Journal Pre-proof Determination of the energetic coefficient of restitution of maize grain based on laboratory experiments and DEM simulations
Lijun Wang, Zhaohui Zheng, Yongtao Yu, Tianhua Liu, Zhiheng Zhang PII:
S0032-5910(19)31129-5
DOI:
https://doi.org/10.1016/j.powtec.2019.12.024
Reference:
PTEC 15031
To appear in:
Powder Technology
Received date:
8 July 2019
Revised date:
18 November 2019
Accepted date:
12 December 2019
Please cite this article as: L. Wang, Z. Zheng, Y. Yu, et al., Determination of the energetic coefficient of restitution of maize grain based on laboratory experiments and DEM simulations, Powder Technology(2019), https://doi.org/10.1016/j.powtec.2019.12.024
This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version will undergo additional copyediting, typesetting and review before it is published in its final form, but we are providing this version to give early visibility of the article. Please note that, during the production process, errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
© 2019 Published by Elsevier.
Journal Pre-proof
Determination of the energetic coefficient of restitution of maize grain based on laboratory experiments and DEM simulations Lijun Wang*, Zhaohui Zheng, Yongtao Yu, Tianhua Liu, Zhiheng Zhang (College of Engineering, Northeast Agricultural University, Harbin, Heilongjiang 150030, China)
*Corresponding author: Lijun Wang. Tel.: 86-451-55191897. E-mail address:
oo
f
[email protected].
Abstract:
pr
The coefficient of restitution (COR) is one of the important mechanical properties of particle.
e-
The calculation model of the energetic COR of particle-particle collisions was derived. The
Pr
method used to determine the energetic COR of irregular particles was proposed. Maize particle-particle collisions were simulated using the discrete element method (DEM). The kinetic
rn
al
energy of the particles during the collis ion was investigated. The relationship between the input and calculated energetic COR was obtained via simulation. The energetic COR of maize grain was
Jo u
obtained and verified using laboratory experiments. The results indicate that it is accurate. The results of the simulation indicate that the energetic COR is more accurate than the kinematic COR for maize grain. The effects of the impact velocity and impact angle on the energetic COR were investigated. The results of the paper will be helpful for maize grain simulation and maize-processing machine design.
Keywords: Maize grain, Energetic COR, Particle-particle collision, DEM simulation, Laboratory experiment
1
Journal Pre-proof
1.Introduction The DEM has been increasingly applied to study granular materials. In various industries such as pharmaceutical, food, agricultural production, and chemical processing, the DEM plays an important role [1-4]. To simulate the real process of the particle moving, it is necessary to accurately input their material and mechanical parameters. The COR is one of the important
f
mechanical parameters required to simulate the kinematics and kinetics of particles using DEM. It
oo
characterizes the energy loss during collis ion [5]. The COR has many definitions that are divided
pr
into three categories: kinematic, kinetic, and energetic definitions [6]. To date, the COR has
e-
mainly been obtained using the bulk calibration approach and experimental measurements.
Pr
Some scholars obtained the mechanical parameters of particles using the bulk calibration approach. Hu [7] considered the angle of repose and bulk density of expanded graphite particles as
al
macroscopic responses and the particle density, sliding frictions, COR, and Poisson's ratio as
rn
microscopic variables. The optimization combinations of the microscopic variables were
Jo u
determined. Ghodki [8] obtained the best set of the coefficient of static friction, coefficient of rolling friction, and COR by comparing the angle of repose of soybeans in the simulations and experiments. Ma [9] used the coefficient of rolling friction, COR, JKR surface energy parameter, and coefficient of static friction as microscopic variables and the soil repose angle as the response value. The regression model of the soil repose angle was established and the regression model was optimized using the soil repose angles obtained via physical experiments. The optimal solution of the contact model parameters of clayey black soil particles was obtained. Based on these studies, it is customary to choose the appropriate combination of mechanical parameters that can describe the bulk response of the solids using the bulk calibration approach. Although this method can 2
Journal Pre-proof
obtain the corresponding COR, the physical meaning of the COR might be difficult to understand. The kinematic COR is widely studied because the velocity of the particle before and after collision can be directly obtained using experimental measurements. Some scholars derived various calculation models of the COR based on the velocity to obtain the COR of the particle under different collision conditions. Jafarpur [10] simplified the kinematic COR to the function
f
derived based on the ball height before and after the impact and experimentally determined the
oo
COR of different sizes of balls using a drop test. Crüger [11] derived the calculation function of
pr
the COR based on the relative velocities of colliding partners and experimentally measured the
e-
COR of glass particles with two different diameters at varying relative velocities and liquid layer
Pr
thicknesses. Wang [12] ascertained the calculation function of the COR based on the grain velocities in the direction of the contact force of the particle-particle collision and experimentally
al
determined the COR of frozen maize partic le-particle collision. However, studies shown that the
rn
kinematic COR that is applied to oblique collis ion, with dry friction collision, and irregular
Jo u
particle collision can violate the law of conservation of energy [13-15]. The kinetic and energetic COR were primarily studied theoretically as it is challenging to obtain work done by the contact force and impulse during collision using experimental measurements. Stronge [13,16] found that for planar oblique collis ion with friction, the kinematic and kinetic definitions of the COR yield results that are energetically inconsistent at some initial velocities when sliding stops or changes direction before separation. If the initial sliding does not stop before separation or centric collision occurs between rough bodies, the kinematic, kinetic, and energetic definitions of the COR are equivalent. Djerassi [17] investigated collis ion with friction and found that the energetic COR yielded energetically consistent results. Pandrea [18] 3
Journal Pre-proof
investigated collision without friction of two rigid bodies and proved the equality of the kinematic, kinetic, and energetic definitions of the COR. Ivanov [19] studied collisions with friction of two rigid bodies and found that the kinematic COR applied to an oblique impact can violate the law of conservation of energy. With the kinetic COR, the calculated energy dissipation is always positive and energetic COR is more realistic. Based on these studies, for collisions without friction or
f
centric collis ions with friction, the kinematic, kinetic, and energetic definitions of the COR are
oo
equivalent. For eccentric collision with friction, the energetic COR is superior to the kinematic
pr
and kinetic definitions of the COR.
e-
A method was proposed to determine the energetic COR of irregular particles. The calculation
Pr
model of the energetic COR of particle-particle collision was derived based on Stronge's impact mechanics model [20] and Newton-Euler dynamic equations [21]. The translational and rotational
al
kinetic energy of the particles during collision were investigated. The energetic COR of maize
rn
grain at different impact veloc ities and angles was determined using laboratory experiments and
Jo u
simulations. The applicability of the kinematic and energetic COR for maize grain was compared. The effects of the impact velocity and impact angle on the energetic COR were investigated.
2.Theory The classic definition of the energetic COR was defined by Stronge [6]. The square of the energetic COR eE
2
is the negative of the ratio of the elastic strain energy released during
res titution to the internal energy of deformation abs orbed during compress ion [20]. The corresponding equations as follows:
eE 2
Wr Wz ( p f ) Wz ( pc ) Wc Wz ( pc ) 4
(1)
Journal Pre-proof Wz ( p)
t ( p)
0
p
Fz vcz dt vcz dp
(2)
0
Where Wr is the elastic strain energy released during restitution (J); Wc is the internal energy of deformation absorbed during compression (J); Wz ( p f ) is the work done by the normal contact force during collis ion (J); Wz ( pc ) is the work done by the normal contact force during compression (J); Fz is the normal contact force (N); v cz is the normal relative velocity between
f
the collision bodies at the contact point (m·s -1); p f is the normal component of impulse when the
oo
contact points finally separate (N · s). pc is the normal component of impulse when the
pr
deformation starts recovery (N·s).
e-
The translational and rotational motion of the particles can be described using
Pr
Newton-Euler dynamic equations. Therefore, the calculation model of the energetic COR of particle-particle collision is derived based on Stronge's impact mechanics model and
al
Newton-Euler dynamic equations.
rn
The particle-particle collis ion is shown in Fig.1. In global coordinate system XYZO , two
Jo u
particles collide at contact point. The contact point is defined as CG located at particle 1 and located at particle 2 for convenient analysis. The particles have centroids located at respectively. There is a position vector rG from GG to and CG ' , equal but opposite interaction forces FG and
CG ,
FG ' act
GG
and rG ' from GG ' to
CG '
and GG ' ,
CG ' .
