ), which can be calculated by equation (4): Where E 1 and E 2 stand for the elastic modulus of Particle 1 and Particle 2, respectively.v 1 and v 2 stand for the Poisson's ratio of Particle 1 and Particle 2, respectively.
The adhesive force W (J m −2 ) can be calculated by equation (5): Where γ 1 and γ 2 represent the surface energy of Particle 1 and Particle 2, respectively (J m− 2 ), and γ 12 represents the interface energy between the two particles (J m− 2 ), which is equal to 0 when the two particles are of the same material.In other words, when γ 1 = γ 2 = γ, γ 12 = 0 and W = 2γ.
The normal elastic contact force F JKR based on the Hertz-Mindlin with JKR contact model can be obtained by equation (6) [18][19][20]:

DEM-MBD coupling model
RecurDyn software, as a MBD software, can perform complex motion by applying motion pairs, flexible joints and drivers.While, EDEM software, as a DEM software, can generate many particles and calculate their collisions and motions.Through combining EDEM software and RecurDyn software, a DEM-MBD coupling model is established for simulating the flip-flow screening process.Firstly, a three-dimensional (3D) model of     the flip-flow screen is created by SolidWorks.Secondly, RecurDyn software is used to convert the 3D model into an interchangeable graphic file format for the MBD model.Thirdly, EDEM software is adopted to create a DEM model imported from the MBD model.Finally, the DEM-MBD coupling model of screening adhesive particles using a flip-flow screen is established.By using the DEM-MBD coupling model, the expanding and contracting motion of the flip-flow screen is performed, and both the collisions among particles and the impact between the particles and the screen panels are realized.The flip-flow screen is composed of five screen panels with the dimensions of 250 mm × 150 mm × 5 mm, each screen aperture is 15 mm × 4 mm with a space of 5 mm.The screen panels are made of polyurethane, which is a kind of superelastic material.
In RecurDyn software, the Mooney-Rivlin Material model with two parameters (C 1 and C 2 ) is used to define the superelastic characteristics [21,22], and it is expressed as [23]: where U is the potential energy of strain, I 1 is the first-order invariant of the deformation tensor, I 2 is the second- order invariant of the deformation tensor, d is the incompressible parameter and J is the volumetric ratio.In this study, the hardness of the screen panels is 70HA, C 1 = 0.46 MPa, and C 2 = 0.12 MPa [24], more details about the material properties are shown in table 2. The five screen panels are meshed with Solid8 (Hexa8), as shown in figure 5.
All the screen panels perform the same periodic cosine motion, and two adjacent screen panels have a phase difference of 180°.
Translational motion pairs are applied on both sides of the five screen panels.The translational displacement of the 1st screen panel, the 3rd screen panel and the 5th screen panel D 1 is expressed as: While, the translational displacement of the 2nd screen panel and the 4th screen panel D 2 is equal in magnitude but opposite in direction, it is expressed as: Where n is the rotational speed, e is the eccentricity of shaft, Δl is the tensional amount and t is time.
The Hertz-Mindlin with JKR contact model is used to define the contact among adhesive particles [25, 26] and the Hertz-Mindlin(no-slip) contact model is used to define the contact between the particles and the screen panels, and their contact parameters are shown in table 3.
As shown in figure 6, when the DEM-MBD coupling model is underway, RecurDyn transmits the translational data and the rotational data of the screen panels to EDEM, EDEM calculates the displacements and velocities of particles and the forces and the torques of particles to the screen panels and then tranmits them to RecurDyn.At the next step, RecurDyn calculates the new translational data and the rotational data of the screen panels.Through transferring data circularly and interactively, both the motion data and the collision data are updated, so that bidirectional coupled computing is finally achieved [27][28][29][30].
As shown in figure 7, the particle factory generates particles at a given feed rate, and the particles fall to the 1st screen panel.Then, the particles agglomerated together due to the adhesive force among particles.The screen panels provide a high velocity for depolymerization of agglomerated particles and separation of adhesive particles from the screen panels, so the undersized particles have the opportunity to pass through the screen apertures.Meanwhile, five receivers are placed under the five screen panels to collect the undersized particles.While the oversized particles and some unscreened undersized particles will finally enter the discharging end.From figure 7 it is observed that adhesive organic fertilizer particles can be successfully screened by using the flipflow screen, and most of them are screened by the first few screen panels.

