Professional Documents
Culture Documents
with Abstracts
Greetings!
We are pleased that you are able to join us at this, our third EDEM conference.
We have an excellent program lined up for you with speakers and posters detailing
EDEM successes in a wide variety of industry sectors and engineering disciplines.
Thank you to all who will be sharing their work with us over the next two days.
Despite all that we have achieved since developing the first version of EDEM software
in 2002, we have only just begun. We will continue to focus our software development
efforts on helping engineers to solve the real, everyday challenges faced in designing
bulk material handling, processing, and manufacturing operations and in researching
particle behavior in complex granular systems.
This is an exciting time for both DEM Solutions and our customers. I look forward to the
presentations, to speaking with you over the next few days, and to working with you in
the future.
Best regards,
John Favier
CEO DEM Solutions
March 31st 2011
Contents
Technical Sessions
Richard LaRoche
Vice President of Engineering, General Manager of DEM Solutions (USA)
Greater Boston Area, Massachusetts, United Sates of America
DEM Solutions provides the world-leading DEM simulation technology and the
simulation know-how to address the needs of companies who handle and process bulk
materials ranging from coal, ores, and soil to pellets, tablets and powders. Our mission is
to support our customers in-house engineering expertise with our software and
applications know-how to generate substantial return-on-investment through reduced
prototyping and testing costs, lower risk of rework and equipment malfunction, improved
control over final product and process quality, and accelerated product innovation.
Peter Wypych
Founder and General Manager of Bulk Materials Engineering Australia;
Associate Professor, Faculty of Engineering, University of Wollongong
Wollongong, New South Wales, Australia
Bulk Materials Engineering Australia (BMEATM) is licensed consultancy of the
University of Wollongong and provides optimal designs and solutions for bulk materials
handling plants and processes in all sectors of industry in Australia and around the world
(e.g. mining, minerals processing, export infrastructure, bulk ports and terminals,
shipping, power generation, and so on). Over the past 20 years, we have completed over
1000 projects for industry, and more than 300 companies and organizations have made
use of our bulk materials handling expertise and facilities. Our operation now is a part of
the SMART Infrastructure Facility (focusing on Simulation, Modeling and Analysis for
Research and Teaching)
Abstract
The reliable design and operation of bulk materials handling and processing plants can be
difficult when dealing with complex geometries and difficult-to-handle materials. Often a lack
of detailed analysis of bulk material flow and process boundary interactions can lead to
costly mistakes which can typically be identified easily once in operation. These problems
can occur due to inaccurate characterization during design, miscalculation of particle
trajectories and velocities, and a lack of engineering tools to thoroughly visualize and
analyze material flow through complex designs. This presentation describes the application
of EDEM simulation to bulk material plant design by identifying current issues and presenting
new methods of calibration and length-scale/dynamic validation. Examples and case studies
are presented to answer these key questions:
In this presentation, a partnership between Bulk Materials Handling Australia (BMEA) and
DEM Solutions will be introduced which will provide expert guidance and material calibration
services for EDEM customers worldwide.
Johannes Quist
Project Assistant, Chalmers Rock Processing Research (CRPR)
Gteborg, Sweden
Chalmers Rock Processing Research (CRPR) is a research group within the
Department of Product and Production Development focused on scientific and
industrial research on machines and systems for production of rock materials. The
objective is to produce rock material products in a cost-efficient and resourceeconomical way, thereby contributing to a sustainable society. The work of the
research group involves developing algorithms, operator interfaces, and simulation
and optimization techniques for optimal operation with an augmented degree of
usage
The Chalmers University of Technology Department of Product & Production
Development focuses on product design, product development and production
systems development and in the interplay between these disciplines. Research at the
Department focuses largely on shortening the lead-time from needs to finished
products while simultaneously achieving added value for the customer. Our largest
customers are in the vehicle and manufacturing industries.
Johannes Quist, Research Engineer, Dep. Product & Production Development, Chalmers
University of Technology, Gteborg, Sweden
Introduction
The usage of CAE tools for simulation and modelling of machines and processes is
continuously increasing at engineering departments and in academia. However efforts
towards ensuring trustworthy results by performing calibration tests are often a scarcity. This
may lead to inaccurate or incorrect results, leading to poor decisions. This reliability issue is
vital when it comes to a potential increase in the use of DEM modelling in the industry. When
conducting DEM modelling efforts aimed towards simulating granular media flow behaviour
there are few methods available for calibrating contact model parameters. In this paper a
calibration device is presented as a solution to the problem of choosing correct contact
model parameters that correspond to the flow behaviour of real media.
Approach
The device has been designed with the intension to create three different flow behaviour
situations that can be studied in a sequence. An illustration of the device can be seen in
Figure 1. The flow through the device is filmed as a reference for later reproduction of the
flow in EDEM. The device can be configured in different ways in order to create several
different flow scenarios. By configuring contact model parameters towards different
scenarios the validity of the model will increased compared to a static configuration. Also,
the device is adoptable to different kind of media shape and size.
Figure 1 - CAD model of the calibration device. Red areas represent surfaces important for
the media-geometry interaction behaviour.
Results
In the top section two plates form a chute controlled by a trap-door mechanism. The aperture
length can be varied between four different positions. The angle for the top plates can be
varied continuous between two positions as well as independently of each other. When
releasing the trapdoor mechanism the material will flow through the chute and fall onto the
angled plane in the middle section of the device. The angle of the plane can be between
three discrete steps. Finally the media will reach the bottom of the device and form an
angled bed of material.
Figure 2 - Snapshots from DEM simulation of the flow through the device.
Mical W. Johnstone
Product Support Engineer, Fluid Technology
Grimsby, United Kingdom
The Fluid Technology unit focuses on technology for developing, producing and
dealing with hoses and hose assemblies. Products developed in this unit are used to
control media flow in cars, trucks and most industrial applications, including offshore
activities.
Dunlop Oil and Marine, a member of the ContiTech Group, a specialist in rubber
and plastics technology, is a world leader in the design, manufacture and supply of
hoses for the oil, gas, petrochemical and dredging industries, for both offshore and
onshore-based operations.
At
was determined manually based on the mid-section of the surface profile. Simple
produced less accurate results as stray particles along the surface would significantly affect
the angle determination.
Confined compression apparatus
A confined compression test is used to investigate the mechanical response of a granular
bulk material subject to vertical compression in a confined cylinder. The confined
compression tester used is a modification of the Masroor et al. (1987) [3] apparatus. The
vertical load applied to the bulk solid is applied by a top platen driven by an INSTRON [4]
machine at a constant rate of 1mm/min. The INSTRON records the applied load and the
vertical displacement. A diagram of the experimental setup is shown in Figure 3 (b). In this
experiment, the bottom edge of the cylinder is free to move, allowing compression from both
the bottom platen as well as the top platen. The sample goes through a loading and
unloading cycle (to 65kPa) to observe both the initial compression and the elastic unloading
response. The speed of loading and unloading is kept constant throughout the experiment.
As the compression stress increases, the sample strain increases and vice versa. The void
ratio of a granular solid e is the ratio of the volume of void to the volume of solid. As the
sample is compressed, the volume of void decreases and therefore e decreases; the sample
is essentially being consolidated. The change in e can be evaluated using the dimensions of
the sample and the platen heights.
Figure 3
logarithmic scale. The advantage of this is that the unloading-reloading curve for a granular
solid is often linear in the semi-logarithmic plot. The unloading trend (triangles in Figure 3c)
ften used in geotechnical
engineering and is a good representation of the bulk unloading response. This means that
equation e = D
145
184
fr
Sample
380
Attached to
INSTRON
Top
plate
Strain gauges
250
Bottom
plate
140
Bottom
load cell
0.586
Sample void ratio, e
FT
Loading
Unloading
0.582
0.578
0.574
0.570
10
FB
The second step in the methodology is to create a numerical dataset that describes how the
DEM parameters influence the bulk responses when simulating the laboratory devices
numerically. The optimisation procedure described in this paper focuses on determining
DEM interparticle model parameters. The other parameters such as the DEM model
parameters for particle-boundary interactions are largely dependent on specific application
scenarios and should be determined separately. Previous parametric studies have been
carried out by studying the influence of a single parameter on the systems response [5,6],
however, this may not adequately capture the combined effects of the model parameters.
Increasing the number of independent model variables rapidly increases the total number of
simulations required to generate the numerical dataset. With 4 DEM parameters and 3
values per variable (e.g. for the shear modulus 1E6, 1E7 and 1E9 are used), running every
possible combination will require 34=81 runs. Design of experiment (DOE) methods were
used to reduce the number of simulations required to create the dataset using partial three
level factorial design [7].
The non-spherical particles in this study were represented using 2 equal overlapping
spheres. More accurate shape representation is possible but will require more spheres and
increase computational time. Some studies have suggested that accurate representation of
particle shape may not be necessary to produce satisfactory predictions, at least for densely
packed granular media under a variety of loading conditions [5,8]. Key DEM implementation
information is given in Table 1. The rotating drum was modelled to the full experimental
apparatus scale. The confined compression cylinder was scaled to 60% of experimental
scale as a previous parametric study revealed that scaling the system down to 60% only
produced a small effect on the unloading stiffness parameter (<10%) that will be factored
into the results.
1000
Poisson's ratio
0.3
1E6/1E7/1E9
Contact model
Coefficient of friction
PP
0.1/0.3/0.6
PB
0.3
PP
0.1/0.5/0.9
PB
0.5
PP
0.0/0.04/0.1
PB
0.0
Coefficient of restitution
2.5
1.20
compression tester of 1.50E-03 (n=3, COV=2.3%) where the initial void ratio was e0=0.46.
Step 4: Parametric optimisation using calibration data
The final step in the methodology determines optimised parameters by calibrating the
numerical results with the measured data. There are two main parts to the parametric
optimisation. First, a model is created using analysis of variances (ANOVA) based on the
numerical dataset and second, the model is calibrated using the experimental data by
response profiling [9] to determine a set of optimised parameters. The statistical analysis
package chosen to create the model in this paper was Statistica [10]. Using the dataset, a 3
level factorial ANOVA model is created based on the gravitas of the DEM parameters on the
bulk responses. The optimised parameters are determined using the response desirability
profiling algorithm in Statistica which is based on the simplex method of function optimisation
[11].
3. S.A. Masroor, L.W. Zachary, and R.A. Lohnes (1987) A test apparatus for determining
elastic constants of bulk solids, SEM Spring Conference on Experimental Mechanics,
Houston, TX, USA.
4. Instron (2009) www.instron.com. Norwood, MA, US.
5. Y.C. Chung (2006) Discrete element modelling and experimental validation of a granular
solid subject to different loading conditions, University of Edinburgh.
6. J. Hrtl (2008) A study of granular solids in silos with and without an insert, The
University of Edinburgh.
7. G.E.P. Box and D.W. Behnken (1960) Some new three level designs for the study of
quantitative variables, Technometrics, vol. 2, p. 455475.
8. Hrtl J. and Ooi J.Y. (2008) Experiments and simulations of direct shear tests: porosity,
contact friction and bulk friction Granular Matter, 10, 263-271.
9. G. Derringer and R. Suich (1980) Simultaneous optimization of several response
variables, Journal of quality technology, vol. 12, p. 214219.
10. StatSoft (2009) Statistica.
11. R. ONeill (1971) Function minimization using a simplex procedure, Journal of the
Royal Statistical Society, Series C (Applied Statistics), vol. 20, pp. 338-345.
Acknowledgements
The authors would like to thank DEM Solutions Ltd and the University of Edinburgh for their
support and discussion.
Nima Gharib
PhD Candidate, Neptec Rover Team (NRT),
Delft, The Netherlands
The Neptec Rover Team (NRT), which includes some of the industrys leading
technology experts, was brought together to investigate, conceptually design, and
test lunar mobility systems for the Canadian Space Agency. This highly experienced
team has been working together to develop technology for the new Lunar Exploration
Light Rover (LELR). The McGill University team focuses on the definition,
development and validation of a compliant wheel; on the effect of operating one or
more of the recommended mobility systems while in the presence of the fine,
abrasive dust on the lunar surface; and on the identification of strategies to mitigate
dust infiltration and component wear.