At
CG
on the particles, respectively. The
particles have translational and angular velocities of VG , VG ' , G , and G ' , respectively. Information on particle-particle collis ion in the global coordinate system XYZO such the coordinates of the centroid and contact point, contact force including the normal contact force and tangential contact force, translational velocity, and angular velocity are obtained using the export data function available in EDEM2.6 software (DEM solution Ltd.,Edinburgh,UK). It is necessary 5
Journal Pre-proof to establish a local coordinate system xyzo whose origin is at CG for convenient analysis. The
z axis is parallel to the normal contact force
FGn . The collis ion contact plane is established. The
z axis is perpendicular to the collision contact plane and the origin of the local coordinate system is at the collis ion contact plane. The collision contact plane equation is solved based on the point normal form equation of a plane. A random line that passes through the origin of the local
f
coordinate system and is located at the collision contact plane is defined as the x axis. The y
oo
axis is obtained based on the mathematical idea of the vector product as the x , y , and z axes
pr
are perpendicular to each other. Information on the particle-particle collision is transferred to the
XYZO by coordinate
e-
local coordinate system xyzo from the global coordinate system
transformation. The calculation model of the energetic COR is derived based on the information
Pr
on the particle-particle collision in the local coordinate system xyzo .
al
The translational and rotational motion of the particle can be described using
Jo u
rn
Newton-Euler dynamic equations. The corresponding equations can be written as: dV M V F dt
(3)
d I r F dt
(4)
M
I
Where M is the mass of particle (kg); V is the translational velocity of particle (m·s -1 ); the angular velocity of particle (rad·s -1);
F
is
is the interaction force ( N ); I is the moment of
intertia of particle (kg·m2 ); r is a position vector from centroid to the contact point (m). From Eqs. (3) and (4), we can obtain Eqs.(5) and (6), respectively.
dV M -1dP Vdt
d I 1r Fdt I 1 Idt
(5) (6)
Where dP is the impulse (N·s). From Eqs. (5) and (6), the change values of the translational and angular velocities of particle 1 6
Journal Pre-proof
and particle 2 in a unit of time can be obtained, respectively.
dVi M -1dPi i Vi dt
(7)
dVi ' M '-1 dPi 'i 'Vi ' dt
(8)
1
1
di I ij rj Fk dt I ij i I iji dt
(9)
di ' I ij '1 rj 'Fk ' dt I ij '1 i 'I ij 'i ' dt
(10)
(11)
oo
v V r
f
The velocity of the contact point of the particle can be calculated based on Eq. (11).
pr
From Eq.(11), we can obtain the velocities of the contact point of the particle 1 and particle 2,
e-
respectively.
(12)
vi ' Vi ' j 'rk '
(13)
Pr
vi Vi j rk
al
The relative velocity between the particles at the contact point can be calculated based on Eq. (14).
vci vi - vi '
rn
(14)
Jo u
The change value of the relative velocity between the particles at the contact point in a unit of time can be calculated based on Eq. (15).
dv ci dvi - dvi '
(15)
Combining Eqs (7), (8), (9), (10), (12), (13), (14), and (15), dvci is expressed as:
dvci mij dpi Bi dt Where
1
mij (M -1 M '-1 ) ij ikm jln (I kl rm rn I kl '1 rm ' rn ')
7
(16) (17)
Journal Pre-proof Bi ijk( - jVk j 'Vk ')
1
1
1
ijk - rk I ji Ai I jj A j I jk Ak rk ' I ji ' Ai ' I jj ' A j ' I jk ' Ak ' 1
1
1
(18)
Based on the index of B , the indices of A and A' can be determined. A and A' can be calculated based on Eqs. (19) and (20), respectively. Ai ijk j ( I kii I kj j I kk k )
(19)
Ai ' ijk j ' ( I ki ' i ' I kj ' j ' I kk ' k ' )
(20)
if i j .
ijk , ikm and jln
the indices are in cyclic order,
if i j
are the permutation tensor,
ijk ikm jln -1
ijk ikm jln 1
if
if the indices are in anticyclic order, and
if the indices are in other orders.
e-
ijk ikm jln 0
ij 1
f
ij 0
is the Kronecker delta defined as
pr
and
ij
oo
Where i j k m l n x, y, z .
Pr
When the relative velocity between the particles at the contact point is generated in the collision contact plane, sliding motion occurs. The sliding velocity is the relative velocity between
al
the particles at the contact point in the collis ion contact plane. v cx and v cy are the sliding
. The friction force hindering the sliding motion of the particles
Jo u
denoted as
rn
velocities in the x and y directions. The angle between v cx and the sliding velocity is will be generated.
The friction force is represented by Coulomb’s law and its direction is opposite the direction of the n sliding motion. The normal contact force FG
is defined as Fz in the local coordinate system
xyzo for convenient analys is. The x and y components of the friction force are defined as
Fsx and Fsy , respectively. Fsx and Fsy can be calculated based on Eqs. (21) and (22), respectively.
Fsx
v cx v cx v cy 2
2
f s Fz
8
Fsx - f s cos Fz
(21)
Journal Pre-proof vcy
Fsy
vcx vcy 2
Fsy - f s sin Fz
f s Fz
2
(22)
Where f s is the coefficient of static friction . From Eqs. (21) and (22), the x and y components of the impulse can be expressed as:
dpx - f s cos dpz
(23)
dp y - f s sin dp z
(24)
mxz dpx Bx m yz dp y By dt mzz dpz Bz
oo
mxy m yy mzy
(25)
pr
dv cx mxx dv m cy yx dv cz mzx
f
Eq. (16) can be written as:
Combining Eqs. (23), (24), and (25), dv cz can be written as:
e-
dvcz ( f s cos mzx f s sin mzy mzz )dpz Bz dt
(26)
dvcz Cdpz Bz dt
(27)
al
Pr
f s cos mzx f s sin mzy mzz is defined as C , Eq.(26) can be simplified as:
rn
v cz nonlinearly changes as pz changes because the angular velocity, translational velocity,
Jo u
position vector, and contact force constantly change during the collision. The calculation of the work done by the normal contact force is simplified based on the mathematical idea of the integral. The collision period is divided into n stages. The stages are defined as m ( 1 m n ). The start and end times of the m stage are defined as
m 1 and m , respectively. We assume that the
information on the particle-particle collision in the m stage is the same as that of the time m . Therefore, v cz linearly changes as
pz changes in the m stage.
v cz can be obtained as a function of pz : v cz Cm ( pz pzm1 ) vczm-1 Bzmt Where t is the interval between
m 1 and m . 9
(28)
Journal Pre-proof The work Wm done by the normal contact force in the m stage can be calculated based on Eq. (29).
Wm
p zm
p zm1
(29)
v cz ( pz )dpz
Combining Eqs. (28) and (29), Wm can be written as:
0.5v czm 0.5v czm1 0.5Bzm t 2 Bzm v czm1t Cm 2
Wm
2
(30)
eE can be calculated based on Eq.(31). n
eE
Wr Wc
oo
f
The energetic COR
2
W
m c 1 c
m
(31)
Wm
pr
m 1
Pr
e-
Where c is the time that the deformation of the particle starts recovery.
rn
3.1. Contact model
al
3. Numerical simulation
The maize particle-particle collis ion was simulated using a three-dimensional DEM based on
Jo u
the soft-sphere model. In the DEM simulation, the translational and rotational motion of the particle was described by Newton-Euler dynamic equations. The cohesive force and liquid bridge between the particles were ignored as the moisture content of the maize grain was 18.6%. The literatures shown that the Hertz-Mindlin contact model can be applied to model the collision of the particle [22-24]. Thus, the no-slip Hertz-Mindlin contact model was chosen to calculate the contact force and torque of each maize partic le. The normal and tangential contact forces were calculated based on the Hertz’s theory and the Mindlin and Deresiewic z theory, respectively [25-26]. The EDEM2.6 was chosen to simulate the maize particle- particle collision. 5% of the Rayleigh time step was used as the fixed time step that is 9.19×10 -7 s to ensure the 10
Journal Pre-proof
accuracy of the simulation results. The cell size was 2 Rmin and the Rmin was defined as the radius of the smallest sphere that is 1.9 mm. To increase the simulation efficiency, the data sampling interval was 1×10-6 s for the period between the time that the particles moved out of oblique plate and the time that the particles fell after the collis ion and that of the remaining
f
simulation time was set as 1×10-4 s.
oo
3.2. DEM model of particle and geometry
pr
In the DEM simulation, multi-sphere approach was often used to construct the model of
e-
irregular particle [27-29]. The number of overlapping sphere should be adopted as much as
Pr
possible to approximate the actual shape of an irregular particle. However, the excessive amount of the overlapping sphere would cause excessive computational expense and can't significantly
al
improve the accuracy of the simulation results [30]. Therefore, the reasonable number of the
rn
overlapping sphere should be adopted. The wedge maize grain was the object used because its
Jo u
quantity accounts for more than 90% of the total maize grain [31]. Thirty wedge maize grains were randomly selected to measure their dimension parameters including the heights, widths of the lower and upper shoulders, and thicknesses of the lower and upper shoulders (see Fig.2). The averages of the dimension parameters were calculated. A geometrical model of the wedge maize grain was attained using SolidWorks2016 (SP02, Dassault Systèmes Americas Corp., United States of America) and imported into EDEM2.6. The model was composed of 42 overlapping spheres to fit its real shape as shown in Fig. 3. The platform for testing the kinematic COR of the particle-particle collision is shown in Fig. 4(a), and the simple geometry of the DEM simulation is presented in Fig. 4(b). The simple 11
Journal Pre-proof
geometry includes two oblique plates, two rotation shafts, and a trestle. a is the distance from the generation position of the particle to the lower end of the oblique plate. b is the width of the trestle. c is the 72 mm distance from the lower end of the oblique plate to the rotation shaft. d is the 12 mm thickness of the oblique plate. β is the angle between the oblique plate and the horizontal surface. The impact velocity of the particle can be adjusted by changing the values of a and b. The
f
impact angle of the particle can be adjusted by changing the value of β. The values of a, b, c, d,
oo
and β are the same as the laboratory experiment settings to ensure the accuracy of the simulation
e-
pr
results.