Model verification
To evaluate the accuracy of this simulation model, a translational motion experiment of a single screen panel is conducted using a hyperelastic polyurethane screen panel with the dimension of 250 mm × 150 mm × 5 mm, as shown in figure 8.A camera is employed to take photographs of the screen panel.After conversion between the image dimension and the real dimension, the real location of the screen panel is determined and the experimental trajectory is obtained.
Meanwhile, a translational motion simulation is also conducted using RecurDyn software.As shown in figure 9, a screen panel with the same dimension is meshed with Solid8 (Hexa8), and 15,453 nodes and 7828 elements are obtained.The translational motion pairs are applied to both sides.By the translational motion simulation, the location of the screen panel is acquired and the simulation trajectory is obtained.Moreover, the theoretical trajectory is also calculated by a classical elastic compression bar theory [31].The translational motion is carried out when the rotational speed, the tensional amount and the inclination angle are 60 r min −1 , 0 mm and 20°, respectively.The eccentricity is set to 6 mm, so the slack amount ranges from 0 to 12 mm.Next, the simulation trajectory is compared with the experimental trajectory and the theoretical trajectory, as shown in figure 10.Obviously, the simulation trajectory has the same trend as the experimental and theoretical trajectories.As the slack amount increases from 0 to 12 mm, the screen panel drops slowly.Moreover, the simulation trajectory agrees well with both the theoretical trajectory and the experimental trajectory and the maximum error does not exceed 3.8 mm, which verifies a high prediction accuracy of this simulation model.

Results and discussion
Based on the DEM-MBD coupling model, screening adhesive organic fertilizer particles is performed when the inclination angle and the eccentricity the feed rate are 20°and 6 mm, respectively.

Observation of screening process
After reaching a steady state, the velocity, the distribution and the trajectory of the particles during the screening process are observed when the rotational speed, the tensional amount, the surface energy and the feed rate are 240 r min −1 , 4 mm, 0.3 J m −12 and 20 kg min −1 , respectively.

Particle velocity during the screening process
Particles are generated by the particle factory and then fall to the 1st screen panel.Under the action of gravity, the falling velocity gradually increases, as shown in figure 11.Then, particles are thrown up by the screen panel in the front and then fall to the screen panel at the back.Particles closer to the midpoint of the screen panel are provided with high velocity, which is favorable to depolymerization and separation.While, the particle velocity at two sides of the screen panels (namely at junctions of two adjacent screen panels) is too small, so some particles stay motionless and agglomerate together.In addition, many small arrows appear under the five screen panels, especially the first few screen panels, which means undersized particles pass through the screen apertures.Meanwhile, some arrows appear at the discharging end, which means oversized particles and some unscreened undersized particles finally enter the discharging end.

Particle distribution during the screening process
Next, the 2nd screen panel is chosen for observing the particle distribution, as shown in figure 12.In a period (7.525 s-7.775 s), the 2nd screen panel expands once (from figures 9(a)-(d) and contracts once (from figures 12(e)-(i).When the distance between the two sides of the screen panel reaches the minimum value, many particles accumulate and agglomerate on the screen panel, as shown in figure 12(a).In the expanding process, the distance between the two sides of the screen panel gradually increases, and the screen panel state changes from relaxed to tensioned.At this moment, the particles closer to the midpoint are thrown up and become scattered.While, the particles at two sides are hard to be thrown up due to the small provided velocity.In the contracting process, the distance between the two sides of the screen panel gradually decreases, and the screen panel state changes from tensioned to relaxed.The particles first continue to go up for an instance and then fall to the screen panels.As thrown up by the screen panels and provided with a high velocity, agglomerated particles are depolymerized and adhesive particles are separated from the screen panels, so undersized particles successfully pass through the screen apertures.through the screen apertures directly.While, a lot of undersized particles will first collide with other particles (Particle 2) or slide along the screen panels (Particle 3), and then pass through the screen apertures.Some undersized particles (Particle 4) unfortunately fail to pass through due to the improper falling points, although   they have contacted the screen panels for many times.Finally, all the oversized particles (Particle 5) and some unscreened undersized particles (Particle 4) will enter the discharging end.

Influence of surface energy, rotational speed, tensional amount and feed rate
Next, the effects of the surface energy γ, the rotational speed n, the tensional amount Δl and feed rate M are investigated, and their parameters are selected according to the practical engineering application, as shown in table 4.  To achieve both a large total flow rate and a high screening efficiency, high surface energy needs to be decreased before screening.