The Department of Mechanical Engineering at McGill University has a long
history of excellence in research and teaching. For more than a century, we have
been committed to train the next generation of innovators, industrial leaders and
academics.
Nima Gharib (PhD Candidate), Robin Briend (Graduate MSc Student), Nasim KavehMoghaddam
(Graduate MSc student), Peter Radziszewski (Associate Professor)
Department of Mechanical Engineering, McGill University, 3480 University St.,
Montreal, Quebec, Canada H3A 2A7
Introduction
Lunar mobility studies require a precise knowledge of the geotechnical properties of the
lunar soil in order to design efficient traction systems. Since the last Apollo missions, the
immense progress of computers on one hand, and the development of the discrete element
method on the other hand, provide ways to test traction system prototypes with simulation.
Before simulating vehicle displacements on the soil strictly speaking and measuring
terramechanics properties of the wheel-soil interactions, one needs to know exactly how to
model a given granular soil and make sure that its mechanical properties, for example its
dynamical response to loads or stresses, are accurately rendered. The angle of repose of a
soil is one of most basic soil property and can be helpful in the calibration of the contact
model and interaction parameters in DEM software. This study focuses on this characteristic
and tries to summarize the effects of different parameters on it, among them particle shape,
electrostatic charge, and gravity.
Simulation Setup
The main parameters used to model the soil are given in Table 1. In order to give accurate
3D extension of [2] results, we used the same values as they did for the simulation
parameters.
Parameter
Definition
value
R0
50 m
Number of particles
3000
Particle density
3000 kg/m3
0.2
5107 Pa
Particle-particle restitution
coefficient
0.5
es
0.5
ss
0.5
rs
0.2
To calibrate the model first a simple model was created in EDEM. It consists of a horizontal
steel plate on which a 1.5 mm diameter steel tube rests vertically. 3,000 particles were
generated inside the tube at the beginning of the simulation. The tube is then lifted upward at
the speed of 5 mm/s. After the tube walls lose its contact with the particle pile and the
particles have reached a static state, the simulation was then run for another half a second
in order to make sure that the pile has reached its stationary state. Finally the results were
compared to experimental data obtained using same tube diameter filled with Ottawa sand
(Fig. 1).
After calibrating the software and finding the proper parameters for the soil, the same
approach were carried out to investigate the effect of gravity, particles electric charge, and
particles shape on the formation of Angle of Repose (AOR).
Single spherical particles and the mixture of paired particle created by overlapping two
particles with either same radius or different radius were used in the simulations (Fig. 2).
Each case was studied under moon and earth gravity. Also the effect of electrical charge on
AOR was investigated by having uniform charge distribution for the particles stack.
Results
The angle of repose were calculated at the end of each simulation using a MATLAB code
uses a least-squares fit of the surface particles to define the slopes of the two sides of the
pile at twelve different planes. The angle of repose for the pile is the average angle from the
total 24 slopes. As shown in Fig. 3 the angle of repose of single particles is much smaller
than that of paired particles. Increasing electrostatic charges decreases the angle of repose.
However a weak electrostatic charge has almost no effect on it. While when the electrostatic
charge increases a certain point the angle of repose drops sharply. The gravity in the other
hand doesnt have much effect on angle of repose.
It is worth to mention the above-mentioned results are from the cases where friction between
the wall of the tube and the particles and the particles themselves was 0.5. Changing
coefficient of friction might change the results.
Fig. 1 Comparison between experimental and discrete element modeling of Angle of Repose
Fig 3. Angle of repose vs. electric charge for different particle shape
at earth and moon gravity
Discussion
In order to decrease the simulation time and have consistency between the results in was
decided to import the position of the grains from an External Factory. That would
dramatically reduce the simulation time specially when the particles were assigned with
electrical charge. The problem that one would face in EDEM it is not possible to import
charged particles. To overcome this problem first the position of whole particles in the stack
were imported without charge except a few particles which were created by a built-in factory
containing the charge of whole stack. The Hertz-Mindlin contact model was modified to
transfer the electrical charge from charged particle to their neighbors which is in contact with.
In This way after a few run times, the whole stack gets charged uniformly and then the
cylinder starts moving up.
To better model the situation on the moon it is recommended to bring in the tribocharging
feature during the simulation. In this way the particles charge will vary when they contact
each other and also the inner surface of the tube. And of course the charge distribution
wouldnt be uniform as it was in this study.
Acknowledgement
The authors would like to thank Neptec and CSA as well as NSERC CRD program for the
financial support of this project and also DEM Solutions. Ltd., Edinburgh, Scotland, UK, for
their help and their advices
References
[1] DEM Solutions, Ltd. (2010), EDEM 2.3 User Guide, Copyright 2010, Edinburgh,
Scotland, UK
[2] S. Ji and H. Shen. Two-dimensional simulation of the angle of repose for a particle
system with electrostatic charge under lunar and earth gravity. Journal of Aerospace
Engineering, 22:10_14, 2009.
[3] R.D. Mindlin. Compliance of elastic bodies in contact. Journal of Applied Mechanics,
16:259_268, 1949.
[4] T. Tanaka Y. Tsuji and T. Ishida. Lagrangian numerical simulation of plug flow of
cohesionless particles in a horizontal pipe. Powder Technology, 71:239-250, 1992.
David Curry
Senior Consulting Engineer
Edinburgh, Scotland
DEM Solutions provides the world-leading DEM simulation technology and the
simulation know-how to address the needs of companies who handle and process
bulk materials ranging from coal, ores, and soil to pellets, tablets and powders.
Michael Thorenson
ESTEQ Engineering is dedicated to helping companies develop products more
efficiently by enabling innovation through simulation and testing technology. This is
made possible through two separate and yet highly integrated disciplines, i.e.
Simulation and Testing.
ED Birch
Manager, Bulk Materials Handling Department
Johannesburg, South Africa
Brian Moore
Senior Consultant. Materials Handling
Wollongong, NSW, Australia
Moore has more than 30 years experience in bulk materials handling. His expertise
includes system design and life extension of materials handling systems for heavy
industry; mass flow bin design and stockpile reclaim systems and belt conveyor and
transfer chute design.
Martin Geissler
Design Engineer, Engineering Plow Systems, Longwall Product Group
Grimsby, United Kingdom
Engineering Plow Systems from Bucyrus, with features offered only by Bucyrus,
the inventor of the plow, our plow systems offer our customers future-oriented
solutions for international hard-coal mining.
Bucyrus is the world leader in the design and manufacture of high productivity
mining equipment for surface & underground mining and the global market leader
and supplier of complete longwall systems. Our equipment and systems are meeting
the demands of underground mining under the most stringent conditions around the
globe. Adapted to the mining challenges faced by our customers today, Bucyrus
customized systems range from hydraulic roof supports, automated plow systems,
shearers, face conveyors and drives to automation and roof support carriers.
David Curry
Senior Consulting Engineer
Edinburgh, Scotland
DEM Solutions provides the world-leading DEM simulation technology and the
simulation know-how to address the needs of companies who handle and process
bulk materials ranging from coal, ores, and soil to pellets, tablets and powders.
Robert Fitzpatrick
Experimental Officer for Mineral Processing
Cornwall, United Kingdom
As Experimental Officer for Mineral Processing, Fitzpatrick is responsible for
independent research into the field of Minerals Processing, teaching students and
undertaking contract research for external bodies. Current research areas include
investigations into the relationship between throughput and separation efficiency for
Sensor-Based Sorters, investigations into the use of unsupervised clustering and
related techniques as alternate methods of training and monitoring sensor-based
sorters to improve their flexibility and adaptability to minerals applications, and
Investigations into a multi-sensor approach to sensor based sorting, combining
optical, inductive and near infrared (NIR) sensor data to improve separation
efficiency.
Camborne School of Mines (CSM) at the University of Exeter is one of the
worlds most famous mining schools. Founded in 1888, CSM now has a unique
combination of scientific and engineering expertise in renewable energy, geology,
mining and minerals processing and applies this to world-leading research and
teaching. The research undertaken in the Minerals Processing Department is an
important aspect of this work.
Introduction
The hand sorting of objects into groups of similar properties may be considered as one of
the earliest forms of technology. Sensor-based sorting is a technique which seeks to
replicate or improve on the hand sorting process by replacing the human eye and hand with
machine vision and automated ejectors. Specifically, sensor based sorting is an automated
separation technique which exploits measurable differences in the physical properties of
particles, either natural or induced, to produce a distinct response to an applied force
(1Manouchehri, 2003; 2Walsh, 1989). A schematic of a typical sensor-based sorter is shown
in figure 1.
The sorter can be considered as four interlinked unit operations, shown diagrammatically in
figure 2.
Data Analysis
Input Feed
Feed
Preparation
Particle
Examination
Ejection System
Output Streams
The first operation is the preparation of the feed material which can consist of several stages
including, sizing, washing and/or wetting followed by feed rate control, particle alignment,
acceleration and stabilisation (1Manouchehri, 2003). This stage is essential to maximise the
likelihood of collecting sensor data which is representative of the true physical properties of
particles and to ensure that particles are examined singularly. The next operation is the
actual collection of data on individual particles. This is undertaken using one or more
sensors operated in series or parallel. The third operation is that of data analysis, where the
sensor data is utilised by a CPU to classify particles according to pre-determined rules and
thresholds. In the last operation the particles are physically separated into two or more
output streams based on the classification decision. A number of methods have been
employed to this end including air and water jets and mechanical paddles.
The focus of this paper is on the use of sensor-based sorting in the mining and minerals
industry. The primary aim of a mining operation is to completely separate valuable, sellable
minerals from waste rock as cheaply as possible. As such a sensor-based sorting process is
evaluated based on the degree to which it can separate mineral types, i.e. separation
efficiency, and its cost.
For sensor-based sorting the separation efficiency is a function of a machines ability to:
generate sensor data which is representative of physical properties; correctly classify
particles based on this sensor data and then to accurately and reliably actualise the
separation of particles. The cost of sorting is minimised by operating at a high throughput.
Unfortunately, each of the factors which determine separation efficiency is adversely
affected by an increase in throughput due to the increased probability of particles being
examined in close proximity (see figure 3).
Low Throughput
Particle Flow
Direction
High Throughput
Decreasing
Average
Particle
Proximity
Approach
To move towards a solution to this problem, it was decided to investigate the use of discrete
element modelling (DEM) to predict the relationship between throughput and separation
efficiency of a sensor-based sorter by replicating physical experimentation undertaken on a
CommoDas optical sorter and comparing the results. The aim was to compare the proximity
of particles at the end of the feed conveyor for both the simulation and actual tests. These
proximities can be directly related to the expected efficiency of the machine. To test the
models, they would then be used to predict the particle proximities and therefore efficiencies
at other throughputs. These would again be compared to actual test data.
To undertake the modelling, Autodesk AutoCAD 2008 was used to create a representation
of the feed mechanism for the optical sorter which was then imported into the EDEM 2.3
particle simulation software, provided by DEM Solutions. Ltd., Edinburgh, Scotland, UK.
Figure 4 on the next page contains an image of the CommoDas optical sorter and of the
CAD representation. It includes:
The feed hopper and vibrating feeder used to control the throughput to the sorter
The chute used to stabilise and disperse the particles
The conveyor used to accelerate and separate the particles
For the DEM simulations, particles were placed in the feed hopper and by the careful
manipulation of sinusoidal translation the throughputs were calibrated to match those used
during physical testing by matching the rate of particles arriving at the end of the conveyor.
The behaviour of particles at these set throughputs was then investigated and their velocity
and positioning tracked. This information was utilised to determine the particle proximity
using a methodology developed by the research team.
Feed Hopper
Conveyor
Chute
Particles were created to simulate three size and two shape fractions used during physical
testing. The material models for these and the feed mechanism were calibrated by
comparing the behaviour of the particles with video footage taken of the physical testing,
focussing on the change in the distribution of particles along the width of the conveyor over
time.
Results
The results of the physical separation of particles using the CommoDas optical sorter are
currently incomplete so comparisons with the simulations have not, as yet, been undertaken.
However, initial test work suggests that the DEM simulations provide an accurate
representation of the feed mechanism. For example, table 1 summarises the distribution of
particles along the width of the feed conveyor for both simulated and actual tests.