Pr
3.3. Material properties
The influence of the irregular shape in the simulation was considered.
Studies have shown
al
that the contact parameters such as the coefficient of rolling friction and the COR are affected by
rn
the shape of the particle [32,33]. The kinematic COR of the wedge maize grain at the different
Jo u
impact velocities and angles was experimentally measured and the process is the same as our previous research [12]. When the impact angle varies from 40° to 60° and the impact velocity changes from 0.9 m·s -1 to 1.7 m·s -1, the kinematic COR of the wedge maize grain is shown in Table 1. The coefficient of rolling friction of the wedge maize grain was referenced that we previously determined [34]. The coefficients of static friction between the maize grain and the oblique plate and maize grain were obtained using the inclined plane method, respectively. For the coefficient of static friction between the maize grain and the oblique plate, a maize grain is placed on a galvanized steel plate. For the coefficient of static friction between the maize grains, some maize grains are glued on the plate, and then a maize grain is placed on the maize plate. The 12
Journal Pre-proof
inclination angle of plate is increased slowly. When the maize grain starts to slip, the inclination angle is measured. The tangent value of the inclination angle is the coefficient of static friction [35]. The wedge maize grain density was calculated based on its mass and its volume. The other physical parameters of the wedge maize grain and oblique plate were referenced from the literature [34]. The physical parameters of the wedge maize grain and galvanized steel plate are
f
reported in Table 2. The physical parameters of the wedge maize grain and galvanized steel plate
pr
oo
were input into EDEM2.6 software to simulate the collision between the wedge maize grains.
Pr
4.1. Behaviors of particle-particle collision
e-
4. Result & analysis
The microscopic properties and macroscopic behaviors of the particle-particle collision were
al
investigated based on the DEM simulation results to better understand the mechanism of the
rn
particle-particle collision. The behaviors of the particle-particle collision are described by setting
as an example.
Jo u
the collision between the particles at an impact angle of 60° and their impact velocity at 0.9 m·s -1
4.1.1. Microscopic properties of particle-particle collision The microscopic properties of the particle-particle collis ion such as the interaction force and kinetic energy of the particle during the collision were investigated. The collision between the particles occurred under the effect of gravity. When two particles collide, they exert the same forces on each other that are in the opposite direction [36]. So the interaction force acting on particle 1 and the kinetic energy of particle 1 were investigated. The forces acting on particle 1 are 13
Journal Pre-proof
plotted in Fig.5. The magnitude of the gravity is 0.00303129 N. As the collis ion time increases, the magnitudes of the normal and tangential contact forces and friction force initially increase and then decrease. The maxima of the normal and tangential contact forces and friction force are 0.847827 N, 0.224173 N, and 0.3153916 N, respectively. The gravity is much smaller than the other forces. Thus, the gravity of the particle was ignored [36,37].
f
The interaction force acting on the contact point affects not only particle’s translational motion
oo
but also rotational motion [38]. Thus, the interaction force was transformed into a component
pr
force through the centroid of the particle and a rotational torque acting on the contact point of the
e-
particles as shown Fig.8(a) and Fig.9(a). The deformations and sliding motion of the particles
Pr
occur due to the interaction between the particles and change the interaction force. The interaction force includes the contact force and friction force. The contact force is affected by the particle
al
deformation [39]. The deformation process is reflected by the normal relative velocity between the
rn
particles at the contact point. The deformation period was divided into a compression period from
Jo u
0 to 0.000049 s wherein vcz 0 and the restitution period from 0.000049 to 0.00142 s wherein
vcz 0 based on the normal relative veloc ity (see Fig.6). The friction force is affected by the sliding motion [40]. The direction and speed of the sliding motion are reflected by the sliding velocity. The state of the sliding motion was divided based on the variations in the sliding velocity and the angle
between it and its component in the x direction (see Fig.7). The state of the
sliding motion is divided into the deceleration slip in a positive direction whose duration is 0 to 0.000035 s wherein
90 and the sliding velocity decreases, the acceleration slip in a reverse
direction whose duration is 0.000035 to 0.000085 s wherein
90 and the sliding velocity
increases, and the deceleration slip in a reverse direction whose duration is 0.000085 to 0.000142 s 14
Journal Pre-proof
wherein
90 and the sliding velocity decreases. The analyses of the work done by the
interaction force and energy transformation during the collis ion were divided into four stages based on the deformation periods and the states of the sliding motion. In stage Ⅰ (0~0. 000035 s), the particles are in a compression period and the state of the sliding motion is deceleration slip in a positive direction. The particle deforms due to the
f
compression and the translational kinetic energy transforms into internal energy of deformation of
oo
the collis ion partner due to the work done by FR (see Fig.8(a)). The collision between the T
( see
pr
particles is eccentric and the particles have no angular velocity. The rotational torque
e-
Fig.9(a)) acting on the contact point is generated. The translational kinetic energy transforms into T
. Thus, the
Pr
the rotational kinetic energy of the particle due to the work done by rotational torque
rotational kinetic energy of the particle increases from 0 J to 2.35×10-6 J and its increase rate is
al
6.714×10-2 J ·s -1 (see Fig.9(b)). The translational kinetic energy of the particle decreases to
rn
1.147×10-4 J from 1.260×10-4 J and its decrease rate is 3.138×10-1 J ·s -1 (see Fig.8(b)).
Jo u
In stage Ⅱ (0.000035~0.000049 s), the particles are in a compression period and the state of the sliding motion is acceleration slip in a reverse direction. The direction of the FRf changes (see Fig.8(a)) as the direction of the sliding motion reverses. The transformation rate between the translational kinetic energy and internal energy of deformation of the collision partner reduces. The work done by FR deforms the particle.
T
increases with the particle deformation. The
transformation rate between the translational kinetic energy and rotational kinetic energy increases. The rotational kinetic energy increases from 2.35×10-6 J to 4.55×10-6 J due to the work done by
T
and its increase rate is 233% that of stage Ⅰ (see Fig.9(b)). The translational kinetic energy decreases to 1.114×10-4 J from 1.147×10-4 J and its decrease rate is only 75.11% that of stage Ⅰ 15
Journal Pre-proof
(see Fig.8(b)). In stage Ⅲ(0.000049~0.000085 s), the particles are in a restitution period and the state of the sliding motion is acceleration slip in a reverse direction. The particle deformation starts to recover. The elastic part of the internal energy of deformation of the collision partner transforms into the translational kinetic energy during the restitution.
T
decreases with the recovery of the particle
f
deformation (see Fig.9(a)). The transformation rate between the translational kinetic energy and
T and
its increase rate is only 75.3% that of stage Ⅱ (see
pr
8.81×10-6 J due to the work done by
oo
rotational kinetic energy decreases. The rotational kinetic energy increases from 4.55×10-6 J to
e-
Fig.9(b)). The translational kinetic energy decreases to 1.08×10-4 J from 1.114×10-4 J and its
Pr
decrease rate is only 40.05% that of stage Ⅱ(see Fig.8(b)).
In stage Ⅳ (0.000085~0.00142 s), the particles are in a restitution period and the state of the
al
sliding motion is deceleration slip in a reverse direction. The energy transformation between the
rn
elastic strain energy and translational kinetic energy continues with the deformation recover. The
Jo u
rotational kinetic energy transforms into the translational kinetic energy when the particles have a large angular velocity [5]. The rotational kinetic energy decreases to 7.68×10-6 J from 8.81×10-6 J (see Fig.9(b)) and the translational kinetic energy increases from 1.08×10-4 J to 1.09×10-4 J (see Fig.8(b)). The mechanical energy transforms into heat due to the frictional slip. The increment of the translational kinetic energy is less than the decrement of the rotational kinetic energy.
4.1.2. Macroscopic behaviors of particle-particle collision Studies have shown that the macroscopic behaviors of particle collision are affected by the microscopic properties during the collisions [5,37]. The macroscopic behaviors of the 16
Journal Pre-proof
particle-particle collis ion such as the translational and angular velocities of the particle were investigated. Fig.10(a)and (b) show the translational and angular veloc ities of the particles before and after the collision, respectively. The particles have only X and Z components of the translational veloc ity before the collision and have X, Y, and Z components of the translational velocity after the collis ion. The X component of the translational velocity decreases and the Z
f
component of the translational velocity slightly decreases. Overall, the translational velocities of
oo
the particles decrease after the collision. The particles have a small angular veloc ity before the
pr
collision. After the collision, the X component of the angular velocity increases slightly, the Y
e-
component of the angular velocity increases most obviously, and the Z component of the angular
Pr
velocity increases significantly. Overall, the angular velocities of the particles increase after the collision. The variation tendenc ies of the translational and angular velocities of the particles are
al
the same as that of the translational and rotational kinetic energies of the particles, respectively.
rn
The aforementioned analys is shows that the translational and angular velocities of the particle
Jo u
change after the collis ion as the kinetic energy is affected by the work done by the interaction force. This also proves that compared with the kinematic COR and kinetic COR, the energetic COR is better able to reflect the nature of the collision. The translational and angular velocities of the particle change after the collision. Therefore, the energetic COR of the maize grain obtained was verified by comparing the translational and angular velocities of the particle after the collision in the simulation and experiment.