Influence of rotational speed n
When n = 160 r min −1 , adhesive organic fertilizer particles cannot be thrown up easily (figure 16(a)) and agglomerated particles cannot be provided enough high velocity (figure 17(a)) for depolymerization and separation, leading to a thick particle layer (figure 16(a)), a large accumulation mass (figure 17(b)), a small total flow rate (only 3.4 kg min −1 ) and a low screening efficiency (only 47.94%) (figure 17(f)).When n increases from 160 r min −1 to 360 r min −1 , the maximum particle velocity approximatively linearly increases (figure 14(a)), leading to a rapid decrease of the accumulation mass of particles on every screen panel, especially on the 1st screen panel (figure 16(b)).This is because after being thrown up by the screen panels and provided with enough high velocity, agglomerated particles are depolymerized and adhesive particles are separated from the screen panels, so a lot of undersized particles pass through the screen apertures successfully (figures 16(b)-(f).Meanwhile, the particle velocity along the screen panel also increases with the increase of the rotational speed, especially along the 5th screen panel (figure 17(c)).Due to sufficient depolymerization, complete separation and quick forward movement of particles, the average residence time of both oversized particles and undersized particles decreases, especially from 160 r min −1 to 200 r min −1 (figure 17(d)), leading to a gradual decrease in the particle layer thickness (figures 16(b)-(f).For the same reason, when n increases from 160 r min −1 to 240 r min −1 , the flow rate of the first two screen panels increases (figure 17(e)), leading to a 109% increase in the total flow rate (figure 17(f)).Meanwhile, the screening efficiency also increases to 100% as the n increases from 160 r min −1 to 240 r min −1 (figure 17(f)).However, when n increases from 240 r min −1 to 360 r min −1 , agglomeration particles have already been sufficiently depolymerized and cannot be further depolymerized, while the particle velocity along the screen panel continues to increases, so a few undersized particles do not have enough residence time to pass through but enter the discharging end directly, resulting in a slight decrease in both the total flow rate and the screening efficiency (figure 17(f)).
To acquire both a large total flow rate and a high screening efficiency, enough but not too high rotational speed n should be chosen.In this study, the optimal rotational speed n is considered to be in the range of 200-320 r min −1 .

Influence of tensional amount Δl
Next, the screening process is carried out at n = 200 r min −1 and n = 320 r min −1 , respectively.First, the effect of the tensional amount is investigated at n = 200 r min −1 .When Δl = −2 mm, the screen panel is always in a relaxed state, whether in the expanding process or the contracting process.The relaxed screen panels cannot provide particles with enough high velocity for depolymerization and separation (figure 19(a)), leading to a thick particle layer (figure 18(a)), a large accumulation mass (figure 19(b)), a small total flow rate (only 3.4 kg min −1 ) and a low screening efficiency (only 47.96%) (figure 19(f)).When Δl increases from −2 mm to 4 mm, the screen panel state at the end of the expanding process changes from relaxed to tensioned, so the maximum particle velocity gradually increases (figure 19(a)), leading to a rapid decrease of the accumulation mass of particles on every screen panel (figure 19(b)).This is because the tensioned screen panels can provide particles with enough high velocity, agglomerated particles are depolymerized and adhesive particles are separated from the screen panels, so a lot of undersized particles pass through the screen apertures successfully (figures 18(b)-(f).Meanwhile, the particle velocity along the screen panel also increases with the increase of Δl (figure 19(c)).Due to sufficient depolymerization, complete separation and quick forward movement of adhesive particles, the average residence time of both oversized particles and undersized particles decreases (figure 19(d)), leading to a gradual decrease in the particle layer thickness (figures 18(b)-(f).For the same reason, when Δl increases from −2 mm to 4 mm, the flow rate of the first two screen panels increases (figure 19(e)), leading to a 105% increase in the total flow rate (figure 19(f)).Meanwhile, the screening efficiency also increases to 98.31% as Δl increases from −2 mm to 4 mm (figure 19(f)).
Next, the effect of the tensional amount is investigated at n = 320 r min −1 .Similar to screening at n = 200 r min −1 , both the maximum particle velocity (figure 21(a)) and the particle velocity along the screen panel (figure 21(c)) increase and both the accumulation mass of particles (figures 21(b) and 20) and the average residence time decrease (figure 21(d)) with the increase of Δl.However, different from screening at n = 200 r min −1 , the screening efficiency and the flow rate do not increase but slightly decrease with the increase of Δl (figures 21(e) and (f).This is because the particle velocity at n = 320 r min −1 is much higher than that at n = 200 r min −1 , no matter what Δl is equal to.Even if Δl = −2 mm, enough high velocity is provided for depolymerization of agglomerated particles and separation of adhesive particles from the screen panels, so a lot of undersized particles pass through the screen apertures successfully and the screening efficiency reaches 100% (figure 21(f)).However, when Δl increases from −2 mm to 4 mm, agglomeration particles have already been sufficiently depolymerized and cannot be further depolymerized, while the particle velocity along the screen panel continues to increase, so a few undersized particles do not have enough residence time to pass through but enter the discharging end directly, resulting in a slight decrease in both the total flow rate and the screening efficiency (figure 21(f)).
To acquire both a large total flow rate and a high screening efficiency, a proper tensional amount Δl should be chosen according to the rotational speed n.