The data shows a good correlation between the simulated and actual horizontal distribution
of particles on the feed conveyor. This suggests that the simulations are an accurate
representation of the physical properties and interactions of the feed mechanism and
particles.
Using particle data obtained from the DEM simulations it was possible to create a distribution
of particle proximity. When physical testing is complete this will be compared with data
collected from the optical sensor. An example proximity distribution is shown in figure 5.
Cumulative Frequency
80.0%
60.0%
40.0%
20.0%
0.0%
10
100
1000
10000
Discussion
The ability to quickly and accurately predict the relationship between throughput and
separation efficiency would drastically reduce the costs involved in implementing a sensorbased sorter within the mining and minerals industry. The experimentation undertaken is a
step towards this goal; by using DEM to simulate the feed mechanism of a sorter, the affect
References
1. Manouchehri, H.R., 2003. Sorting in Mining, Mineral Processing and Waste Utilization:
(History, Innovations, Applications, Possibilities, Limitations, and Future) Swedish Mineral
Processing Research Organisation, Stockholm.
2. Walsh, D.E., 1989. What Sort of Ore Sorter?, Alaska Science Forum. Geophysical
Institute, Fairbanks, Alaska.
Rodrigo M. de Carvalho
Research Assistant, Laboratory of Mineral Processing, Department of
Metallurgical and Materials Engineering
Rio de Janeiro, Brazil
The Laboratory of Mineral Processing is well reputed worldwide for its research in
comminution, ranging from the understanding of fundamentals of breakage to the
development and application of modeling and simulation tools to improving industrial
applications involving size reduction.
Universidade Federal do Rio de Janeiro (UFRJ) is the second largest university in
Brazil, and houses the graduate school of engineering (COPPE). Part of COPPE, the
Department of Metallurgical and Materials Engineering is recognized in Brazil as one
of the leaders in research and development in the various fields from mineral
processing to materials science.
Introduction
The discrete element method (DEM) has been used actively for the last 20 years or so to
describe what happens inside grinding mills [1] and, more recently, crushers [2]. Whereas
this technique found almost immediate application in aiding the operator in identifying and
preventing the mill speed and filling that would lead to ultraprojection of grinding media in
semi-autogenous or ball mills for a given mill liner configuration, the application of the
technique as a quantitative tool to predict media and liner wear and size reduction has not
yet reached the maturity required for industrial application [3].
One of the limitations identified in the past has been overcome by the rapid evolution of
computational power, making it possible to simulate, in 3D, the motion of particles both
grinding media and balls inside industrial-scale grinding mills and crushers. In the case of
predicting size reduction in comminution machines, the greater challenge remains in
coupling the information that DEM provides, regarding the mechanical environment and on
the flow of solids through the vessel, to material breakage properties that lead to the main
outcome of a comminution process, which is the size distribution of the product. In order to
tackle this problem, three different approaches can be used, which are described as follows.
In the first case, DEM simulates not only the mechanical environment, but also the entire
breakage processes. Particles are built from an aggregate of smaller particles united by a
cohesion force, and breakage occurs when the aggregate receives enough energy to break
the cohesion force among these smaller particles [4,5]. In this case, the outcome from each
breakage event, that is, the breakage of an aggregate, has to be simulated before simulation
starts, which makes it extremely computationally intensive. The second type of models is the
one in which, although particles are also present in the DEM simulations, the outcome from
each breakage event is not calculated by the method, being rather provided by an empirical
model calibrated using particle breakage tests. After each breakage event the broken
particle is replaced by an aggregate of smaller particles, the breakage product. This
approach, called fast breakage model has been applied successfully to prediction of cone
crushing [2]. Although more computationally efficient than the first approach, both groups of
models are very demanding of computer power, since new particles are almost continuously
being created in the system.
The third approach, which is the one proposed by the authors, is to use DEM solely to
simulate motion of media and to track the collisions among particles, leaving the population
balance model the task of coupling the collision energy information to the models describing
particle response to each stressing event in order to calculate the overall breakage product.
This approach is particularly well suited to media mills, such as ball, stirred and semiautogenous mills. By using this approach the computational power required to run DEM
simulation is significantly smaller than the one required by the first two types of models. This
is illustrated in Figure 1, which demonstrates that the differences in computing power
required by the different approaches become particularly significant when mills of large
diameters are simulated.
Approach
The model initially relies on detailed ore breakage properties which can be obtained by a
number of laboratory tests, describing the ore response to the different breakage
mechanisms (body and surface breakage), the breakage probability and the response to
unsuccessfully breakage events (particle damage) [6].
Once the material (ore) breakage properties are all characterized the mechanistic model
developed by the Authors can be used to simulate many different comminution processes, if
appropriate description of how the energy is transferred to particles are given. This
information can be obtained using DEM simulations.
Figure 2 shows the model framework, having on the left side what model inputs are
necessary to get predictions. As such, PBM requires information related to ore properties,
including breakage properties and feed size distribution. DEM simulations require contact
parameters from ore and the equipment and also equipment design and operational
parameters. Then DEM gives PBM information of the impact energy distribution. The PBM
calculations, which are the core of this model approach, then give predictions of product
particle size distribution, fracture energies (particle strength) and flowrates. This approach
can also be coupled with CFD simulations in order to get information of transport rates and
they can also affect DEM and PBM calculations.
One particular comminution machine of interest is the ball mill. In this device, the mechanical
environment is dominated by the grinding media (typically steel balls), so that nearly all
breakage results from collisions between balls. Particle breakage occurs when ball hits a
particle bed sat upon another steel ball. If particles within this bed have fracture energy
bellow the energy given by ball collision, it may fracture. For this device, no ore particles are
modeled inside DEM and the ore phase is considered as a continuous phase. This requires
an appropriate characterization of DEM contact parameters as the impact between two steel
balls is assumed to involve a virtual bed composed of ore particles. As particle size
distribution has almost no effect on impact energy spectrum given by DEM simulations it is
called one-way coupling, as is also illustrated in Figure 2 where arrows represents the
information flow between model blocks. Details of the model applied to batch ball mills may
be found elsewhere [7].
In contrast to that, two-way coupling must be used whenever particles contained in the ore
also act as grinding media. This is the case of the semi-autogenous mill, in which coarser
ore particles also act as grinding media. These particles have to be considered in order to
get appropriated description of the impact energy spectrum. However, since they change
their size during the process, DEM simulations have to be updated for every significantly
changes in the mill hold-up in an iterative process with PBM simulations. This is called twoway coupling and is illustrated in Figure 2 by the dashed arrows.
Model Inputs
Model Outputs
CFD
Particles
(ore)
Particle Size
Distribution
Material flow
Breakage
properties
Size
distribution
Contact
parameters
PBM
Operational
Fracture
energies
Flow rates
Particle Size
Distribution
Impact energy
distribution
Equipment
Design
Particle Size
Distribution
DEM
Figure 2: Model framework showing the coupling between PBM and DEM and also CFD
Results
In order to validate the model, batch grinding experiments were carried out using a
laboratory ball mill (30x30cm), rotating at 54 rpm. The mill is loaded with 467 steel balls of 25
mm diameter. A certain amount of ore particles were fed to the mill in order to get 100% of
voids between balls filled. DEM simulations of the ball load movement were made using
EDEM 2.2 particle simulation software provided by DEM Solutions. Ltd., Edinburgh,
Scotland, UK. The mill is shown on Figure 3 as well as its DEM simulation snapshot.
Figure 3: Real lab ball mill (left) and a snapshot of it running on EDEM (right).
Experimental particle size distributions after each grinding time were compared to that
predicted by the mechanistic model, Figure 4 shows the results for two different ores at
different initial conditions in terms of particle size distribution. An estimate of the time
required in the DEM simulations is about 15 minutes, whereas solution of the PBM equations
for the longer grinding times required between 2 and 4 hours of computation using a Xeon
Quadcore X3370 processor.
100
100
10
0.1
0.01
Initial
1 min.
4 min.
8 min.
15 min.
0.1
Granulite -4.75 mm
10
Initial
1 min.
5 min.
10 min.
12 min.
1
0.01
10
(b)
0.1
10
Figure 4: Experimental (dots) versus simulated (lines) batch grinding results for copper ore
(left) and granulite (right)
Discussion
The simulation results showed very good agreement to the experimental data, which
indicates that the approach proposed is a very powerful tool to predict grinding results. The
PBM calculations can benefit from DEM simulations that can predict how the energy is
transferred to ore particles as a function of design and operational equipment parameters,
without demanding high computational effort as is the case of other approaches that use
DEM.
References
1.
Mishra, B.K. and Rajamani, R.K. (1990), Motion analysis in tumbling mills, KONA
Powder and Particle, No. 8, pp. 92-98.
2.
Lichter, J., Lim K., Potapov, A., Kaja, D (2009), New developments in cone crusher
performance optimization, Minerals Engineering, Vol. 22, pp. 613-617.
3.
Powell, M.S. and Morrison, R.D. (2007), The future of comminution modelling,
International Journal of Mineral Processing, October, Vol. 84, , pp. 228-239.
4.
Herbst, J.A. and Potapov, A.V. (2004), Making a discrete grain breakage model
practical for comminution equipment performance simulation, Powder Technology, Vol. 143144, pp. 144-150.
5.
Herbst, J.A (2004), A microscale look at tumbling mill scale-up using high fidelity
simulation, International Journal of Mineral Processing, December, Vol. 74, pp. S299-S306.
6.
Tavares, L. M. (2007), Breakage of single particles: quasi-static, in Handbook of
Powder Technology, Vol. 12, pp. 3-68.
7.
Tavares, L. M. and Carvalho, R.M. (2009), Modeling breakage rates of coarse
particles in ball mills, Minerals Engineering, Vol. 22, pp. 650-659.
Acknowledgements
The authors thank AMIRA, Vale, CNPq and CAPES for sponsoring the investigation. The
authors would also like to thank DEM Solutions for providing the EDEM software through its
academic program.
Dingena L. Schott
Assistant Professor, Section of Transport Engineering and Logistics,
Delft, The Netherlands
Within the Section of Transport Engineering and Logistics Schott focuses on the
field of Dry Bulk Transport and Storage, including the logistics and environmental
impact involved. Her teams current DEM work is on equipment design and
calibration and validation with the use of EDEM.
The Delft University of Technology Department of Marine and Transport
Technology focuses on the development, design, building, and operation of marine,
dredging and transport systems and their equipment. This requires the further
development of the knowledge of the dynamics and the physical processes involved
in transport, dredging and marine systems, the logistics of the systems and the
interaction between the equipment and control systems.
Abstract
This research investigates the possibilities of acquiring reliable parameters through
validation by means of experiments. In addition the focus is on validation and calibration of
simulations. Glass beads were used as test material. The advantage of glass beads is that a
spherical particle model is close to the real shape.
First, analyzing the sensitivity of the parameters used was done. Parameters analyzed are
the shear modulus, coefficient of restitution, static and rolling friction coefficients, Poissons
ratio and the particle density. Numerous simulations have been performed using different
combinations of parameters to study the influence of each parameter. Based on this
sensitivity analysis three sets of parameters are selected and used in validation simulations
using particles of different size.
Physical experiments using glass beads are used to verify the results of the validation
simulations. The angle of repose of glass beads is the bulk behavior characteristic measured
in the experiments and simulations of this study.
Nima Gharib
PhD Candidate, Neptec Rover Team (NRT),
Delft, The Netherlands
The Neptec Rover Team (NRT), which includes some of the industrys leading
technology experts, was brought together to investigate, conceptually design, and
test lunar mobility systems for the Canadian Space Agency. This highly experienced
team has been working together to develop technology for the new Lunar Exploration
Light Rover (LELR). The McGill University team focuses on the definition,
development and validation of a compliant wheel; on the effect of operating one or
more of the recommended mobility systems while in the presence of the fine,
abrasive dust on the lunar surface; and on the identification of strategies to mitigate
dust infiltration and component wear.
The Department of Mechanical Engineering at McGill University has a long
history of excellence in research and teaching. For more than a century, we have
been committed to train the next generation of innovators, industrial leaders and
academics.
