Professional Documents
Culture Documents
Azdez
Azdez
net/publication/271434423
CITATIONS READS
2 229
3 authors:
Chris Saunders
University of South Australia
56 PUBLICATIONS 463 CITATIONS
SEE PROFILE
Some of the authors of this publication are also working on these related projects:
Optimising high speed no-till seeding with bentleg furrow openers View project
All content following this page was uploaded by Mustafa Ucgul on 29 April 2019.
Paper Number:
131618800
Abstract. Modeling of soil-implement interactions is a complex process due to the variability of the soil profile,
non-linear behavior of the soil material and the dynamic effect of the soil flow. An approach that gives a further
insight is the technique of discrete element modeling (DEM). Although ample eminent research has been
conducted regarding modeling of the soil-implement interaction in 2D, there is just a few studies regarding to
3D modeling of the soil-implement interaction. Although cohesion was taken into account in the available 3D
DEM studies, there is no study that considers both soil adhesion and cohesion. Additionally, in the available 3D
DEM studies soil particle contacts were assumed as pure elastic, which is not realistic. As such most of the
available literature has been focused on draft forces with no satisfactory prediction of the vertical forces. In this
study a DEM model is presented that considers the plasticity, cohesion and adhesion of the soil. The approach
to determine the required DEM parameters was to firstly model a non-cohesive soil with elastic/plastic contacts
that does not adhere to the tool and to then include cohesion and adhesion to the model. The DEM parameters
for a non-cohesive soil were determined by performing and then simulating an angle of repose and two
penetration tests. Cohesion and adhesion was added to the model using cohesive energy density values
between particle/particle and particle/tool, respectively. The validation of the model for a cohesive soil was
carried out using shear box tests and simulations in terms of three different levels of moisture contents of 4, 8,
and 12% for a sandy loam soil. The parameters; namely coefficient of friction, coefficient of rolling friction, and
integration time, were determined using a trial and error approach for a 10 mm radius spherical particle. EDEM
software was used in the simulations. As a last step of the study, the interaction between the soil and a sweep
tool which was experimentally studied by Fielke (1988) was simulated. A good correlation was obtained
between the measured and predicted draft and vertical forces showing that the method proposed is valid.
Keywords. DEM, draft force, simulation, sweep, vertical force.
The authors are solely responsible for the content of this meeting presentation. The presentation does not necessarily reflect the official
position of the American Society of Agricultural and Biological Engineers (ASABE), and its printing and distribution does not constitute an
endorsement of views which may be expressed. Meeting presentations are not subject to the formal peer review process by ASABE
editorial committees; therefore, they are not to be presented as refereed publications. Citation of this work should state that it is from an
ASABE meeting paper. EXAMPLE: Author's Last Name, Initials. 2013. Title of Presentation. In 2013 ASABE International Meeting 2013.
St. Joseph, Mich.: ASABE. For information about securing permission to reprint or reproduce a meeting presentation, please contact
ASABE at rutter@asabe.org or 269-932-7004 (2950 Niles Road, St. Joseph, MI 49085-9659 USA).
1. Introduction
The accurate modeling of soil-tool interaction is important to design energy efficient tillage tools. However,
optimization of a tillage tool requires expensive and time consuming field tests which may only be undertaken
at certain times of the year. Therefore precise modeling of the soil-tool interaction would be very useful to
provide fast, cheap, and accurate prediction of performance. Commonly used methods to model the soil-tool
interaction are; analytical (McKyes and Ali, 1977 and Godwin et al., 2007), empirical (Hendrick and William
1973 and Zhang and Kushwaha, 1999), and continuum numerical methods (Kushwaha and Shen., 1995;
Fielke, 1999; Karmakar and Kushawa, 2005; and Karmakar et al., 2009). Analytical methods can model the soil
failure but not the soil movement due their quasi-static assumptions. Even though empirical methods provide
practical information; performing experiments for each field and operation condition is not viable (Raji, 1999).
The continuum numerical methods are based on an assumption of continuity and can be used to predict tillage
related forces when a proper constitutive law is used. However, there is always a change in the soil structure,
with tillage creating soil translocation, deformation and separation. Mixing of soil layers cannot be modeled
using continuum numerical methods (Asaf et al., 2007). To overcome the deficiencies of the analytical,
empirical and continuum numerical models, the Discrete Element Modeling (DEM) which is a dis-continuum
numerical method used for modeling the mechanical behavior of granular materials, can be employed. In DEM,
interactions between the particles are calculated by using contact models governed by physical laws. After
calculating forces acting upon a particle, the position and orientation of the particle is then calculated by
integrating Newton’s second law of motion. An explicit scheme is used for the computation. The calculation
cycle is repeated until the system reaches balance or a specified total simulation time (EDEM, 2010). DEM
simulations can be run in 2D or 3D. Ideally, to get accurate results, the size and shape of the particles used in
the DEM simulations should be as close as possible to actual particle shape and size. However, as the number
of particles studied increases, more calculations and simulation time is required. In the case of 3D modeling of
soil-tool interaction, the use of actual soil particle sizes in DEM simulations is not viable due to the extremely
large number of particles. Therefore, the particle sizes used in the simulation must be larger than that of the
actual particle size (to decrease the number of the particles) and in this case the physical properties of the
large particles must be adjusted.