4.2. Determination of the energetic COR The energetic COR of irregular particle was determined based on laboratory experiments and 17
Journal Pre-proof
DEM simulations. The method of determination is as follows. The single-factor simulation was used to select the contact parameter of the particle [41,42]. The range of the energetic COR is important. We measured the kinematic COR of the maize grain (see Table 1) and referenced the kinematic COR of the maize grain determined by other scholars [33,43]. The energetic COR initially ranged from 0.75 to 1.25 times the kinematic COR obtained. The golden-section method
f
has the advantages of fine stability, fast convergence speed, and high precision compared with
oo
other single dimensional search method [44]. The initial range of the energetic COR was shrank to
pr
a relatively small range through the golden-section method and comparison of the input and
e-
calculated energetic COR in the silulation. The number of simulation experiments could be
Pr
decreased. Within the minimum range of the energetic COR, the relationship between the input and calculated energetic COR can be obtained through single-factor simulation. The more accurate
al
the parameters input, the more realistic the collis ion behaviors of the particle simulated by DEM
rn
[45]. When the energetic COR calculated is the same as the energetic COR input, the energetic
Jo u
COR can be obtained based on this relationship obtained. The energetic COR obtained was verified by comparing the translational velocities and the Y components of the angular velocities of the particles after the collision in the simulation and experiment.
4.2. 1. Selection of the energetic COR The 0.75 and 1.25 times the kinematic COR at the specific impact velocity and angle are denoted as A and B , respectively. The initial range of the corresponding energetic COR is denoted as A, B. In this range, two golden section points are selected, namely, points a and
b , as shown in Eqs. (32) and (33). 18
Journal Pre-proof
Where 0.618 . a and
a B ( B A)
(32)
b A ( B A)
(33)
b were input into the software to simulate partic le-particle collision,
respectively. The corresponding energetic CORs that are calculated based on the simulation results are F a and F b , respectively. if
F (a) a F (b) b
oo
f
Let a A . Then, the range is [a, B] .if
(34)
A, b .
e-
Let b B . Then, the range is
(35)
pr
F (a) a F (b) b
Pr
New points are selected in the new range to continue the calculation and comparison. After several loops, the initial range eventually shrinks to a very small range. Within this range, the
al
relationship between the input and calculated energetic COR can be obtained through single-factor
rn
simulation. If the energetic COR calculated is the same as the energetic COR input, the energetic
Jo u
COR can be obtained based on this relationship obtained. To better understand the process of the selection of the energetic COR, the collis ion between the particles when their impact angle is 60° and impact velocity is 0.9 m·s -1 is used. The experimentally determined kinematic COR of the maize grain is 0.438. The initial range of the energetic COR is 0.3285-0.5475. Three iterations are used in the given initial range. The results obtained during the iteration process are shown in Table 3. The energetic COR of the maize grain ranges from 0.443 to 0.495. Within this range, the COR einput input into EDEM2.6 software is a factor, and the energetic COR ecalculated calculated based on the simulation result is a target. A single-factor simulation is conducted, The result is shown in Fig. 11. 19
Journal Pre-proof The data points in Fig.11 are fitted to determine the relationship between einput and the
ecalculated as shown in Eq.(36). ecalculated 11.41einput 12einput 3.598 2
(R2 =0.9805)
(36)
When ecalculated einput , an energetic COR of 0.474 is obtained. Based on the aforementioned method, the energetic COR of the maize grain at the different
oo
f
impact velocities and angles was obtained as shown in Table 4.
pr
4.2.2. Physical verification experiment
e-
The energetic COR of the maize grain obtained was verified by comparing the translational
Pr
velocities and Y components of the angular velocities of the particles after the collision in the simulation and experiment. The particle-particle collisions at the different impact velocities and
al
angles were simulated based on the energetic COR shown in Table 4. The translational veloc ities
rn
and Y components of the angular velocities of the simulation partic les are obtained using the
Jo u
export data function available in EDEM2.6. Based on the platform for testing the kinematic COR of the particle-particle collision shown in Fig.4(a), the actual maize partic le-particle collis ion experiments were conducted under the same conditions used to simulate the particle-particle collis ion. The particle images were obtained using a high-speed digital video camera (Phantom v9.1, Vis ion Research, America) and processed via Phantom motion analysis software (8.0.606.0-CPhCon:606). At time t 0 , the coordinates of maize grains 1 and 2 were ( X10 , Y10 , Z10 ) and ( X 20 , Y20 , Z 20 ) , respectively. At time t n , the coordinates of maize grains 1 and 2 changed to ( X1n ,Y1n , Z1n ) and ( X 2n , Y2n , Z 2n ) , respectively. 20
Journal Pre-proof
At time tn / 2 , the translational velocity of the particle is the same as the average translational velocity of the particle in the period from times t 0 to t n . The translational velocities of maize grains 1 and 2 at time tn / 2 can be calculated as shown in Eqs. (37) and (38), respectively.
X 1n X 10 t n - t0
V1Y
Y1n Y10 t n - t0
V2 X
X 2 n X 20 t n - t0
V2Y
V1Z
Y2 n Y20 tn - t0
Z1n Z10 t n - t0
V2 Z
Z 2 n Z 20 t n - t0
(37)
(38)
f
V1 X
oo
Due to restrictions in the measurement device and instrument, only the Y component of the
pr
angular velocity was measured. The positions of the maize grains are represented by the dotted lines. The angles between the dotted lines of maize grains 1 and 2 and the horizontal axis were
1n and 2 n respectively (see Fig.12 (b)) at time t n .
Pr
as
10 and 2 0 , respectively (see Fig.12 (a)) at time t0 , while the angles were defined
e-
defined as
al
At the time tn / 2 , the angular veloc ity of the particle is the same as the average angular
rn
velocity of the particle in the period from times t 0 to t n . The Y components of the angular
respectively.
Jo u
velocities of maize grains 1 and 2 at time tn / 2 can be calculated as shown in Eqs. (39) and (40),
1Y
2Y
1n 10 t n - t0
2 n 20 t n - t0
(39)
(40)
The simulation and experiment were repeated 5 times. The translational velocity and Y component of the angular velocity were obtained by averaging all of the data from the experiments and simulations. The standard deviations of the translational veloc ity and Y component of the angular veloc ity were calculated to understand the dispersion of the data. The 21
Journal Pre-proof
average values and standard deviations of the translational velocity and Y component of the angular velocity are shown in Tables 5, 6, 7, and 8, respectively. Fig.13 and 14 are based on Tables 5, 6, 7, and 8. The simulation result is close to that of the experiment. Hence, the determined energetic COR is accurate and the method used to determine the energetic COR of the maize grain is valid. However, there is still a difference between the
f
simulation and experiment. On the one hand, the collis ion positions of the maize grains are
oo
different in the experiment, while they are set exactly in the simulation. On the other hand,
pr
compared with the experiment, the simulation environment is ideal and the maize grain energy
Pr
e-
loss is smaller.
4.2.3. Comparison between the kinematic and energetic COR
al
The validity of the method used to determine the energetic COR of the maize grain was
rn
demonstrated by comparing the translational velocities and Y components of the angular velocities
Jo u
of the particles after the collision in the simulation and experiment. For the maize grain-grain collision, the applicability of the kinematic and energetic COR was compared based on the simulation.
The particle-particle collisions at the different impact angles and impact velocities were simulated based on the kinematic COR shown in Table 1 and the energetic COR shown in Table 4, respectively. The corresponding energetic COR was calculated based on the simulation results. The input and calculated CORs were compared. The simulation experiments of the particle-particle collis ions at the different impact angles and velocities were repeated 5 times. The average values and standard deviations of the energetic COR 22
Journal Pre-proof
obtained were obtained as shown in Table 9. For convenient analysis, the kinematic and energetic COR that were input to simulate particle-particle collis ions are defined as eKinput and eEinput , respectively. The energetic COR eEcalculated
eKcalculated
was calculated based on eKinput and the energetic COR
was calculated based on eEinput .
As shown in Fig.15, the difference between eKinput and
eKcalculated
is large. The difference reaches
f
a maximum of 0.11798 when the impact angle is 40° and the impact velocity is 0.9 m·s -1. The
1.3 m·s -1 . The difference between eEinput and
is small. The difference reaches a maximum
pr
eEcalculated
oo
difference reaches a minimum of 0.0565 when the impact angle is 60° and the impact velocity is
e-
of 0.005 when the impact angle is 60° and the impact velocity is 0.9 m·s -1. The difference reaches
Pr
a minimum of 0.0004 when the impact angle is 60° and the impact velocity is 1.5 m·s -1 . The energetic COR is more accurate than the kinematic COR for maize grain. However, there is still a eEcalculated
because the calculation of the work done by the normal
al
difference between eEinput and
Jo u
rn
contact force during collision was simplified.