Houshin Nejati (PhD candidate) 1, Nima Gharib (PhD candidate)2, Peter Radziszewski
(Associate Professor)2, Faramarz Hassani (Full professor) 1
1 Department of Mining and Materials Engineering, McGill University, 3450 University,
Montreal, QC, Canada H3A 2A7,
2 Department of Mechanical Engineering, McGill University, 3480 University St., Montreal,
Quebec, Canada, H3A 2A7
Introduction
Considering the relation between rigid bodies, Cundall and Strack (1979) proposed the
Discrete Element Method (DEM), for modeling of geomaterials such as rocks and soils,
whose micromechanical behaviour are discontinuous. In this model, soils and rocks are
represented as spherical either rigid or deformable particles. Each particle is modeled by its
trajectory along the system due to the contact forces and external forces acting on each
particle, such as gravity.
DEM models provides better insight of engineering problems in these fields; for instance, the
effect of micro-structural rock characteristic such as existing micro-cracks on the overall
physical behaviour of rock sample (Fu, 2005). Moreover, some emergent properties of
macroscopic system automatically come from discrete model such as transition from brittle
to ductile behaviour, and nonlinear mechanism in deformation (Cundall, 2005). Fracture
toughness is one of the material properties, which defines the resistance of rocks to
deformation and crack propagation when rocks already fractured, hence; fracture toughness
is one of the essential parameters for designing mining structures and tunnelling in cracked
bodies. Moreover, there is an approximate linear relationship between fracture toughness
and other physical properties of rock such as hardness index, Uniaxial Compressive
Strength, and Youngs modulus in many rocks (Whittaker et al, 1992) and (Nasseri, 2005). In
addition, previous studies show the significant relationship between rock micro-structural
characteristics on fracture toughness, initiation and propagation cracks in rocks. (Nasseri et
al, 2005).
Three basic crack propagation modes in fracture process can be seen: Mode I (tension,
opening), mode II (shear, sliding), and mode III (tearing). As shown in Figure 1, a SemiCircular Bending (SCB) specimen is a semi-disc of radius R, placed on two roller support
with 2S span. A prefabricated crack with length a, which makes 90 degree angle with
respect to horizontal, is created in centre of the disc. Herein, EDEM (a DEM based software
developed by DEM Solutions. Ltd.) was employed to model the mode I fracture toughness,
crack propagation and the Crack Tip Opening Displacement (CTOD) of the (SCB) basalt
specimen.
In order to evaluate the accuracy and validity of DEM simulations, its results compared with
expected results from fracture mechanics and continuum modeling using ANSYS. For FEM
based model following assumptions are made: 1-Linear Elastic Fracture Mechanics: LEFM
uses derived elasticity solutions to determine the stress intensity factor KI at a crack tip. 2-
Plane Strain Problem 3- Cracked tip region is meshed using eight- node quadrilateral
element (Barsoum, 1977), also KCALC command is used to calculate KI.
K Ic s t R
s t f Rt
Where t is the tensile strength of the model, R is particle radius and n is tensile strength of
contact-bond. The initial tensile bonds for this simulation determined by using the calculated
value of Kic from ANSYS and the value of Basalt tensile strength measured in Brazilian
Tensile test from above relationships.
Table 1. Material properties and DEM parameters
SCB Radius
25 mm
Modulus of elasticity
41930 MPa
Poissons ratio
0.16
Density
3000 kg/m3
243.60 MPa
0 degree
Crack width
0.5 mm
Crack length
15 mm
a/R
0.6
Particle size
0.5 mm
Number of particles
14325
Results
CTOD of EDEM simulations are very similar to the one obtained ANSYS as shown in Figure
2. Approximately a 6.1 % is differential with ANSYS outcome.
In Figure3, the comparison of CTOD for the S/R 0.5 and 0.7 are presented. One of the
advantages of DEM simulations over FEM simulation is that crack propagation (broken
bonds) automatically comes from discrete model as shown Figure 3. It is worth to mention
that the micro-cracks creating by this model match with samples that are broken with tension
experimental tests.
Figure 4 demonstrates the stress intensity distribution of basalt sample with S/R equal to 0.5
modeled in ANSYS which can imply crack propagation and eventually breakage from the tip
of the crack. Ongoing fracture toughness experiments at McGgill University will bring light to
this discrepancy.
Discussion
At first, inbuilt factory was used to generate particles and afterward a plate was defined to
compress the particles, the process was extremely lengthy and the generated bonds
between particles were very poor so that it leads to breaking many bonds even before
applying the compressive load. In order to reduce the pre-processing time, using MATLAB,
the coordinates of the particle centres was generated then imported to EDEM by an external
factory
Authors believe that the using the DEM features such as possibility to simulate roughness of
rock simply with modifying the particle size as well as possibility to model different minerals
with different physical properties will lead to more advance model to study the crack growth
trajectories.
Figure 2. Comparison of the Crack Tip Opening Displacement of DEM and FEM modeling of
Mode I fracture toughness
References
CUNDALL, PA., 2002 .Discontinuous future for numerical modeling in soil and rock. In
proceeding of third International Conference on Discrete Element Method- Geotechnical
Special publication, No. 117.
CUNDALL PA, and Strack ODL., 1979 .A discrete numerical model for granular assemblies,
Geotechnique, 29, pp. 47-65.
CUNDALL PA., 2005 .Discontinuous Future for Numerical Modeling in Soil and Rock. In
proceeding of third International Conference on Discrete Element Method- Geotechnical
Special publication, No. 117.
AYATOLLAHI, M.R., ALIHA, MRM., and HASSANI, MM., 2006 .Mixed mode brittle fracture
in PMMAan experimental study using SCB specimens. Materials Science and Engineering
A 417, pp.348356.
BARSOUM, R., 1977 .Triangular quarter-point elements as elastic and perfectly-plastic crack
tip element. International Journal for Numerical Engineering. Vol. 11, pp.85-98.
DONEZE, F., RICHEFEU. V., and MAGNIER, SA, 2009.Advances in Discrete Element
Method applied to soil, rock and concrete Mechanics. in: State of the art of geotechnical
engineering, Electronic Journal of Geotechnical Engineering, p. 44,
EDEM 2.1.1, 2009 .User Guide. DEM Solutions.
HAZZARD, JF. and YOUNG, RP., 2004Dynamic modeling of induced seismicity,
International
NASSERI, MHB, MOHANTYA, B., and Robin, PY F, 2005 .Characterization of
microstructures and fracture toughness in five granitic rocks, International Journal of Rock
Mechanics & Mining Sciences 42 , PP. 450460.
NASSERI, MHB., SCHUBNEL, A., YOUNG, RP., 2007.Coupled evolutions of fracture
toughness and elastic wave velocities at high crack density in thermally treated Westerly
granite. International Journal of Rock Mechanics & Mining Sciences 44, 601616.
SINGH,RN, and GEXIN, S.1990 .A numerical and experimental investigation for determining
fracture toughness of Welsh limestone. Mining science and technology, 10, 61-70.
WANG, Y., and MORA, P., 2008 .Modeling wing crack extension: implications for the
ingredients of Discrete Element Model. Pure appl. geophys. 165, 609620
WHITTAKER BN, SINGH RN., and SUN G., 1992. Rock fracture mechanics principal,
design and applications. Amsterdam: Elsevier.
Nima Gharib
PhD Candidate, Neptec Rover Team (NRT),
Delft, The Netherlands
The Neptec Rover Team (NRT), which includes some of the industrys leading
technology experts, was brought together to investigate, conceptually design, and
test lunar mobility systems for the Canadian Space Agency. This highly experienced
team has been working together to develop technology for the new Lunar Exploration
Light Rover (LELR). The McGill University team focuses on the definition,
development and validation of a compliant wheel; on the effect of operating one or
more of the recommended mobility systems while in the presence of the fine,
abrasive dust on the lunar surface; and on the identification of strategies to mitigate
dust infiltration and component wear.
The Department of Mechanical Engineering at McGill University has a long
history of excellence in research and teaching. For more than a century, we have
been committed to train the next generation of innovators, industrial leaders and
academics.
Introduction
Lunar mobility studies require a precise knowledge of the geotechnical properties of the
lunar soil when it comes to design adapted and efficient traction systems. Since the Apollo
missions, the remarkable progress of computers allows direct testing of new design
prototypes performances through soil-structure interaction simulations based on the discrete
element method (DEM).
Before simulating traction system displacements on the soil, the virtual soil parameters need
to be calibrated. This study presents a systematic method for the calibration of a granular
soil through four steps: (1) measure of some of the real material properties through a few
experiments, (2) determination of the design variables defining the virtual soil, (3)
construction of surrogate models for the virtual material properties as a function of the
design variables via simulated experiments, and (4) optimization of the design variables
values to fit the virtual soil properties to the real ones.
Two different experiments, the direct shear test and the angle of repose measure, are used
to determine the following material properties: cohesion, internal angle of friction, and angle
of repose. Optimum DEM parameters are computed to characterize two types of soil: silica
sand, based on experimental direct shear test and angle of
repose measures, and lunar regolith, based on data from the
literature.
Approach
To characterize the real soil that needed to be modeled, it was
decided to choose experiments that would give soil properties
relevant to our study the displacement of a traction system on
a deformable soil while being simple to model with our DEM
software to build the virtual soils response surface.
The direct shear test meets these criteria as its experimental setup consists of only three
parts and as it allows both determination of the soils cohesion c and the internal angle of
friction , which are crucial in terramechanics. Indeed, the Mohr-Coulomb law shows that the
maximum shear stress max of a soil depends on c and :
max c p tan
(1)
Figure 4. The direct shear test
with p being the pressure in the soil [1]. The maximum shear
experimental setup (image
stress is important as it determines the maximum thrust of a
from
British
Standards
wheel or a track on the soil. The setup of the direct shear test is Institution BS 1377-1:1990).
made of a three-part box filled with the studied granular material (cf. Fig. 1). A constant load
is applied on the top part so that the soil specimen is subjected to a constant pressure p.
The bottom frame of the box remains stationary while an increasing longitudinal force F
applied on the upper frame makes it glide on the bottom one. The longitudinal force F and
the displacement of the upper frame relatively to the bottom one are recorded. As the
contact between the two frames is frictionless, F equals to the shear stress over the box
cross-section area S: F = S. F ultimately reaches a threshold, which corresponds to the
shear failure of the soil on the plane between the two frames: max = Fmax/S. This
experiment is run under various pressures p by applying different normal loads on the top
part of the box, and the linear regression of the maximum shear stresses max plotted with
respect to p gives the soil cohesion c and internal angle of friction according to the MohrCoulomb law (Eq. 1).
The second experiment used for our calibration study is the angle of repose experiment. The
low cohesion and the high deformability of granular soils induce an important phenomenon
on the wheel-soil interaction. Indeed, as a wheel rolling on a granular soil tends to sink, the
resistance to motion of the soil on the wheel depends on how the soil will be moved by the
front of the wheel and recover the side faces. This avalanching process occurring in a
sloping soil can be illustrated by the angle of repose experiment. The angle of repose A of a
granular material is one of its most distinct properties: it imposes the shape of a heap of
gravel or a sand dune, for instance. However, it is not an intrinsic property of the material
and can depend on the experimental conditions. Different experimental setups can be used
to measure it, the most common being the slow lifting of a vertical straw filled with material
and initially laying on a plate, or the lifting of one side of a filled box. In our case, we chose
the straw setup, as it was used by S. Ji and H. Shen in their 2D DEM angle of repose
simulations [2]. We extended their study to 3D simulations to build the surrogate model of
the virtual soil's angle of repose.
To conclude, the three material properties used in this virtual soil calibration process are:
cohesion and internal angle of friction (determined with the direct shear test), and angle of
repose. Table 1 gives the values of these properties for a silica Barco sand (measured by N.
Kaveh-Moghaddam) and lunar regolith (from literature [3]).
0.1 to 1 kPa
0.024 kPa
30 to 50
26.6
65
30
The discrete element method software EDEM models granular soils from pre-defined particle
prototypes and particle creation rules, called factories. It computes the interactions between
the granular material and rigid geometries, built from simple polygons through EDEMs
graphical user interface or imported from CAD models. A particle prototype consists of a
sphere or a union of spheres, each one characterized by its radius r and center position.