Much research has been carried out to model the soil-tool interaction in 2D DEM (Tanaka et al., 2000; Momozu
et al., 2003; Zhang and Li, 2006; Zhang et al., 2008; and Shmulevich et al., 2007), however, just a few studies
were carried out to model the soil-tool interaction in 3D DEM (Obermayr et al., 2011; Tsuji et al., 2011; Mak et
al., 2012; and Chen et al., 2013). The cohesion was considered in some of the available 3D DEM studies but
there was no 3D DEM study to date that has considered the soil cohesion and adhesion simultaneously. To
date, the DEM particle contacts were assumed as purely elastic without considering the plastic deformation
behavior of the soil. In addition, the main motives of the available studies were to predict and calibrate the draft
forces while no accurate prediction was achieved for the vertical forces. In the current study, a 3D DEM model
that considers the plastic deformation behavior (Hysteric Spring Contact Model) of the soil, the cohesion, and
adhesion was proposed to model the soil-tool interaction. The approach to determine the required DEM
parameters was to firstly examine a non-cohesive soil that did not adhere to the tool and to then include
cohesion and adhesion to the model. To validate the determined DEM parameters the interaction between the
soil and a sweep tool which was experimentally studied by Fielke, (1988) in a range of soil conditions was
simulated.
The calibration process was carried out by performing an angle of repose and two penetration tests, and
matching the simulation results using a modeled spherical particle with radii of 10 mm to test results. A trial and
error method was used to match the simulation to actual tests. The simulations were performed by using a
DELL Precision T7500 Intel (R) Xeon (R) CPU X5667 @ 3.07 GHz and 48GB RAM computer. EDEM 2.4
software was used for the simulations.
Figure 1. (a) Angle of repose test equipment, (b) Typical discharge pile, and (c)-(d) Simulation of angle of repose test
using EDEM.
Figure 2. (a) Penetration test equipment, (b) Tool dimensions of the circular disc, (c) Tool dimensions of the cone, and
(d)-(e) Simulation of penetration test using EDEM.
60 2 2
Energy (J)
Energy (J)
1.5 1.5
40
1 1
20
31.5° 0.5 0.5
0 0 0
0 20 40 60 80 100 120 0 20 40 60 0 20 40 60
Horizontal distance (mm) Displacement (mm) Displacement (mm)
Figure 3. (a) Pile profile results for angle of repose test, (b) Cumulative energy versus penetration for disc, and (c)
Cumulative energy versus penetration for cone
------Experiment, - - - -Simulation
2013 ASABE Annual International Meeting Paper Page 4 of 9
4. Integration of cohesion and adhesion with the HSCM
In order to model the cohesion and adhesion, their forces were added to the total normal contact forces in the
HSCM. The friction was assumed to restrict the tangential element motion in the governing equations (HSCM
equations), so the adhesion and cohesion forces were not added in the tangential direction. The magnitude of
the adhesion or cohesion forces were calculated, as per EDEM (2011) as,
Fa/c=ξ·Ac (19)
where ξ is the cohesion energy density which is defined as the energy needed to remove a particle from its
nearest neighbors divided by the total volume of the removed particle. In the current study the cohesion energy
density was assumed as the cohesion strength. Ac is the contact area which is calculated as,
Ac=π·rc2 (20)
where rc is the contact radius, defined by Hertz (1882) as,
rc=((3·req·Fns) / (4·Eeq))1/3 (21)
where req, Eeq, and Fns
are the equivalent radius, equivalent Young’s modulus, and normal contact force,
respectively. Eeq is defined as per EDEM (2011) as,
1/Eeq= (1-υa2)/Ea+ (1-υb2)/Eb (22)
where E and υ are the Young’s modulus and Poisson’s ratio for the individual particles a and b, respectively.