4.3. Effects of different factors on the energetic COR The determined energetic COR was verified by the laboratory experiment, which indicates that the determined energetic COR is accurate. The maize particle-particle collisions at the different impact velocities and angles were simulated based on the determined energetic COR of the maize grain shown in Table 4. The effects of the impact angles and velocities on the energetic COR were investigated based on the simulation results. Fig.16 is based on Table 4. Fig.16(a)and (b) show the effects of the impact angle and velocity on the energetic COR of the maize grain, respectively. When the impact velocity is fixed at 0.9 23
Journal Pre-proof m·s -1 , the energetic COR increases as the impact angle increases. The contact force decreases as the impact angle increases as shown in Fig.17(a). The energy loss decreases and the energetic COR increases because the particle deformation decreases as the contact force decreases. When the impact angle is fixed at 60°, the energetic COR decreases as the impact velocity increases. The contact force increases as the impact velocity increases as shown in Fig.17(b). The energy loss
f
increases and the energetic COR decreases because the particle deformation increases as the
pr
oo
contact force increases.
e-
5. Discussion
Pr
The calculation of the work done by the normal contact force during the collision was simplified based on the mathematical idea of the integral as the information on the particle-particle
al
collision constantly changes. The number of section points that divide the collision period should
rn
be adopted as much as possible to approximate the real value of the work. However, an excessive
Jo u
number of section points causes unreasonable computational expense. The interval between the two section points was 1×10-6 s. The energetic COR determined was verified by comparing the results in the simulation and experiment. The results of the simulation and experiment are close. Only the energetic COR of the wedge maize grain was investigated as its quantity accounts for more than 90% of the total maize grain [31]. Only the effects of the impact velocities and angles on the energetic COR were investigated. The effect of the collision position of the maize grain will be investigated in the future. The energetic COR was compared with the kinematic COR that we previously measured [12]. The energetic COR and kinematic COR have similar trends with the variations in the impact 24
Journal Pre-proof
velocities and impact angles as the kinematic and energetic definitions of the COR can reflect the energy loss during collisions. The value of the kinematic COR compared with the energetic COR is smaller as the kinematic COR does not separate the energy dissipation due to deformation from that due to friction. The result that the energetic COR is more accurate than the kinematic COR for
f
maize grain is consistent with the findings of Stronge [13].
oo
6.Conclusions
pr
The calculation model of the energetic COR of particle-particle collis ion was derived based on
e-
Stronge's impact mechanics model and Newton-Euler dynamic equations. -1
Pr
The collision period of the maize particle-particle when their impact velocity was 0.9 m·s
and their impact angle was 60° was divided into four stages: the deceleration slip in a positive
al
direction and acceleration slip in a reverse direction during the compression period and the
rn
acceleration slip and deceleration slip in a reverse direction during the restitution period.
Jo u
The translational and rotational kinetic energies of the particle are affected by the work done by the interaction force. Thus, the translational and angular velocities of the particle change. The energetic COR derived based on the work done by the normal contact force reflects the nature of the collision. When the impact velocity and impact angle are 0.9 m ·s -1 and 60°, respectively, the energetic COR of the maize grain was 0.474. When the impact velocity is fixed at 0.9 m·s -1 and the impact angle increases from 40° to 60°, the energetic COR of the maize grain increases from 0.334 to -1
0.474. When the impact angle is fixed at 60° and the impact velocity increases from 0.9 m·s to 1.7 m·s -1 , the energetic COR of the maize grain decreases from 0.474 to 0.304. 25
Journal Pre-proof
The simulation result is consistent with the result of the experiment, which demonstrates the validity of the method used to determine the energetic COR of irregular particles based on laboratory experiments and simulations. The energetic COR is more accurate than the kinematic COR for maize grain.
f
Acknowledgments
oo
This work was supported financially by the Chinese Natural Science Foundation (51475090),
pr
the Natural Science Foundation of Heilongjiang Province, China (E2017004), New Century
e-
Excellent Talents of General Universities of Heilongjiang Province, China (1254-NCET-003) and
Pr
the Science Backbone Project of the Northeast Agricultural University, China.
al
References
rn
[1] A. Tamrakar, R. Ramachandran, CFD-DEM-PBM coupled model development and validation
249-270.
Jo u
of a 3D top-spray fluidized bed wet granulation process, Comput. Chem. Eng. 125 (2019) https://doi.org/ 10.1016/j.compchemeng.2019.01.023.
[2] J.E. Kim, Y. M. Chung, CFD-DEM simulation of the fluidized-bed granulation of food powders, Biotechnol. Bioproc. Eng. 24 (2019) 191-205. https://doi.org/ 10.1007/s12257-018-0382-6. [3] J. Mellmann, F. Weigler, H. Scaar, Research on procedural optimization and development of agricultural drying processes, Drying Technol. 37 (2019) 569-578. https://doi.org/ 10.1080/07373937.2018.1494186. [4] B. Blaisa, D.Vidal, F. Bertrand, G. S. Patience, J. Chaouki, Experimental Methods in Chemical 26
Journal Pre-proof
Engineering: Discrete Element Method-DEM, Can. J. Chem. Eng. 97 (2019) 1964-1973. https://doi.org/ 10.1002/cjce.23501. [5] B. Buck, Y.L. Tang, S. Heinrich, N.G. Deen, J.A.M. Kuipers, Collision dynamics of wet solids: Rebound and rotation, Powder Technol. 316 (2017) 218-224. https://doi.org/ 10.1016/j.powtec.2016.12.088.
https://doi.org/ 10.13465/j.cnki.jvs.2015.19.007.
oo
Vib. Shock. 34 (2015) 43-48.
f
[6] W.L. Yao,R. Yue, Advance in controversial restitution coefficient study for impact problems, J.
pr
[7] H.L. Zhou, Z.Q. Hu, J.G. Chen, X. Lv, N. Xie, Calibration of DEM models for irregular
https://doi.org/ 10.1016/j.powtec.2018.03.064.
Pr
(2018) 210-223.
e-
particles based on experimental design method and bulk experiments, Powder Technol. 332
[8] B.M. Ghodki, M. Patel, R. Namdeo, G. Carpenter, Calibration of discrete element model
al
parameters: soybeans, Comput. Part. Mech. 6 (2019) 3-10.
rn
https://doi.org/10.1007/s40571-018-0194-7.
Jo u
[9] J.W. Li, J. Tong, B. Hu, H. B. Wang, C.Y. Mao, Y. H. Ma, Calibration of parameters of interaction between clayey black soil with different moisture content and soil-engaging component in northeast China, Transactions
of
the
CSAE. 35 (2019) 130-140.
https://doi.org/10.11975/j.issn.1002-6819.2019.06.016. [10] A. Aryaei, K. Hashemnia, K. Jafarpur, Experimental and numerical study of ball size effect on restitution coefficient in low velocity impacts, Int. J. Impact. Eng. 37 (2010) 1037-1044. https://doi.org/10.1016/j.ijimpeng.2010.04.005. [11] B. Crüger, V. Salikov, S. Heinrich, S. Antonyuk, V.S. Sutkar, N.G. Deen, J.A.M. Kuipers, Coefficient of restitution for particles impacting on wet surfaces: An improved experimental 27
Journal Pre-proof
approach, Particuology. 25 (2016) 1-9. https://doi.org/10.1016/10.1016/j.partic.2015.04.002. [12] L.J. Wang, B.X. Wu, Z.C. Wu, R. Li, X. Feng, Experimental determination of the coefficient of restitution of particle-particle collis ion for frozen maize grains, Powder Technol. 338 (2018) 263-273.
https://doi.org/ 10.1016/j.powtec.2018.07.005.
pr
https://doi.org/ 10.1016/10.1115/1.4030459.
oo
with friction, J. Appl. Mech. 82 (2015) 081003.
f
[13] W.J. Stronge, Energetically consistent calculations for oblique impact in unbalanced systems
e-
[14] Y. Wang, M.T. Mason, Two-dimensional rigid-body collisions with friction, J. Appl. Mech. 59
Pr
(1992) 635- 642.https://doi.org/ 10.1115/1.2893771. [15] D.B. Marghitu, E.D. Stoenescu, Rigid body impact with moment of rolling friction, Nonlinear. https://doi.org/ 10.1007/s11071-006-9176-z.
al
Dyn. 50 (2007) 597-608.
rn
[16] W.J. Stronge, Comment: collision with friction; part B: Poisson’s and Stronge’s hypotheses, https://doi.org/ 10.1007/s11044-010-9204-0.
Jo u
Multibody. Syst. Dyn. 24 (2010) 123-127.
[17] S. Djerassi, Stronge’s hypothesis-based solution to the planar collision-with-friction problem, Multibody. Syst. Dyn. 24 (2010) 493-515.
https://doi.org/ 10.1007/s11044-010-9200-4.