Consequently, a granular material made of non-spherical particles (corn seeds for example)
can be modeled by a particle prototype including several spheres. The particle creation
rules, or factories, define how the particle should be created. The main parameters
characterizing a factory are: number of particles to create, time and place of creation,
particle prototype, size distribution. Each particle prototype or other geometry element used
in a simulation is associated to a material, with density , Poisson's ratio and shear
modulus G. The contact model used to compute the interaction forces between two
contacting spheres belonging to two different particles is detailed in EDEMs User Guide [4]
and is based on Hertz-Mindlin model [5], [6]. It characterizes the interactions with three
coefficients: restitution e, static friction s and rolling friction r. The
coefficient of rolling friction, specific to EDEM, models the effect of
surface roughness on non-spherical shaped particles. Indeed, as the
virtual particles are made of spheres, they can roll on each other
without friction. To avoid this artificial feature, the rolling friction
coefficient introduces an artificial torque in the contact model
opposed to this rolling motion.
A
(average
of 3 sim.)
Standard
deviation
Value of the
modified
parameter
A
(average
of 3 sim.)
Standard
deviation
r=100m
26.37
0.53
s=0.5
31.48
0.64
=4000kg/m
27.67
0.34
r=0.2
36.57
0.76
=0.4
27.95
0.28
e=0.3
27.63
0.61
G=7107Pa
27.06
0.44
Initial set:
27.47
0.52
The angle of repose simulation has been described in Section 3.2. Let us describe briefly the
direct shear test simulation. To simulate this experiment within an acceptable time, we
modeled only a slice of the box used in the experimental setup (cf. Fig. 1), setting periodic
boundaries on both sides of the slice to simulate an infinitely wide box. The slide width was
set to four times the mean particle radius to make sure that a particle had no chance to
interact with itself because of the periodic boundaries. In our simulations, the upper frame
remained stationary while a translation of constant velocity was imposed to the lower frame
(cf. Fig. 3). A plane macro-particle subject to a constant vertical force applied the desired
pressure on the soil. The horizontal component Fx of the total force of the soil on the lower
frame was recorded. Fig. 4 shows the force Fx plotted with respect to time for different
pressures. Fx reaches a plateau after approximately 0.2s, which allows us to compute the
maximum shear stress max = Fmax/S. Cohesion and internal angle of friction were then
computed with the linear regression of max with respect to the pressure p (cf. Eq. 1).
Second-order fit
Each surrogate model of the soils properties c, and A was then built as a product of two
independent functions: f(r= r0, s, r) describing the response surface for r= r0=50m, and a
dimensionless function of r: g(r). f was computed as the second-order fit of the data points
because this fitting technique is simple and adapted to curved response [7].
Results
For the silica Barco sand, the algorithm gave a satisfactory result. With equal weighting
factors and a convergence criterion of = 0.001, it converged after 81 iterations to the
solution x=(r, s, r)=(70.85m, 0.609, 0.0811). The virtual soil estimated properties were
then: c=23.9Pa, =29.5, A=27.8, whereas the real soil measured values are c=24Pa,
=26.6, A=30 (cf. Tab. 1).
On the other hand, the algorithm did not converge in the regolith case (objective properties:
c=0.3k, =40, A=65, property of the virtual soil: c=0.621k, =31.2, A=36.7 with the
solution x=(r, s, r)=(66.21m, 1.44, 0.150) ). One reason for that could be the very high
angle of repose of the regolith. As our simulations never showed an angle of repose higher
than 50 (cf. Fig 5), the response surface of the virtual soil never reaches A=65 in the
explored region of the design space. Decreasing the weighting factor of the objective
function 1 (responsible for the angle of repose fit) to one tenth of the other ones gave a
better approximation: c=0.359k, =38.7, A=32.9 with x=(r, s, r)=(172.7m, 1.21, 0.136).
Discussion
This study presents a methodology for the DEM parameter calibration of a virtual soil,
involving geotechnical experiments on the real material and their simulations with a DEM
software. The example given in this paper shows how a virtual granular material can be
calibrated to replicate the behaviour of a silica sand. However, the lunar regolith behaviour
could not be modeled accurately, probably because of a mediocre angle of repose surrogate
model combined with a lacking design space exploration.
This methodology could be greatly improved by a better surrogate model management.
Indeed, an automated exploration of the design space in potentially optimum zones during
the optimization process, which would allow an automated update of the response surfaces,
would ensure high-fidelity surrogate models. Such an automated process was not doable
with our DEM software release, but can definitely be explored in future work.
Acknowledgments
The authors are grateful to Neptec, the Canadian Space Agency, and the NSERC CRD
program for the financial support of this project, and also DEM Solutions. Ltd., Edinburgh,
Scotland, UK, for their help and their advices
References
[1] M.G. Bekker. Theory of Land Locomotion. The University of Michigan Press, 2008.
[2] S. Ji and H. Shen. Two-dimensional simulation of the angle of repose for a particle
system with electrostatic charge under lunar and earth gravity. Journal of Aerospace
Engineering, 22:10-14, 2009.
[3] G. Heiken. Lunar sourcebook : a user's guide to the moon. Cambridge University Press,
1991.
[4] EDEM 2.1.1 User Guide. DEM Solutions, 2009.
[5] R.D. Mindlin. Compliance of elastic bodies in contact. Journal of Applied Mechanics,
16:259 {268, 1949.
[6] T. Tanaka Y. Tsuji and T. Ishida. Lagrangian numerical simulation of plug ow of
cohesionless particles in a horizontal pipe. Powder Technology, 71:239{250, 1992.
[7] Raymond H. Myers and Douglas C. Montgomery. Response Surface Methodology,
second edi-tion. Wiley, 2002.
[8] Jorge Angeles. MECH 577 Optimum Design, Lecture notes. McGill University, 2008.
Klaus Mutschler
Design & Welding Engineer, Mechanical Engineering, Components R&D
Luxembourg, Luxembourg
Lionel Hausemer
Project Engineer, Mechanical Engineering, Components R&D
The Paul Wurth Group is one of the world leaders in the design and supply of the
full-range of technological solutions for the iron making industry. The Group also
provides tailor-made equipment and systems for the steel making industry and
affiliated sectors.
Abstract:
Raw material transport using conveyor transfer and chutes may become critical due to
details not covered in the design phase or the low maintenance frequency in the process.
Problems such as material spillage, belt misalignment, breakage and premature wear of
peripherals associated with the system can cause damage to the process. Currently, part of
the raw material is lost in transportation in the Ipatinga ironmaking plant (Usiminas). Aiming
to reduce this loss, a study was done on particular system of pellets transport, which
comprises two conveyor belts and a set of transfer chute, using the EDEM. Experiments
were conducted to obtain some pellet, belt and metal properties needed in the simulation.
Analyzing the simulation results for current conditions it was possible to identify problems
that have occurred in the process. Changes in the model geometry and new simulations
were made and a new geometry that solved the problem of material spillage in the system
was obtained.
Guilherme Defendi
Ferrous Technology Center
Belo Horizonte, Minas Gerais
The Ferrous Technology Center brings together laboratories and pilot plants along
with equipment devoted to iron ore application studies. Through R&D projects, the
center develops products and technical solutions with an integrated view of the
mining and steelmaking chains.
Vale explores for, produces and sells iron ore and pellets, nickel, copper, coal,
bauxite, alumina, aluminum, potassium, kaolin, manganese, ferro-alloys, cobalt,
platinum-group metals and precious metals. Vale also operates in the logistics,
energy and steelmaking sectors with the mission of transforming mineral resources
into prosperity and sustainable development.
Abstract:
For steelmaking, coke and sinter and/or lump and/or pellet must be feed in a blast furnace
so that reduction happens and after melting of metallic charge and pig iron formation. Steel
formation happens in forward stages of refine. Material distribution during blast furnace
charging is very important for a better control on blast furnace operational performance.
Segregation of different materials can happen in a non uniform way in the furnace center
and walls, leading to a higher difficult to reach desired goals. This work present a Discrete
Element Method (DEM) simulation performed in a blast furnace. Material data was firstly
calibrated and the EDEM software was used to analyze material flow during blast furnace
charging to help predict material segregation. Also voidage condition was mapped in
different regions of the furnace, in order to check how segregation will affect voidage and by
consequence gas flow. EDEM was also used to predict compressive forces to understand
how blast furnace charging operation can cause particle breakage in different regions.
Yongxiang Yang
Yuko Enqvist, Materials innovation institute (M2i), Delft, The Netherlands; Allert Adema,
Materials innovation institute (M2i), Delft, The Netherlands; Vilas Tathavadkar, Materials
innovation institute (M2i), Delft, The Netherlands; Yongxiang Yang, Delft University of
Technology, Delft, The Netherlands; Rob Boom, Materials innovation institute (M2i), Delft,
The Netherlands, and Delft University of Technology, Delft, The Netherlands.
Introduction
The cohesive zone in the blast furnace, where iron ores soften and melt, greatly increases
the resistance to the ascending gas flow since the permeability of the ore layers becomes
varied and restricted, accordingly the gas can predominantly flow through intermediate coke
layers. The cohesive zone acts as a gas distributor that has a remarkable impact on the
performance and stability of blast furnace operation. The softening and melting process in
the cohesive zone (Fig. 1) is generally distinguished into three stages, softening, exudation
and melting-down, which depend on ore properties (such as the degree of reduction and
carburisation, the porosity, the slag former composition, and the distribution of the phases)
as well as global furnace operating conditions (temperature, solid packing structure, load,
bed porosity, PCO/PCO2, and so on). Considering the complex operating conditions and a
number of reactions taking place in the blast furnace, the mathematical modeling, coupled
with physical modeling could be a powerful tool for understanding the effect of the
microscopic ore properties and the macroscopic furnace flow conditions on the formation of
the cohesive zone.
The aim of this work is to develop a comprehensive model for predicting the cohesive zone
properties, such as its shape, location, structure, permeability and mineralogical changes, in
conjunction with upper and lower zones to the cohesive zone. The model should allow
predicting the changes of status in the cohesive zone in response to the changes in the
operating conditions.
Modeling Approach
In this work the Discrete Element Method (DEM) Computational Fluid Dynamics (CFD)
coupling approach is applied to model the solid gas flow in the blast furnace cohesive
zone. The commercial DEM particle simulation software package EDEM2.3 (DEM
solutions Ltd., UK) and CFD software package ANSYS FLUENT 12.0 (Ansys Inc., USA) are
used together with EDEM CFD coupling module for FLUENT (DEM solutions Ltd., UK).
The lower half of an experimental blast furnace (EBF) [1], excluding the hearth is employed
as the basic geometry of the model (Fig. 2). The hot air is introduced to the furnace at
tuyeres, located close to the bottom of the geometry, flows upward through the descending
packed burden bed, and exits at the top of the furnace. Ore and coke burdens are
alternatively charged on the top of the bed. Ore is charged relatively close to the furnace wall
compared to coke. The burden charging conditions used in this study are listed in Table 1.
As solid burden descends, ore heats up to its melting temperature, and is disappeared, while
coke is removed when entering the raceway. The burden descent velocity is primarily
determined by the coke removal rate at the raceway as well as ore melting. In this study the
burden descent velocity is approximately set to 40 mm/s. It should be noted that the descent
velocity applied is much higher than one in the EBF (0.5 mm/s) in order to generate burden
layer distribution with acceptable CPU time. The motion of the descending solid burden is
determined by the DEM, while the flow of continuum gas is obtained by the CFD over
computational grids. The drag force, and conductive and radiative heat transfer are
calculated by DEM CFD coupling. The input parameters used in the CFD and DEM
simulations are summarised in Table 2.
The simple softening and melting model is integrated in EDEM in order to predict the
formation of the cohesive zone. The softening is defined when iron ore deforms 50%. In the
model iron ore deforms 50% at 1200C and completely melts (disappears) at 1400C as preset criteria. The deformed particle shape is described by a rubber elasticity model. The
softening and melting model as well as the burden charging and coke removal model are
implemented in EDEM simulation using the EDEM application programming interface (API),
namely Particle Body Force API and Particle Factory API.