Influence of feed rate M
When M = 20 kg min −1 , adhesive organic fertilizer particles can be thrown up easily (figure 11(a)) and provided enough high velocity (figure 23(a)) for depolymerization and separation, leading to a thin particle layer (figure 22(a)), a small accumulation mass (figure 23(b)) and a high screening efficiency (99.6%) (figure 23(f)).When M increases from 20 kg min −1 to 60 kg min −1 , the accumulation mass of particles on all the screen panels gradually increases (figure 23(b)), leading to a gradual increase of the particle layer thickness (figures 22(b)-(e).Due to the increased gravity of the thick particle layer, the particles become hard to be thrown up, and both the maximum particle velocity (figure 23(a)) and the particle velocity along the screen panels (figure 23(c)) gradually decrease, especially along the 5th Due to the blockage of the thick particle layer and the decreased particle velocity along the screen panels, the average residence time of both oversized particles and undersized particles gradually increase (figure 23(d)).As a result of the screen blockage induced by the thick particle layer (figures 22(d)-(e), undersized particles become hard to pass through the screen apertures, so the screening efficiency decreases from 99.6% to 62.8% as M increases from 20 kg min −1 to 60 kg min −1 (figure 23(f)).Unlike the screening efficiency, the total flow rate first gradually increases when M increases from 20 kg min −1 to 50 kg min −1 and then slightly decreases when M increases from 50 kg min −1 to 60 kg min −1 (figure 23(f)).This is because when M is small, the first few panels screen most of particles and the last few panels screen only a few particles(figure 23(e)); when M increases, the last few panels begin to screen more particles, leading to an increase of the total flow rate(figure 23(e)).However, when M continues to increase, the screen panels can no longer screen more particles but only could screen fewer particles due to the increasingly severe screen blockage, leading a decrease of the total flow rate.
To achieve both a large total flow rate and a high screening efficiency, a proper feed rate M should be chosen.In this study, the optimal feed rate M is considered to be in the range of 30-40 kg min −1 .

Experimental realisation
For verifying the simulation model, a flip-flow screen is installed and it includes a frame, baffles, linkages, two eccentric wheels, an electric motor, a toothed belt and a speed governor, as shown in figure 24.As most of adhesive particles are screened by the first few screen panels, not five but four screen panels are adopted.Similarly, these screen panels have the dimensions of 250 mm × 150 mm × 5 mm, and each screen aperture is 15 mm × 4 mm with a space of 5 mm.Every screen panel has two sides, one side is fixed to the frame and the other side is fixed to the linkage.The eccentric wheels are driven by the electric motor and then they spur the linkages, by which the motion is changed from rotary to swinging.As the swinging angle is very small, the swinging motion is approximately a linear reciprocating motion, leading to the periodically expanding and contracting motion of the screen panels.Finally, all the screen panels perform the same periodic cosine motion, and two adjacent screen panels have a phase difference of 180°.Adhesive organic fertilizer particles with the moisture content of 26% are chosen.The eccentricity is 6 mm, so the slack amount ranges from 0 to 12 mm.The inclination angle, the tensional amount and the feed rate are set to 20°, 4 mm and 20 kg min −1 , respectively.By using the speed governor, the rotational speed can be adjusted in the range of 0-450 r min −1 .
Consistent with the above trend of the screening efficiency with n, the screening efficiency first increases and then decreases with the increase of n, and it approaches 100% when n = 200 r min −1 .
Thus, the rotational speed n is set to 200 r min −1 (namely the period is 0.3s), and the screening process in a random period is observed, as shown in figure 25.In a period, every screen panel expands once and contracts once.In figure 25(a), the distance between the two sides of the 1st and 3rd panels reaches the minimum value and particles have accumulated on the two screen panels, while the distance between the two sides of the 2nd and 4th panels reaches the maximum value.Then, the 1st and 3rd panels expand, while the 2nd and 4th panels contract, so some particles begin to accumulate on the 2nd and 4th panels, as shown in figure 25(b).After half a period, the distance between the two sides of the 1st and 3rd panels reaches the maximum value, while the distance between the two sides of the 2nd and 4th panels reaches the minimum value, as shown in figure 25(c).In the expanding process of the 1st and 3rd panels, particles on the two screen panels are provided with a high velocity, so they will soon be thrown up to the 2nd and 4th screen panels, respectively.
Next, the 1st and 3rd panels contract, while the 2nd and 4th panels expand, as shown in figure 25(d).After a period, the distance between the two sides of the 1st and 3rd panels reaches the minimum value again, while the distance between the two sides of the 2nd and 4th panels reaches the maximum value again, as shown in figure 25(e).Similarly, in the expanding process of the 2nd and 4th panels, particles on the two screen panels are provided with a high velocity, so they are thrown up to the 3nd screen panel and the discharging end, respectively.During the screening process in a period, undersized particles gradually pass through the screen apertures, so the particle layer along the screen panels become thinner and thinner.Finally, oversized particles enter the discharging end.Consistent with the above simulation result, most of adhesive organic fertilizer particles are screened by the first few screen panels and it is observed that undersized particles screened by the 1st and 2nd panels are significantly more than them screened by the 3rd and 4th panels.Overall, adhesive organic fertilizer particles can be successfully screened by using the flip-flow screen.