[18] N. Pandrea, N.D. Stănescu, A new approach in the study of frictionless collisions using inertances, Proc. Inst. Mech. Eng. 229 (2015) 1989-1996. https://doi.org/ 10.1177/0954406214553983. [19] A. P. Ivanov, Energetics of a collision with friction, J. Appl. Mech. 56 (1992) 527-534. https://doi.org/ 10.1016/0021-8928(92)90008-V. [20] W.J. Stronge, Impact mechanics, first ed., The press syndicate of the university of Cambridge, 28
Journal Pre-proof
Cambridge, 2000. [21] J.F. Liu, Z.Q. Hu, 3D analytical method for the external dynamics of ship collisions and investigation of the coefficient of restitution, Chin. J. Sh. Res. 12 (2017) 84-91. https://doi.org/ 10.3969/j.issn.1673-3185.2017.02.011. [22] M. Nishida, Y. Tanaka, DEM simulations and experiments for projectile impacting
https://doi.org/10.1007/s10035-010-0173-z.
oo
(2010) 357-368.
f
two-dimensional particle packings including dissimilar material layers, Granul. Matter. 12
[23] H. Zhang, T.J. Li, M.L. Liu, Z.Y. Huang, H.L. Bo, DEM Simulation and analysis of collision
e-
https://doi.org/10.7538/yzk.2017.51.12.2212.
pr
between single coarse particle and wall, At. Eng. Sci. Technol. 51 (2017) 2213-2217.
Pr
[24] G.K.P. Barrios, R.M. de Carvalho, A. Kwade, L.M. Tavares, Contact parameter estimation for DEM simulation of iron ore pellet handling, Powder Technol. 248 (2013) 84-93.
al
https://doi.org/10.1016/j.powtec.2013.01.063.
Jo u
156-171.
rn
[25] H. Hertz, Ueber die Berührung fester elastischer Körper, J. Reine Angew. Math. 92 (1882)
[26] R.D. Mindlin, H. Deresiewicz, Elastic spheres in contact under varying oblique forces, J. Appl. Mech. 20 (1953) 327-344.
https://doi.org/10.1007/978-1-4613-8865-4_35.
[27] D.D. Han, D.X. Zhang, H.R. Jing, L. Yang, T. Cui, Y.Q. Ding, Z.D. Wang, Y.X. Wang, T.L. Zhang, DEM-CFD coupling simulation and optimization of an inside-filling air blowing maize precision seed-metering device, Comput. Electr. Agri. 150 (2018) 426-438. https://doi.org/ 10.1016/j.compag.2018.05.006. [28] R. Cabiscol, J.H. Finke, A. Kwade, Calibration and interpretation of DEM parameters for simulations of cylindrical tablets with multi-sphere approach, Powder Technol. 327 (2018) 29
Journal Pre-proof
232-245.
https://doi.org/ 10.1016/j.powtec.2017.12.041.
[29] D. Markauskas, Á. Ramírez-Gómez, R. Kačianauskas, E. Zdancevičius, Maize grain shape approaches for DEM modeling, Comput. Electr. Agri. 118 (2015) 247-258. https://doi.org/ 10.1016/j.compag.2015.09.004. [30] Y.L. Han, F.G. Jia, Y. Zeng, L.W. Jiang, Y.X. Zhang, B. Cao, Effects of rotation speed and
f
outlet opening on particle flow in a vertical rice mill, Powder Technol. 297 (2016) 153-164.
oo
https://doi.org/10.1016/j.powtec.2016.04.022.
pr
[31] X.M. Wang, J.Q. Yu, F.Y. Lv, Y. Wang, H. Fu, A multi-sphere based modelling method for
e-
maize grain assemblies, Adv. Powder. Technol. 28 (2017) 584-595.
Pr
https://doi.org/10.1016/j.apt.2016.10.027.
[32] C.M. Wensrich, A. Katterfeld, Rolling friction as a technique for modelling particle shape in
al
DEM, Powder Technol. 217 (2012) 409-417.
rn
https://doi.org/ 10.1016/j.powtec.2011.10.057.
Jo u
[33] J. Hlosta, D. Žurovec, J. Rozbroj, Á. Ramírez-Gómez, J. Nečas, J. Zegzulka, Experimental determination of particle–particle restitution coefficient via double pendulum method, Chem. Eng. Res. Des. 135 (2018) 222-233.
https://doi.org/ 10.1016/j.cherd.2018.05.016.
[34] L.J. Wang, R. Li, B.X. Wu, Z.C. Wu, Z.J. Ding, Determination of the coefficient of rolling friction of an irregularly shaped maize particle group using physical experiment and simulations, Particuology. 38 (2018) 185-195.
https://doi.org/10.1016/j.partic.2017.06.003.
[35] M. Molenda, S.A. Thompson, I.J. Ross, Friction of wheat on corrugated and smooth galvanized steel surfaces, J. Agric. Eng. Res. 77(2000) 209-219. https://doi.org/10.1006/jaer.2000.0591. 30
Journal Pre-proof
[36] H.N. Pishkenari, H.K. Rad, H.J. Shad, Transformation of sliding motion to rolling during spheres collision, Granul. Matter. 19 (2017) 69-81. https://doi.org/10.1007/s10035-017-0755-0. [37] N.V. Brilliantov, N. Albers, F. Spahn, T. Pöschel, Collision dynamics of granular particles with adhesion, Phys. Rev. E 76 (2007) 302-313. https://doi.org/10.1103/PhysRevE.76.05130. [38] B. Buck, J. Lunewski, Y.Tang, N.G. Deen, J.A.M. Kuipers, S. Heinrich, Numerical
oo
f
investigation of collis ion dynamics of wet particles via force balance, Chem. Eng. Res. Des. 132 (2018) 1143-1159. https://doi.org/10.1016/j.cherd.2018.02.026.
spheres,
Phys.
Rev.
e-
elastoplastic
pr
[39] H.A. Burgoyne, C. Daraio, Strain-rate-dependent model for the dynamic compression of E.
89
(2014) 032203.
Pr
https://doi.org/10.1103/PhysRevE.89.032203
[40] R. Cross, Oblique impact of a steel ball, Powder Technol. 351 (2019) 282-290.
rn
al
https://doi.org/10.1016/j.powtec.2019.04.038. [41] Z. Hu, X.Y. Liu, W.N. Wu, Study of the critical angles of granular material in rotary drums
Jo u
aimed for fast DEM model calibration, Powder Technol. 340 (2018) 563-569. https://doi.org/10.1016/j.powtec.2018.09.065. [42] Z. Yan, S.K. Wilkinson, E.H. Stitt, M. Marigo, Discrete element modelling (DEM) input parameters: understanding their impact on model predictions using statistical analys is, Comp. Part. Mech. 2 (2015) 283-299. https://doi.org/10.1007/s40571-015-0056-5. [43] C.G. Montellano, J.M. Fuentes, E. Ayuga-Téllez, F. Ayuga, Determination of the mechanical properties of maize grains and olives required for use in DEM simulations, J. Food Eng. 111 (2012) 553–562. https://doi.org/10.1016/j.jfoodeng.2012.03.017.
31
Journal Pre-proof
[44] Z. Chu, Z.Y. Wu, L.P. Chen, J.W. Ding, Fast MARS based on golden section method, J. Syst. Simul. 24 (2018) 1561-1566.
https://doi.org/10.16182/j.cnki.joss.2012.08.001.
[45] J. Horabik, A. Sochan, M. Beczek, R. Mazur, M. Ryżak, P. Parafiniuk, R. Kobyłka, A. Bieganowski, Discrete element method simulations and experimental study of interactions in 3D granular bedding during low-velocity impact, Powder Technol. 340 (2018) 52–67.
Jo u
rn
al
Pr
e-
pr
oo
f
https://doi.org/10.1016/j.powtec.2018.09.004.
32
Journal Pre-proof
Figure legends Fig.1. Particle-particle collision. Fig.2. Dimension parameters of wedge maize grain. H is the height, W is the width of the lower shoulder, w is the width of the upper shoulder, T is the thickness of the lower shoulder, and t is the thickness of the upper shoulder.
f
Fig.3. Simulation model of the wedge maize grain.
oo
Fig.4. (a) Platform for testing the kinematic COR of the particle-particle collision. (b) Simple
pr
geometry of the platform presented in the DEM simulation.
e-
n Fig.5. The interaction force FG acting on the particle 1, the normal contact force FG , the
Pr
g t f tangential contact force FG , the gravity FG , and the friction force FG .
Fig.6. Variations in the normal relative velocity v cz with the collision time.
between the sliding velocity and the sliding velocity in the
al
Fig. 7. (a) Variation in the angle
rn
x direction with the collision time. (b) Variations in the sliding velocity with the collision time.