As for advanced softening and melting models, the thermodynamic equilibrium of iron ore is
modeled by using the thermochemical software package FactSage (GTT Technology,
Germany), and the FactSage thermodynamic model is linked to FLUENT by the commercial
thermodynamic programming library ChemApp (GTT Technology, Germany) in order to
perform the thermodynamic equilibrium computation in each FLUENT cell. Furthermore the
custom EDEM FLUENT coupling module is implemented in the coupled EDEM FLUENT
model that can allow the custom burden particle properties to be transferred between EDEM
and FLUENT. The effect of burden charging conditions on burden layer distribution, the
formation of the cohesive zone, and heat transfer are investigated with high burden descent
velocity and artificial burden thermal properties. The present study does not include chemical
reactions and advanced softening and melting model.
Figure 1: Softening and melting process of iron ore pellets in the blast furnace cohesive
zone.
I. Front
II. Side
2.3
Raceway
Gas inlet
H=0.1
1.2
0.25
Unit [m]
(a)
(b)
Figure 2: The model geometry used in the DEM and CFD simulations (a) and computational
grid for CFD simulation (b) (28365 cells). For simulation with Case III, the height is
extended by 0.4 m (=8 cells).
Case II
coke
Case III
ore
coke
ore
0.03
0.03
0.02
0.0315
0.02
Density [kg/m3]
2000
1000
4000
1000
4000
1700
1700
5730
1700
5730
close to wall
close to wall
Gas inlet
Blast velocity/ total volumetric flow rate
Blast pressure
2.4 [bar]
2200 [
C]
Burden
Types
Shape
Spherical
Size distributions
Uniform
1107 [Pa]
Poissons ratio
0.25 [-]
Coefficient of restitution
0.05
Charging temperature
900 [
C]
(2 5)104
Time step
Simulation Results
The time evolution of solid burden and gas flow fields are investigated at different charging
conditions. The burden heat capacity and thermal conductivity used are 10 J/kg K and 1000
W/m K, respectively, which are artificially adjusted due to the high burden descent velocity.
The bed is initially packed with coke, having a pre-defined temperature distribution. Ore
charging commences when simulation starts. The burden temperature field is generally
stabilised around simulation time of 20 s.
Figure 3 shows the effect of charging conditions on burden layer distribution (A) and on gas
temperature field (B) at simulation time of 30 s. It can be seen that charging conditions
considerably affect the burden layer distribution and gas temperature field. Consequently the
formation of the cohesive zone is varied with charging conditions. When applying more
realistic burden properties, for case II and III, the bed porosity in ore layers is much lower
than in coke layers, thus, gas tends to flow through coke layers, which generates large
temperature variations along the height. It can be also observed that ore charging condition
slightly alters the gas temperature field (Fig. 3 (b) and (c)). The cohesive zone considerably
increases the gas flow resistance, accordingly, coke slit type flow is realized in the
simulations. It might be difficult for the current model to reproduce the temperature field (i.e.
cohesive zone) obtained in the EBF due to the unrealistic burden descent velocity as well as
burden thermal properties used in the simulations.
(A)
(c) Case III
(a) Case I
(a) Case I
(b) Case II
(b) Case II
(B)
(c) Case III
Figure 3: The effect of charging conditions on burden layer distribution (A) and on gas
temperature field [K] (B) at simulation time of 30 s.
Discussion
The present results demonstrate that the coupled DEM CFD approach can provide a good
way to model anisotropic burden descent behaviour and heat transfer between solid burden
and gas. However, it should be pointed out that the obtained results are based on unrealistic
conditions (i.e. high burden descent velocity and artificial burden thermal properties) in order
to reduce computational time, and therefore, is not able to reproduce the key phenomena
observed in the EBF. It is necessary to use actual EBF conditions once implementing
chemical reactions into the coupled EDEM FLUENT model. Due to the current
computational restriction for EDEM FLUENT coupling, it is not possible to simulate the
formation of the cohesive zone over a long physical time. The couple EDEM FLUENT
model can predict the changes of the status in the cohesive zone within a very short physical
William R. Ketterhagen
Senior Scientist, Process Modeling and Engineering Technology Group,
Pfizer Worldwide Research and Development
Groton, CT, USA
Within the Process Modeling and Engineering Technology Group, Ketterhagen
works to develop models and engineering solutions in support of drug product
development and technology transfer. Modeling efforts range from fit-for-purpose
engineering models to discrete element method (DEM) models for detailed powder
(or tablet) flow predictions. These models are applied in several areas, including
powder characterization, storage, and handling; granulation, and film coating
processes.
At Pfizer, we apply science and our global resources to improve health and wellbeing at every stage of life. We strive to set the standard for quality, safety and value
in the discovery, development and manufacturing of medicines for people and
animals. Our diversified global health care portfolio includes human and animal
biologic and small molecule medicines and vaccines, as well as nutritional products
and many of the world's best-known consumer products. For more than 150 years,
Pfizer has worked to make a difference for all who rely on us.
Gregor Toschkoff
Researcher, Research Center Pharmaceutical Engineering (RCPE
GmbH)
Graz, Austria
The Research Center Pharmaceutical Engineering (RCPE GmbH) is an
interdisciplinary research institute in the area of pharmaceutical process and product
development. A central goal of the RCPE is to transform pharmaceutical product
development and process development from empirical approaches to a rational
science-based endeavor, in accordance with ICHs Quality-by-Design framework. To
fulfill this mission as an innovation company we cooperate with various partners
from science and industry, ranging from small and medium enterprises to global
players from different sectors of the pharmaceutical industry.
Graz University of Technology pursues top teaching and research in the fields of
the engineering sciences and the technical-natural sciences.
Figure 1: Geometries of the two tablet coating machines that were used in this work. Left:
Driam Driaconti continuous coater, right: Bohle BFC5 lab-scale coater. The pictures
generated using the EDEM 2.3 particle simulation software. For details please refer to the
section Approach below.
Approach
The performance of modern coaters strongly depends on the nature of the spray zone where
particles are effectively coated and the transport between the coating zone and the zone
where particles are not seen by the spray. Beside experimental work [2,4], numerical
simulations of particle motion using the Discrete Elements Method (DEM) have become an
extremely important tool in particle technology problems and are frequently used for coater
simulations [8].
The aim of this work is to analyze and understand the effects of parameters like tablet form,
fill volume or pan rotation speed on the intra-tablet coating variability [3] in different coating
devices. To this end, Discrete Element Method (DEM) using the EDEM 2.3 particle
simulation software (DEM Solutions. Ltd., Edinburgh, Scotland, UK) is used to numerically
reproduce the tablet motion inside different coating machines, in this case the geometries of
a Driam Driaconti continuous coater(DRIAM Anlagenbau GmbH, Eriskirch, Germany) and a
Bohle BFC5 lab-scale coater (L.B. BOHLE Maschinen + Verfahren GmbH, Ennigerloh,
Germany) are used, see Fig. 1.
The special material attributes of the tablets are known from experiments. For each
geometry different tablet shapes, namely bi-convex, oval and/or round, are modeled by the
glued spheres approach available in EDEM. Further parameter variations include different
fill volumes or different rotational speeds. For each case, important process attributes (e.g.,
residence time of the tablets under the coating spray, intra-tablet coating variability, tablets
velocities pattern) are investigated
Results and Discussion
For the detailed analysis of the tablets flow inside the bed in terms of mean velocities and
granular temperatures, a MATLAB-based program processing the data exported from EDEM
was used. A main target of the MATLAB post-processing is to evaluate the particle-based
variables on a static grid. In this way, important parameters like particle velocity or rotational
velocity can be averaged over time, and detailed investigation of e.g. local velocity variations
is possible.
DRIAM Driaconti continuous coater
An important quality attribute for tablet coating is the residence time distribution of the tablets
in the spray zone. While this quantity is fastidious to extract by experimentation, it is readily
available from the DEM simulations data. Figure 1 shows the distribution for the DRIAM
continuous coater (see above) for a simulation time of 60 seconds. It can be seen that both
tablet shape and fill level have an influence on the time that a single tablet spends exposed
to the spray. From this information, an expected coating variability and in the end coating
process time can be estimated.
Figure 2: Residence Time Distribution in the spray zone after 60s of simulation, for spherical
tablets (left) and oblong tablets (right)
Every 0.5 seconds, a set of data was exported using the EDEM software export dialog.
Based on these data sets, a mean velocity of the tablets was calculated by interpolating
each particle onto a grid. The result is a normalized average velocity on a static grid,
allowing precise comparison of different process parameters. Figure 2 shows a matrix of
results, for different fill levels and different tablet shapes. For example, it showed that round
tablets develop a qualitatively different flow profile, with pronounced disorder in the upper
region.
Figure 3: Normalized time-averaged tablet velocity on the grid for round, oval and bi-convex
tablets at the different coater fill ratios for a vertical slice in the middle of the coating
apparatus.
Bohle BFC5
Another concern that is connected with a spray coating process is the mixing of tablets [7]. In
the simulation setup, the cylindrical coating drum of a BFC5 lab coater was filled with two
sorts of particles, one sort in the front and one in the back region. Figure 4 shows the relative
standard deviation of the binary mixture for round and biconvex shaped particles and two
different rorational speeds. The RSD is calculated by using bins of appropriate size including
all tablets. A high value means high separation, a low value good mixing. On the abscissa,
the number of revolutions is drawn. As can be seen, the mixing per revolution is nearly
constant, but one has to keep in mind that with a higher rotational speed, the same amount
of revolutions and therefore mixing is achieved in shorter time.
Figure 4: Decrease of the Relative Standard Deviation of binary mixture with the number of
revolutions. In both cases, 50s are simulated, therefore the red curve (10 rpm) ends earlier.
As described above for the Driaconti, a time-averaging of particle velocities was done on a
locally stationary grid for the BFC5 as well. The result is shown in Fig. 5. Although the two
coater geometries are quite different, the qualitative difference of the movement pattern
between round and biconvex tablets is the same, with the round tablets showing disorder
near the top of the tablet bed. In the Bohle BFC5, two ribbons lead to the good axial mixing
properties of the apparatus along the wall. In Fig. 5, this can be seen as circle-shaped
regions of increased velocity near the coater wall.
Figure 5: Normalized time-averaged tablet velocity on the grid for round and bi-convex
tablets at different coater rotation rates for a vertical slice in the middle of the coating
apparatus.
Conclusion
The DEM simulation has proven to be a valuable tool to gain understanding the dynamical
behavior of the tablets under the spray gun. The gathered information is essential to obtain a
satisfactory intra-tablet coating homogeneity, which in turn is necessary to minimize the
number of tablet batches that have to be rejected. The outcomes of this work aims at
demonstrating the utility of numerical simulation in the development and the design of
pharmaceutical tablet coating processes.
References
Adam, S., Suzzi, D., Radeke, C., Khinast, J.G., 2010. An integrated Quality by Design (QbD)
approach towards design space definition of a blending unit operation by Discrete Element
Method (DEM) simulation. European Journal of Pharmaceutical Sciences, In Press.
Alexander, A., Shinbrot, T., Muzzio, F.J., 2002. Scaling surface velocities in rotating
cylinders as a function of vessel radius, rotation rate, and particle size. Powder Technology
126, 174-190.
Freireich, B., Wassgren, C., 2010. Intra-particle coating variability: Analysis and Monte-Carlo
simulations, Chem. Eng. Sci. 65, 11171124.
Ho, L., Mller, R., Rmer, M., Gordon, K.C., Heinmki, J., Kleinebudde, P., Pepper, M.,
Rades, T., Shen, Y.C., Strachan, C.J., Taday, P.F., Zeitler, J.A., 2007. Analysis of sustainedrelease tablet film coats using terahertz pulsed imaging. Journal of Controlled Release 119,
253-261.
Kalbag, A., Wassgren, C., Penumetcha, S.S., Perez-Ramos, J.D., 2008. Inter-tablet coating
variability: Residence times in a horizontal pan coater. Chem. Eng. Sci. 63, 2881-2894.
Suzzi, D., Radl, S., Khinast, J.G., 2010. Local analysis of the tablet coating process: Impact
of operation conditions on film quality. Chemical Engineering Science, Volume 65, Issue 21,
Pages 5699-5715.