Conclusions
(1) Adhesive organic fertilizer particles are successfully screened by using the flip-flow screen and organic fertilizer particles in a easily absorbed range are obtained, which is in favor of increasing fertilizer absorption, reducing fertilizer utilization and improving soil environment.
(2) As successfully screening adhesive organic fertilizer particles with high moisture and high viscosity, flipflow screen become an effective solution for screening adhesive particles.Through adjusting the screening variables, flip-flow screen can also be used to screen other adhesive particles, such as moist coal, wet garbage, paper pulp and so on.(3) By using the DEM-MBD coupling model, not only complex motion can be performed and many particles can be generated, but also both the collisions among particles and the impact between the particles and the screen panels can be realized.Thus, the DEM-MBD coupling model can be used to simulate sand excavating, seed planting, material conveying, particle screening and so on.

Figure 1 .
Figure 1.Adhesive organic fertilizer particles (a) produced by rotary drum granulation (b) after classification.

Figure 5 .
Figure 5. Model of five screen panels.

Figure 8 .
Figure 8.Translational motion experiment.(a) A single screen panel; (b) Screen panel during experiment.

Figure 10 .
Figure 10.Theoretical, simulation and experimental trajectory of a single screen panel under different slack amounts.

Figure 11 .
Figure 11.Particle velocity during the screening process.

3. 1 . 3 .
Particle trajectory during the screening process Next, the trajectories of particles with different diameters are analyzed, as shown in figure 13.In it, blue curves and dark green curves are the trajectories of undersized particles, and red curves and light green curves are the trajectories of oversized particles.If possessing proper falling points, undersized particles (Particle 1) will pass

Figure 13 .
Figure 13.Particle trajectory during the screening process.

3. 2 . 1 .
Influence of surface energy γ When γ = 0.3 J m −2 , adhesive organic fertilizer particles can be thrown up easily (figure14(a)) and provided enough high velocity (figure15(a)) for depolymerization and separation, leading to a thin particle layer (figure14(a)), a small accumulation mass (figure 15(b)), a large total flow rate (10.4 kg min −1 ) and a high screening efficiency (99.4%) (figure 15(f)).When γ increases from 0.3 J m −2 to 1.2 J m −2 , both the maximum particle velocity (figure 15(a)) and the particle velocity along the screen panels (figure 15(c)) change slightly.However, both the accumulation mass of particles on all the screen panels (figure 15(b)) and the average residence time of both oversized particles and undersized particles (figure 15(d)) increases, leading to a gradual increase of the particle layer thickness (figures 14(b)-(d).This is because the adhesive force among particles increases with the increase of the surface energy, the provided particle velocity is no longer enough for sufficient depolymerization of agglomerated particles and complete separation of adhesive particles from the screen panels, so undersized particles become hard to pass through the screen apertures.Because of this, when γ increases from 0.3 J m− 2 to 1.2 J m −2 , the flow rate of the first three screen panels decreases (figure 15(e)), leading to a 46% decrease in the total flow rate (figure 15(f)).Meanwhile, the screening efficiency also decreases to 54.2% as γ increases from 0.3 J m− 2 to 1.2 J m− 2 (figure 15(f)).

Figure 25 .
Figure 25.Screening process in a period.

Table 1 .
Diameter and gradation of adhesive organic fertilizer particles.

Table 2 .
Material property of particles and screen panels.

Table 4 .
Selected parameters for simulation.