After calculating the contact, damping, and adhesion/cohesion force, Equation 10 becomes,
Fn=Fns+ Fnd+ Fa/c (23)
Fielke (1988) conducted a series of direct shear tests to measure the cohesion of his soil samples (sandy loam
soil taken from UniSA Tillage Test Track – 85% sand, 3% silt and 12% clay). In order to validate the contact
model, Fielke (1988)’s experimental work was simulated using EDEM 2.4 software. For the simulations the
DEM parameters determined for a beach sand (100% sand), as detailed in Table 1 were used. The simulations
were made using 10 mm nominal spherical radii particles with an actual radii varying randomly from 0.95 to
1.05 x the nominal size in a shear box of 60 x 60 x 40 mm. The particles were placed in a face centered cubic
packing arrangement with the particle spacing chosen so as to achieve the desired bulk density. In order to
apply the desired normal stress, the top wall of the shear box was designed as a servo wall as suggested by
Sadek et al. (2011). The vertical velocity of the top wall was changed automatically to achieve the desired
normal stress as set by Fielke (1988). The shearing process was simulated by keeping the upper shear box
stationary and moving the lower shear box horizontally at 0.020 mm/s. The shear stress was predicted as the
total force on the opposing wall of the lower shear box divided by the area of the sheared section (Figure 4).
Simulations were carried out for 53, 97.4, and 364.1 N normal forces and the results were compared to the test
results in Figures 5a to 5c. A comparison of the results in Table 2 of the measured and predicted values of
cohesion are plotted in Figure 5d and this shows there was on average an 18% lower value for cohesive
strength for the simulation results when using the beach sand DEM parameters, actual bulk density and the
cohesive energy density equal to the measured cohesive strength. By refining the DEM parameters to more
closely represent the wider mix of particle sizes of the Tillage Teat Track soil should improve the correlation.
(a) (b)
Figure 4. DEM simulation of direct shear test (a) Start and (b) End
120
120
Shear stress (kPa)
120 30
Simulated cohesive
Shear stress (kPa)
100 100
strength (kPa)
80 80 80 20
60 60 60
40 40 10
40
20 20 20
0
0 0 0 0 10 20 30
0 20 40 60 80 100120 0 20 40 60 80 100120 0 20 40 60 80 100120
Measured cohesive
Normal stress (kPa) Normal stress (kPa) Normal stress (kPa) strength (kPa)
Table 3. Tillage Test Track soil properties for compaction levels created by a 1 tonne roller of Fielke (1988)
Figure 6. (a) Definition of the share wing geometry (Fielke, 1988), and (b) and (c) Screen captures of soil-tool simulation taken
from EDEM
In the current study, Fielke (1988)’s experimental work was simulated for 4, 8, and 12 km/h tool speeds. For
the simulations the HSCM plus cohesion/adhesion and the DEM parameters provided in Table 1 were used. A
nominal 10 mm radius spherical particle was used in the simulations. Simulations were run in a virtual soil bin
whose dimensions were 2,500 mm long x 1,500 mm wide x 300 mm deep (Figures 6b and 6c). Each simulation
was repeated three times as there was a variation in results and the averages of the simulation results were
2013 ASABE Annual International Meeting Paper Page 6 of 9
taken as the final result.
The effect of the tool speed on tillage forces (draft and vertical downward force) at varying moisture contents
for 75 mm operation depth are given in Figure 7. As shown in Figure 7 (at all moisture content levels), the
predicted draft forces increased linearly with the increase of the tool speed while predicted vertical forces
increased non-linearly with the increase of the tool speed. The results presented in Figure 7 show that the
simulation results for the draft and vertical forces were closely correlated to the Fielke (1988)’s experimental
results, as shown in Figure 8.
1200 300
800 200
600
400 100
200
0 0
0 4 8 12 0 4 8 12
Speed (km/h) Speed (km/h)
(a) (b)
Figure 7.Effect of the tool speed on (a) Draft and (b) Vertical forces at varying moisture contents
800 300
Predicted draft force (N)
Predicted vertical force (N)
R² = 0.91 R² = 0.9176
600
200
400
100
200
0 0
0 200 400 600 800 0 100 200 300
Measured draft force (N) Measured vertical force (N)
(a) (b)
Figure 8. Correlation of predicted and measured (a) Draft and (b) Vertical forces
6. Conclusion
This study used a Hysteric Spring Contact Model to simulate particle-particle interactions in a plastic manner
and include soil cohesion between particles and adhesion of particles to implements as cohesive energy
density values. In order to run a timely 3D tillage simulation, 10 mm spherical radius particles were chosen and
appropriate DEM parameters were developed using an angle of repose, penetration of disc and cone and
direct shear test. By changing the particle packing density to achieve the actual soil density and setting the
cohesive energy density value to the soil’s measured cohesive strength, simulations were able to match the
measured values of soil cohesive strength within 18%. Simulation results were compared to test results of
Fielke (1988) for a 400 mm wide sweep tool operating at 75 mm depth at a range of speeds and soil
conditions. A close correlation was found for the measured draft and vertical forces showing that the DEM
method proposed to account for larger particle size, plasticity, cohesion and adhesion was valid.
Acknowledgements
The authors acknowledge the support of the University of South Australia for granting of a post graduate
scholarship to Mustafa Ucgul and the Australian Grains Research and Development Corporation (GRDC)
project USA00005 for funding the computer and software.