Tobiska, S., Kleinebudde, P., 2003. Coating uniformity and coating efficiency in a Bohle LabCoater using oval tablets. European Journal of Pharmaceutics and Biopharmaceutics 56, 39.
Ketterhagen, W. R.; am Ende, M. T. & Hancock, B. C., 2009, Process modeling in the
pharmaceutical industry using the discrete element method, Journal of Pharmaceutical
Sciences, , 98, 442-470
Massih Pasha
PhD Candidate, Institute of Particle Science and Engineering
Leeds, UK
The Institute of Particle Science & Engineering (IPSE) is one of three institutes
within the School of Process Environmental and Materials Engineering (SPEME) of
University of Leeds. The research at IPSE focuses on the engineering science of
advanced particulate systems applied to a range of sectors including, healthcare,
which includes foods and pharmaceuticals; personal and household products, which
includes polymers, biomaterials, and fine chemicals; and minerals and fuels, e.g.
nuclear. In all of these areas IPSE focuses its strength and expertise in
measurement, modeling and manufacture.
The University of Leeds is one of the UK's top research universities, with more than
61% of our research rated as 'world leading' or 'internationally excellent.' Our
academics and their cutting-edge research are in high demand throughout the world,
and we regularly share our expertise with businesses to help them grow.
Massih Pasha, Institute of Particle Science and Engineering, University of Leeds, Leeds, UK
Colin Hare, Institute of Particle Science and Engineering, University of Leeds, Leeds, UK
Ali Hassanpour, Institute of Particle Science and Engineering, University of Leeds, Leeds,
UK
Mojtaba Ghadiri, Institute of Particle Science and Engineering, University of Leeds, Leeds,
UK
Introduction
Industrial processes involving powder blending, transfer, storage, feeding, compaction and
fluidisation all require reliable powder flow [1]. There are a number of processes which deal
with small amounts of loosely compacted powders. These include filling and dosing of small
quantities of powders in capsules and dispersion for dry powder inhalers and dry particle
sizing. In other cases, the availability of powders for flowability testing is an issue. For
instance in nuclear and pharmaceutical industries, the amount of testing powder is limited
due to ionising radiation for the former and toxicity and cost of drugs for the latter [2].
There exist a number of test methods for evaluation of flow behaviour of powders such as
uniaxial compression test, shear test, raining bed and the Sevilla powder tester [3]. Most of
these test methods require a relatively large amount of powder and measuring the flow
behaviour at relatively high compaction stresses. Hassanpour and Ghadiri [2] introduced a
testing method by ball indentation which can be performed on small amounts of loosely
compacted powders. In the present paper an attempt is made to evaluate the flowability
measurement of cohesive powders using the ball indentation method. The indentation
process and unconfined compression test are simulated using the Distinct Element Method
(DEM) for particles consolidated to different stresses. The correlation between indentation
characteristics and flow behaviour of powders (unconfined yield stress) are investigated by
comparing indentation results with those of unconfined compression.
Approach
In the indentation process, different samples of powders are consolidated into a cylindrical
die at a pressure that forms weak tablets. The die must be made of low friction materials in
order to reduce the effects of wall friction. The weakly formed tablets are then indented using
a spherical indenter and the depth/load cycle is recorded from which the hardness of the
bulk powder can be calculated. During loading, the load is increased at a specified rate until
a desired maximum load is reached. Then the load is decreased to zero at the same rate.
During unloading, the elastic deformation of the sample will recover. Hardness, H, is given
as the ratio of maximum indentation load, Fmax, to projected area of the impression, A,
(Equation 1).
Fmax
A
(1)
A dh h2
(2)
where d is the diameter of the indent, and h is the depth of impression [2].
In indentation test, during formation of the local plasticity zones around the indenter, the
volume of the powder bed present in a yield condition is surrounded by an elastically
deformable region. This leads to an increase in the local yield strength (i.e. hardness) [5]. A
linear relationship between the hardness and yield stress is usually considered:
H C .Y
(3)
where H is the hardness, Y is the yield stress and the proportionality factor C is known as
constraint factor. It is important to relate hardness to yield stress, since the flow behaviour is
defined based on the yield strength. Wang et al. [4] has concluded that indentation hardness
and unconfined yield stress have a linear relationship with pre-consolidation pressure for a
number of materials. This corroborates the linear relationship between yield stress and
hardness. The constraint factor for a number of testing powders was also determined, and it
was concluded that it is independent of the pre-consolidation pressure but is material
dependant. For particle assemblies, the constraint factor would depend on single particle
properties which needs to be analysed by the Distinct Element Method (DEM) simulations
[4].
DEM simulations and analysis of the ball indentation technique were conducted using
EDEM software provided by DEM Solutions. Ltd., Edinburgh, Scotland, UK. The HertzMindlin elastic contact model alongside a linear cohesion model are used. The cohesion
force between particles is calculated as follow:
F kA
(4)
where A is the contact area and k is a cohesion energy density with units Jm-3. The value k
is chosen so that the work of cohesion of the model equates to that of JKR model. The
material properties used in the simulations are summarised in Table 1.
Property
Particles
Die (Geometry)
Diameter (mm)
1 (1,0.14)
39
Density (kg.m-3)
2500
7800
Poissons Ratio
0.25
0.3
55
182
0.2-1.0
16,000 particles are generated to form a bed height of ~ 15 mm. The indenter diameter is 13
times greater than that of the particles. Figure 1 shows the indentation simulation inside the
EDEM environment.
In order to investigate the correlation between yield stress and hardness, the unconfined
compression test is also simulated.
The speed at which the piston and indenter are moved is controlled to provide a strain rate
of 1. This shows that the processes are carried out within quasi-static regime, where the
effects of dynamics of the geometries on stresses are minimised [6].
Results
Figure 2 shows the hardness values calculated for a range of maximum indentation load for
three different cohesion levels when the powder bed is consolidated to 10 kPa.
12
=1.0
10
=0.5
Hardness (kPa)
=0.2
0
0.09
0.1
0.11
0.12
0.13
Indentation Load (N)
0.14
0.15
0.16
Figure 2: Hardness vs. indentation load for three interface energy values
It can be seen that hardness does not change significantly with indentation load. This shows
that the powder bed is not consolidated during indentation process. If the bed is
consolidated, the hardness value will increase and it will not be representative of the preconsolidation of interest.
Figure 3 shows the hardness and unconfined yield stress results obtained from the
simulations for a range of pre-consolidation pressure. The interface energy between the
particles is 1 Jm-2 and the hardness values are obtained with a maximum indentation load of
0.12 N.
Hardness (kPa)
12
10
10
8
Hardness
Unconfined
Yield Stress
12
5
10
15
Pre-consolidation Stress (kPa)
20
Figure 3: Hardness vs. unconfined yield stress for a range of pre-consolidation pressure
It is clear that there exist a correlation between the hardness and unconfined yield stress.
The constraint factor for the simulated powder with a surface energy of 1 Jm-2 is ~5.
Discussion
The dimensions of the simulated bed and indenter were sufficient to prevent further
consolidation of the bed by indentation in the range tested. These simulations of ball
indentation were coupled with simulations of unconfined compression and showed the
correlation between hardness and unconfined yield stress. For this material the constraint
factor was found to be approximately 5.
The influence of single particle properties such as surface energy, friction, shape and
stiffness on the constraint factor will be analysed in the future simulations. Development of a
relationship between single particle properties and the constraint factor will allow the yield
stress to be inferred from ball indentation experiments alone. Consequently a ball
indentation device could be developed to measure the flowability of powders in-situ, even
when only a small powder quantity is present.
References
1. Prescott J.K., and R.A. Barnum. Pharmaceutical Technology, 2000. 24(10): p. 60.
2. Hassanpour A., and M. Ghadiri, Particle & Particle Systems Characterization, 2007. 24(2):
p. 117.
3. Castellanos A., J.M. Valverde, M.A.S. Quintanilla, Kona, 2004, 22: p. 66.
Andrew Hobbs
Research Engineer
Sheffield, UK
As a Research Engineer with Astec Inc., Hobbs performs CFD and DEM analyses
for product development and optimization. Astec has been using EDEM in their
research since the pre-1.0 beta release, and it has become a vital part of their design
process.
Astec Inc., based in Chattanooga, Tennessee, is a global leader in the hot mix
asphalt and road construction industries. Astec was founded in 1972 with the vision
to apply creative thinking and state-of-the-art technology to traditionally low-tech
industries.
Introduction
Astec, Inc. is a global leader in the production of hot mix asphalt equipment. Hot mix asphalt
(HMA) is the most common road surface in the US, comprising approximately 94% of all
roads. HMA is comprised of sand and various sizes of crushed rock, called aggregate,
which is mixed together with liquid asphalt cement binder at temperatures above 180o C.
Designing equipment to produce high quality HMA presents many engineering challenges.
The physics involved HMA production are quite complex and include multiphase heat
transfer, combustion, dense and dilute particle transport, and pollutant formation to name a
few. The harsh environment in many of the internals of the equipment mean direct
observation is very difficult. Simulation methods including Discrete Element Method (DEM)
provide Astec engineers with valuable insight and help them design better, more efficient
asphalt plants. This paper will present several recent case studies.
Approach
Simulations were undertaken using EDEM 2.3 from DEM Solutions Ltd., Edinburgh,
Scotland, UK. After importing CAD geometry, established particle parameters were input
and the simulations were run. In some cases use of the API was made to add in custom
features to expedite runtimes. These custom features include the motion of slat conveyors,
recorded factory inputs, and wall conduction heat transfer. Ensight 9.2 was used for data
visualization in some of the cases.
Results
Discussion
The use of DEM simulations provides Astec engineers with a tool to both visualize material
behavior inside existing equipment and virtual test new designs before fabrication resulting
in better designs and quicker times to market. In addition the expanded capabilities provided
by the API have permitted simulations that would not have been possible without
customization.
Dingena L. Schott
Assistant Professor, Section of Transport Engineering and Logistics,
Delft, The Netherlands
Within the Section of Transport Engineering and Logistics Schott focuses on the
field of Dry Bulk Transport and Storage, including the logistics and environmental
impact involved. Her teams current DEM work is on equipment design and
calibration and validation with the use of EDEM.
The Delft University of Technology Department of Marine and Transport
Technology focuses on the development, design, building, and operation of marine,
dredging and transport systems and their equipment. This requires the further
development of the knowledge of the dynamics and the physical processes involved
in transport, dredging and marine systems, the logistics of the systems and the
interaction between the equipment and control systems.
Abstract
Cutting processes are common for many geotechnical, mining, dredging and bulk materials
handling cases. Understanding the interactive phenomena between granular materials and
cutting tools is very important for designing or evaluating cutting process. Currently few
researchers are conducting research to analyze bulk materials and cutting machine
mechanical interaction.
Simple dry sand cutting is analyzed with computational experiments in this paper to
supplement the knowledge in this field. Granular dynamics software (EDEMTM) and
Multibody dynamics software (MSC.ADAMSTM) are used to simulate sand and cutting tools
operation respectively. This computational experiment represents the complete cutting
process including initial and steady state.
Previously cutting methods were simulated by discrete element methods without proper
loading effects from blade on bulk materials. It has been overcome in this work by using
MBD software and the output is then compared with sand cutting analytical model of
Miedema (2009).
Introduction
The mixing of solids is a fundamentally important unit operation in the pharmaceutical, food
and agricultural industries, as well as many others. Controlling the mixing mechanisms is key
to achieving the desired characteristics for a final product; this is difficult to design from first
principles since, in spite of considerable research, fundamental understanding remains
incomplete.
The mixing mechanisms will depend on the mixing action of the mixer (a wide range of
possible designs) and the flow behaviour of the particles. Rotating cylinders for example are
widely used as mixers. In batch mode they usually they consist of a horizontal cylinder
rotating around the central axis [1,2,3,4,5].The motion of the granular bed is predominantly
rotation about the cylinder axis with a cascading free surface: mixing occurs predominantly
in the cross-section with some axial dispersion [6].