Jo u
Fig.8. (a) The component forces FR passing through the centroid of the particle of the interaction force, the component force of the normal contact force contact force
FRt ,
FRn ,
the component force of the tangential
and the component force of the friction force FRf . (b) Translational kinetic
energy Ek . Fig.9. (a) Rotational torque T acting on the contact point generated by the interaction force, the X component of the rotational torque TX , the Y component of the rotational torque TY , and the Z component of rotational torque TZ . (b) Rotational kinetic energy Erk .
33
Journal Pre-proof
Fig.10. Translational and angular velocities of the particle before and after the collision: (a) translational velocity and (b) angular velocity. Fig.11. The effect of the input value of the COR einput on the energetic COR ecalculated calculated based on simulation result when the impact angle is 60° and the impact velocity is 0.9 m·s -1 .
f
Fig. 12. Images of the rotating maize grains ( front view): (a) t0 and (b) t n .
oo
Fig.13. Numerical and experimental comparison of the translational velocity and Y component of
pr
the angular velocity at the different impact angles when the impact velocity is 0.9 m·s -1 : (a) X
e-
component of the translational velocity V X , (b) Y component of the translational veloc ity VY , (c)
Pr
Z component of the translational velocity VZ , and (d) Y component of the angular velocity ωY . Fig.14. Numerical and experimental comparison of the translational velocity and Y component of
al
the angular velocity at the different impact velocities when the impact angle is 60°: (a) X
rn
component of the translational velocity V X , (b) Y component of the translational veloc ity VY , (c)
Jo u
Z component of the translational velocity VZ , and (d) Y component of the angular velocity ωY . Fig.15. The comparison between the input and calculated COR at the different impact angles and velocities : (a) the impact angle varying from 40° to 60° and the impact velocity is 0.9 m·s -1 and (b) the impact velocity changing from 0.9 m·s -1 to 1.7 m·s -1 and the impact angle is 60°. Fig.16. The effects of different factors on the energetic COR: (a) impact angle and (b) impact velocity. n
Fig.17. The effects of different factors on the normal contact force FG and tangential contact t
force FG :(a) impact angle and (b) impact velocity.
34
Journal Pre-proof
Jo u
Fig.2.
rn
al
Pr
e-
pr
oo
f
Fig.1.
35
Journal Pre-proof
Fig.3.
(a)
Jo u
rn
al
Pr
e-
pr
oo
f
Fig.4.
(b)
36
Journal Pre-proof
Fig.5.
1.0
FnG FtG FgG FfG
FG /(N)
0.8
0.6
oo
f
0.4
pr
0.2
0 0.2
0.4
0.6 0.8 t /(s)
1
1.2
1.4 1.5 ×10-4
al
Pr
e-
0
rn
Fig.6.
Jo u
v cz /(m·s-1)
0.5
0.0
compression
restitution
-0.5
0
0.2
0.4
0.6
0.8
t /(s)
37
1
1.2
1.4
1.5 ×10-4
Journal Pre-proof
Fig.7.
θ /(°)
300
200
deceleration slip in a positive direction
100
deceleration slip in a reverse direction
acceleration slip in a reverse direction
(a)
v cx
0.10
vcy 0.05
f
v cx vcy /(m·s-1)
0
-0.05
0
0.4
0.2
0.6
0.8
1
1.2
(b)
1.4
1.5 ×10-4
rn
al
Pr
e-
pr
t /(s)
oo
0.00
Jo u
Fig.8.
FR /(N)
1 0.5
FnR FtR FfR
0
Ek /(10-4 J)
(a) 1.4
Ⅰ
Ⅱ
particle 1 particle 2
Ⅳ
Ⅲ
1.3 1.2 (b) 1.1 0
0.2
0.4
0.8
0.6
t /(s)
38
1
1.2
1.4
1.5 ×10-4
Journal Pre-proof
T /(10-3 N·m)
Fig.9.
2 ( a) 0 TX TY TZ
-2
Ⅱ
Ⅰ
particle 1 particle 2
Ⅳ
Ⅲ
f
10 5
0
0.4
0.2
0.6
0.8
1
e-
pr
t /(s)
oo
Erk /(10-6 J)
-4 15
0.9 0.8
0.5
particle 1
particle 2
rn
V /(m·s-1)
0.6
al
0.7
VX VY VZ
Pr
Fig.10.
0.4
Jo u
0.3 0.2 0.1 0.0
Pre-collision
Post-collision
(a)
39
1.2
( b) 1.4
1.5 ×10-4
Journal Pre-proof 80
ωX 70
ωY ωZ
ω /(rad·s-1)
60
particle 1 50
particle 2
40 30 20 10 0
Pre-collision
Post-collision
e-
pr
oo
f
(b)
Pr
Fig.11. 0.52
al
0.48 0.46
rn
ecalculated
0.50
Jo u
0.44
0.45
0.46
0.47
einput
40
0.48
0.49
0.50
Journal Pre-proof
Jo u
rn
al
Pr
e-
(a)
pr
oo
f
Fig.12.
(b)
41
Journal Pre-proof
Fig.13. 0.28
Simulated particle 1 Simulated particle 2
0.26
Experimental particle 1 Experimental particle 2
VX /(m·s-1)
0.24 0.22 0.20 0.18 0.16
40
45
50
oo
f
0.14 55
60
Pr
e-
(a)
pr
Impact angle /(°)
0.07
Experimental particle 1 Experimental particle 2
rn
0.04
Simulated particle 1 Simulated particle 2
0.03 0.02
Jo u
VY /(m·s-1)
0.05
al
0.06
0.01 0.00
-0.01
40
45
50
Impact angle /(°)
(b)
42
55
60
VZ /(m·s-1)
Journal Pre-proof
0.85
Simulated particle 1 Simulated particle 2 Experimental particle 1
0.80
Experimental particle 2
0.75
0.65
50
45
55
pr
40
oo
f
0.70
60
e-
Impact angle /(°)
Pr
(c)
105
95
rn
Simulated particle 1 Simulated particle 2 Experimental particle 1 Experimental particle 2
85
Jo u
ωY /(rad·s-1)
90
al
100
80 75 70 65 60
40
45
50
Impact angle /(°)
(d)
43
55
60
Journal Pre-proof
Fig.14.
0.40
Simulated particle 1 Simulated particle 2 Experimental particle 1
0.35
VX /(m·s-1)
Experimental particle 2 0.30
0.25
0.15 0.9
1.1
1.3
oo
f
0.20
1.5
1.7
Pr
e-
(a)
pr
Impact velocity /(m·s-1)
0.07
Simulated particle 1 Simulated particle 2
Experimental particle 1 Experimental particle 2
rn
0.06
al
0.08
0.04
Jo u
VY /(m·s-1)
0.05
0.03 0.02 0.01 0.00
-0.01 0.9
1.1
1.3
Impact velocity /(m·s-1)
(b)
44
1.5
1.7
Journal Pre-proof 1.6
Simulated particle 1 Simulated particle 2
1.5
Experimental particle 1
VZ /(m·s-1)
1.4
Experimental particle 2
1.3 1.2 1.1 1.0 0.9 0.8 0.9
1.7
1.5
1.3
1.1
f
Impact velocity /(m·s-1)
e-
pr
oo
(c)
Simulated particle 1
160
Pr
Simulated particle 2
Experimental particle 1 Experimental particle 2
120
al
ωY /(rad·s-1)
140
Jo u
80
rn
100
60
0.9
1.1
1.3
Impact velocity /(m·s-1)
(d)
45
1.5
1.7
Journal Pre-proof
Fig.15.
0.55
eEinput eKinput eEcalculated eKcalculated
0.50
e
0.45
0.40
0.35
45
50
55
oo
40
f
0.30 60
Pr
e-
(a)
pr
Impact angle /(°)
0.55
rn
0.40 0.35
0.25
Jo u
e
0.45
0.30
eEinput eKinput eEcalculated eKcalculated
al
0.50
0.9
1.1
1.3
1.5 -1
Impact velocity /(m·s )
(b)
46
1.7
Journal Pre-proof
Fig.16.
0.48
eE
0.44 0.40 0.36 0.32 45
50
Impact angle (°)
60
pr
oo
(a)
0.48
e-
0.44
eE
55
f
40
0.40
Pr
0.36 0.32
1.1
1.3
Impact velocity (m·s-1)
(b)
Jo u
rn
al
0.9
47
1.5
1.7
Journal Pre-proof
Fig.17.
F /(N)
1.8 1.6
40°
45°
1.4
55°
60°
50°
1.2
FnG
1 0.8
FtG
0.6
0 0.2
0.4
0.6
1
0.8
1.5
1.4
×10-4
Pr
e-
(a)
1.2
pr
t /(s)
oo
0
f
0.4 0.2
rn
al
2
1.1m·s-1
1.3m·s-1
1.5m·s-1
1.7m·s-1
FnG
Jo u
F /(N)
1.5
0.9m·s-1
1
FtG
0.5 0
0
0.2
0.4
0.6
0.8
t /(s) (b)
48
1
1.2
1.4
1.6 ×10-4
Journal Pre-proof
Tables Table 1 Kinematic COR of the wedge maize grain. Table 2 Physical parameters and their values in simulation. Table 3 Results from different iterations. Table 4 The energetic COR of the maize grain at the different impact velocities and angles.
f
Table 5 Numerical and experimental comparison of the X component of the translational velocity
oo
V X at the different impact angles and velocities.
e-
VY at the different impact angles and velocities.
pr
Table 6 Numerical and experimental comparison of the Y component of the translational velocity
Pr
Table 7 Numerical and experimental comparison of the Z component of the translational velocity
VZ at the different impact angles and velocities.
al
Table 8 Numerical and experimental comparison of the Y component of the angular velocity ωY
rn
at the different impact angles and velocities.
velocities.
Jo u
Table 9 The comparison between the input and calculated COR at the different impact angles and
49
Journal Pre-proof Table 1 Kinematic COR
Impact velocity(m·s -1 )
40
0.9
Average value 0.304
STDEV 0.032
45
0.9
0.353
0.026
50
0.9
0.401
0.017
55 60
0.9 0.9
0.418 0.438
0.021 0.039
60
1.1
0.404
0.032
60 60
1.3 1.5
0.365 0.308
0.025 0.030
60
1.7
0.257
0.026
Table 2 Parameters
Pr
Type
e-
pr
oo
f
Impact angle(°)
-3
Density, ρp (kg·m )
Galvanized steel
Poisson ratio, νp Shear modulus, Gp (Pa) Density, ρg (kg·m-3 )
al
Maize particle
Jo u
particle— particle
rn
Poisson ratio, νg Shear modulus, Gg (Pa)
particle—galvanized steel
Coefficient of restitution, epp Coefficient of static friction,
Value 1164 0.4 1.37×108 7850 0.28 8.10×1010 0.257~0.438 0.372
μ spp Coefficient of rolling friction, μ rpp
0.0607
Coefficient of restitution, epg Coefficient of static friction,
0.613 0.549
μ spg Coefficient of rolling friction,
0.0311
μ rpg
50
Journal Pre-proof Table 3
Iteration
Input value
Calculated value
The second iteration
0.412 0.463 0.464
0.5334 0.491 0.485
The third iteration
0.495 0.443
0.452 0.518
0.463
0.491
The first iteration
oo
f
Table 4
Impact angle(°)
Impact velocity(m·s -1 )
40 45
0.9 0.9
50
0.9
55 60
0.9 0.9
60 60
1.1 1.3
0.419 0.397
60 60
1.5 1.7
0.326 0.304
pr e-
0.334 0.403 0.425
al
Pr
0.434 0.474
rn Jo u
Table 5
Energetic COR
Simulation
Experiment
Impact Impact
Particle1
Particle2
Particle1
Particle2
velocity angle(°)
(m· s-1 )
Average
Average STDEV
value
Average STDEV
value
Average STDEV
value
STDEV value
40
0.9
0.2558
0.00335
0.25981
0.00324
0.25889
0.01195
0.26066
0.00769
45
0.9
0.22404
0.00773
0.22543
0.00726
0.22954
0.00831
0.2301
0.00659
50
0.9
0.19989
0.0043
0.20052
0.00575
0.20637
0.01248
0.20179
0.01209
55
0.9
0.17621
0.00548
0.17871
0.00468
0.17193
0.0103
0.17906
0.01391
60
0.9
0.15553
0.00715
0.15753
0.00271
0.15413
0.01115
0.15573
0.00762
60
1.1
0.17547
0.01007
0.1754
0.00802
0.16623
0.01217
0.16466
0.0082
60
1.3
0.21299
0.0086
0.21299
0.0086
0.19642
0.01357
0.19642
0.01357
60
1.5
0.27146
0.01088
0.26955
0.00852
0.27004
0.01212
0.26991
0.01055
60
1.7
0.34294
0.01354
0.34092
0.01205
0.33327
0.01037
0.34102
0.01976
51
Journal Pre-proof
Table 6
Simulation Impact
Impact
angle
velocity
(°)
(m· s )
Experiment
Particle1 -1
Particle2
Average
Particle1
Average
Average
STDEV value
Particle2
STDEV value
Average STDEV
value
STDEV value
0.9
0.01753
0.00722
0.01576
0.00974
0.01664
0.01127
0.01749
0.01155
45
0.9
0.00856
0.00575
0.01033
0.00515
0.02749
0.01378
0.00856
0.00575
50
0.9
0.0216
0.01972
0.02183
0.02036
0.01719
0.02267
0.01033
0.00515
55
0.9
0.00991
0.00682
0.01052
0.00604
0.02118
0.01416
0.02749
0.01378
60
0.9
0.02079
0.02105
0.02079
0.02105
0.01907
0.01438
0.02804
0.01623
60
1.1
0.02529
0.01398
0.02954
0.0179
0.01108
0.01207
0.01633
0.01001
60
1.3
0.03189
0.01664
0.03189
0.01664
0.03356
0.02005
0.03356
0.02005
60
1.5
0.02323
0.02006
0.02283
0.02071
0.01823
0.01123
0.01856
0.01089
60
1.7
0.02453
0.01998
0.02453
0.01994
0.01946
0.01537
0.01946
0.01538
e-
pr
oo
f
40
Pr
Table 7
Simulation Impact
angle
velocity
(°)
(m· s )
Particle1 Average
Average
Average
STDEV value
Particle1
STDEV
value
Particle2 Average
STDEV value
STDEV value
rn
-1
Experiment
Particle2
al
Impact
0.0091
0.65338
0.00654
0.65544
0.01027
0.65206
0.00933
0.00922
0.71969
0.01138
0.73146
0.00752
0.72326
0.01352
0.00918
0.76699
0.01423
0.76472
0.00843
0.77862
0.00839
0.79271
0.00678
0.79333
0.00715
0.7935
0.01119
0.79905
0.01872
0.82758
0.00444
0.83021
0.00828
0.83837
0.01173
0.82683
0.00831
0.97085
0.02802
0.98813
0.03064
0.99928
0.04334
1.01661
0.03585
1.3
1.24811
0.0194
1.24955
0.02094
1.25502
0.01519
1.23149
0.01426
60
1.5
1.3629
0.02072
1.35916
0.0212
1.36544
0.0121
1.35753
0.01305
60
1.7
1.56109
0.00878
1.54339
0.0223
1.55757
0.01138
1.56366
0.00512
0.9
0.65463
45
0.9
0.72646
50
0.9
0.77426
55
0.9
60
0.9
60
1.1
60
Jo u
40
52
Journal Pre-proof Table 8
Simulation Impact
Impact
angle
velocity
(°)
(m· s-1 )
Experiment
Particle1
Particle2
Average
Particle1
Average
Average
STDEV value
Particle2 Average
STDEV
STDEV
value
STDEV
value
value
0.9
98.19266
2.42444
99.4543
2.11521
95.606
3.31989
96.98
3.60305
45
0.9
92.33102
1.42073
93.09682
1.00399
89.26
2.61209
88.662
2.79682
50
0.9
81.7773
1.01063
80.69374
1.82674
78.312
0.80061
78.36
2.32013
55
0.9
75.8215
1.28341
76.03728
3.11258
74.62
1.2518
73.9
1.95448
60
0.9
67.08734
0.88628
68.30458
1.07859
65.9
2.05426
65.36
1.35388
60
1.1
75.5266
1.38429
76.89626
1.53528
69.26
1.78269
70.66
2.56671
60
1.3
88.98746
0.82203
88.1176
0.99218
85.26
2.28539
83.84
1.48425
60
1.5
126.9478
2.68354
124.6128
4.21623
121.82
2.96766
120.72
2.84377
60
1.7
153.475
3.15357
153.1688
2.75525
154.62
3.205
150.62
6.80419
Pr
e-
pr
oo
f
40
(°)
velocity -1
(m·s )
45
0.9
0.334
Jo u
40
eEinput
rn
Impact Impact angle
al
Table 9
eEcalculated
eKcalculated
eKinput Average value
STDEV
Average value
STDEV
0.304
0.3306
0.00439
0.42198
0.00581
0.9
0.403
0.353
0.3986
0.00522
0.44006
0.00841
0.9
0.425
0.401
0.427
0.00406
0.45788
0.00579
0.9
0.434
0.418
0.4334
0.00462
0.4774
0.00488
0.9
0.474
0.438
0.469
0.00539
0.52266
0.00605
60
1.1
0.419
0.404
0.421
0.0051
0.50276
0.00558
60
1.3
0.397
0.365
0.3932
0.00789
0.4215
0.00634
60
1.5
0.326
0.308
0.3256
0.00365
0.3887
0.00559
60
1.7
0.304
0.257
0.3076
0.00503
0.3532
0.00466
50 55 60
53
Journal Pre-proof
Jo u
rn
al
Pr
e-
pr
oo
f
Graphical Abstract
54
Journal Pre-proof Highlights · The calculation model of the energetic coefficient of restitution was derived. · The kinetic energy of particle during collision was investigated. · The velocity of particle is affected by the interaction force and kinetic energy. · The energetic coefficient of restitution of maize grain was determined.
Jo u
rn
al
Pr
e-
pr
oo
f
· The collision period was classified based on the sliding velocity and deformation.
55