Matthew T. Hardin, Tony Howes, David A. Mitchell Mass transfer correlations for rotating
drum bioreactors, Journal of Biotechnology 97 (2002), pp. 89-101
A.C. Santomaso, Y.L. Ding, J.R. Lickiss, D.W. York, Investigation of the Granular Behaviour
in a Rotating Drum Operated over a Wide Range of Rotational Speed, Chemical Engineering
Research and Design 81 (2003), pp. 936-945
Powder mixing: some practical rules applied to agitated systems M.Poux, J. Bertrand 1990
The purpose of the work reported here is to evaluate the power of DEM to help understand
flow processes and explain mixing mechanisms in different mixing equipments: horizontal
rotating drum, the hoop mixer and the Turbula mixer.
Approach
The commercial three-dimensional DEM code (EDEM 2.3) has been used in this work.
The three different motions (rotating drum, hoop mixer, Turbula mixer) have been applied to
a cylindrical container, 45 mm in diameter and 80 mm in length, as shown in Fig.1. The
granular system comprises two differently coloured and initially segregated fractions of
otherwise identical monosized spherical particles (3 mm, 9000 particles, 50% fill level) and
10
M. Marigo, D.L. Cairns, M. Davies, M. Cook, A. Ingram, E.H. Stitt, Developing Mechanistic
Understanding of Granular Behaviour in Complex Moving Geometry using the Discrete
two different initial filling conditions have been considered, transverse and axial filling
patterns.
Figure 1: Representation patterns used in the simulation for the three mixers. (a) Transverse
filling (b) Axial filling.
The rate and extent of mixing, quantified using a segregation index based on contacts
between two discretely labelled but otherwise identical fractions, was shown to depend on
equipment motion, operating speed and the initial distribution of the fractions.
Results
As shown in Fig.2 the effect of rotational speed is investigated in case of rotating drum filled
for both the filling patterns and the well known characteristics of the horizontal drum
operating in rolling mode were demonstrated: excellent transverse mixing and poor axial
mixing; both improving with speed as the depth of the active layer is shown to increase:
transverse and axial loading. As expected it can be noticed that in case of axial filling the
mixing is very slow as result of only purely dispersive mechanism in axial direction. In cases
of transverse filling the rate of mixing rate is fast since for a rotating drum the radial mixing is
very effective and higher rotation speed leads enhance mixing performance.
Figure 2: Comparison of the segregation index in case of a rotating drum at different speeds.
The hoop mixer incorporates off-axis rotation, causing periodic tilting of the cylinder axis.
The angle of inclination creates a rocking effect, which forces material movement along the
longitudinal axes of the container as highlighted by the black arrows in Fig.3.
A comparison between the different types of mixers for the characteristic number of rotation
Nmix (reciprocal rate of mixing) is shown in Fig.4. Interestingly, at low speeds the hoop
mixer and simple rotating drum exhibit similar transverse mixing but increasing speed has
the opposite effect: improving transverse mixing in the drum while worsening it in the hoop.
Axial mixing in the hoop mixer, on the other hand improves with speed. The Turbula displays
a very interesting relationship with speed. At low speeds, its transverse mixing performance
is the same as the horizontal drum and hoop but decreases significantly with increasing
speed, going through a minimum at medium speed before recovering completely at high
speed.
Conclusions
The present work is an elementary comparison of the effect of axis of rotation and loading
pattern on the mixing performances for a cylindrical vessel moving according different
motions: rotating drum, hoop mixer and Turbula mixer for a drum filled with spherical
particles.
An exponential law was used to describe the mixing behaviour in terms of a characteristic
number of rotations to achieve mixing. It was observed as expected for the rotating drum
operating in rolling mode the axial mixing is purely a dispersive mechanism and the radial
mixing is dominant.
With the hoop mixer it was observed that the rocking motion causes mixing in axial direction
and that the overall mixing efficiency depends on the operating speed. The axial mixing in
case of a hoop mixer improves with the speed whereas the radial mixing slightly degrades
as the speed increases. In the case of the Turbula mixer, we observe a decrease in mixing
efficiency from 23 to 46 rpm and a subsequent increase as speed increases from 46 to 69
rpm for both axial and radial mixing. This appears to be indicative of a transition in the bed
behaviour and mixing mechanism; further experimental investigations are necessary to
properly validate the DEM model and these observations. Further experimental work has
been carried out and it will be reported by comparing DEM simulations and Positron
Emission Particle Tracking mixing experiments on similar conditions for the Turbula mixer.
Acknowledgement: MM would like to acknowledge the EU for financial support through the
Framework 6 Marie Curie Action "NEWGROWTH", contract number MEST-CT-2005020724, Johnson Matthey Plc for funding and supporting this research.
References
1
Y. L. Ding, R. N. Forster, J. P. K. Seville, D. J. Parker, Scaling relationships for
rotating drums, Chemical Engineering Science 56 (2001), pp. 3737-3750
1
Y. L. Ding, R. Forster, J. P. K. Seville, D. J. Parker Granular motion in rotating drums:
bed turnover time and slumpingrolling transition, Powder Technology 124 (2002), pp. 18-27
1
Matthew T. Hardin, Tony Howes, David A. Mitchell Mass transfer correlations for
rotating drum bioreactors, Journal of Biotechnology 97 (2002), pp. 89-101
1
A.C. Santomaso, Y.L. Ding, J.R. Lickiss, D.W. York, Investigation of the Granular
Behaviour in a Rotating Drum Operated over a Wide Range of Rotational Speed, Chemical
Engineering Research and Design 81 (2003), pp. 936-945
1
Abdel-Zaher M. Abouzeid, Douglas W. Fuerstenau, Mixingdemixing of particulate
solids in rotating drums, International Journal of Mineral Processing 95 (2010), pp. 40-46
1
1990
Powder mixing: some practical rules applied to agitated systems M.Poux, J. Bertrand
1
Carolyn Wightman, Fernando J. Muzzio, Mixing of granular material in a drum mixer
undergoing rotational and rocking motions I. Uniform particles, Powder Technology 98
(1998), pp. 113-124
1
M. Aoun-Habbache, M. Aoun, H. Berthiaux, V. Mizonov, An experimental method
and a Markov chain model to describe axial and radial mixing in a hoop mixer, Powder
Technology 128 (2002), pp. 159-167
1
N. Sommier, P. Porion, P. Evesque, B. Leclerc, P. Tchoreloff, G. Couarraze,
Magnetic resonance imaging investigation of the mixing-segregation process in a
pharmaceutical blender, International Journal of Pharmaceutics 222 (2001), pp. 243-258
1
M. Marigo, D.L. Cairns, M. Davies, M. Cook, A. Ingram, E.H. Stitt, Developing
Mechanistic Understanding of Granular Behaviour in Complex Moving Geometry using the
Discrete Element Method. Part A: Measurement and Reconstruction of Turbula Mixer
Motion using Positron Emission Particle Tracking, Computer Modeling in Engineering and
Sciences 1591 (2010), pp.1-22
Nima Gharib
PhD Candidate, Neptec Rover Team (NRT),
Delft, The Netherlands
The Neptec Rover Team (NRT), which includes some of the industrys leading
technology experts, was brought together to investigate, conceptually design, and
test lunar mobility systems for the Canadian Space Agency. This highly experienced
team has been working together to develop technology for the new Lunar Exploration
Light Rover (LELR). The McGill University team focuses on the definition,
development and validation of a compliant wheel; on the effect of operating one or
more of the recommended mobility systems while in the presence of the fine,
abrasive dust on the lunar surface; and on the identification of strategies to mitigate
dust infiltration and component wear.
The Department of Mechanical Engineering at McGill University has a long
history of excellence in research and teaching. For more than a century, we have
been committed to train the next generation of innovators, industrial leaders and
academics.
Introduction
Lunar dust is expected to be electrostatically charged due to solar UV irradiation and its
exposure to the solar wind and cosmic rays. The charged dust particles hover above the
surface of the moon and cover everything that they come into contact with. The dust
particles are so fine and also very abrasive. From mission documents of the six Apollo
missions that landed on the surface of the moon, dust related problems is categorized into
nine main groups; vision obscuration, false instrument reading, lost of foot traction, dust
coating and contamination, seal failures, clogging of mechanisms, abrasion of materials,
thermal control problems, and inhalation and irritation risk. Thereby keeping dust away from
electrical, mechanical and visual devices is a way to increase life expediency of the parts
and be able to have longer mission duration [1].
Lack of atmosphere, high temperature fluctuation and limitation on material quantity that can
be carried to the moon, restrict us to apply terrestrial approaches for sweeping dust away
from surfaces. In this work the possibility of generating traveling electro-magnetic waves by
electric curtain and using electrostatic and dielectrophoretic forces for dust removal is
investigated. Electric curtain is a device consists of parallel electrodes connected to single or
multi AC power source(s). It generates travelling electromagnetic waves so that particles
within the generated field would move based on their polarity along or against the direction
of the field [2, 4, 6]. The electro-magnetic field can acts as a contactless conveyor which
reduces the potential of damaging delicate surfaces.
Approach
Planar, circular, and tubular configurations have been selected bearing in mind the potential
application they might be used. In the case of tubular configuration both inside and outside
of the tube is studied. Each device is connected to a 3-phase AC power source with the
frequency of 50 Hz. There is a phase lag of /3 between each phase which provides
continuous moving waves that will act as an electro-magnetic conveyor.
In the first step the electric fields generated by each configuration need to be determined.
ElecNet software developed by Infolytica Corporation utilize with Finite Element Method to
determine the electric fields around each geometry. The resulting electric field is shown in
Figure 1.
In the second step the calculated electric field is divided into seven time steps and then
imported to EDEM 2.3 particle simulation software with the additional field manager
module. Material properties used during the simulations are listed in Table 1.
(a)
(b)
(c)
Fig. 1. Moving electric field generated by (a) planar (b) tubular (c) configuration while
connected
3 phase AC power source
Results
After the particle created and deposited on the surfaces the modified External Force is
added to the model and the simulations were run for another few more time steps. The API
was modified so that it reads seven data series from field manager to calculate the forces
on the particles. As shown in the Fig. 2 after the device is turned on, the particles experience
the electric field and move along or against it based on their polarity.
(a)
(b)
(c)
(d)
Fig. 2 Dust removal by electrostatic forces in (a) planar (b) tubular-outside (c) tubular-inside
(d) circular configurations
Discussion
The efficiency of this method is depends on the particle size, the activation frequency,
voltage profile, distance between electrodes, electrode diameters, and the medium the
device is working in. Therefore to obtain a smooth movement of dust particles above
mentioned variables need to be optimized.
We plan to do some experiments in very low temperature and vacuum condition to simulate
the working condition on the moon and show the potential use of electric curtain for future
space missions.
Acknowledgement
The authors would like to thank Neptec and CSA as well as NSERC CRD program for the
financial support of this project and also DEM Solutions. Ltd., Edinburgh, Scotland, UK, for
their help and their advices
References
[1] J. R. Gaier (2005) NASA, GRC. NASA/TM, Abstract #2005-213610.
[2] A. S. Biris, D. Saini et. al. (2004) IEEE, 2, 12831287.
[3] S. Masuda, et al. (1988) IEEE, 24, 217-222.
[4] M. K. Mazumder, R. Sharma, et al, (2007), Particulate Science and Technology, 25, 5-20.
[5] S. Masuda, T. Kamimura (1975) Journal of Electrostaics,1, 351-370.
[6] F. M. Moenser (1995) IEEE, Abstact # 0-7803-2503-6.
[7] DEM Solutions, Ltd. (2010), EDEM 2.3 User Guide, Copyright 2010, Edinburgh,
Scotland, UK.
John P. Morrissey
PhD Candidate, Silos and Granular Solids Research Group, Institute for
Infrastructure & Environment, School of Engineering
Edinburgh, Scotland, United Kingdom
The Silos and Granular Solids Research Group has conducted research and
consultancy in the areas of shell structures, particulate solids mechanics and bulk
handling in support of innovative engineering solutions for over 20 years. The Group
have worked on a wide spectrum of topics including computational modelling of
solids and structures, functional and structural design of silo structures, including
their codification in design standards, material characterization and experimentation,
including solids flow and silo pressures. The focus of recent research is to transform
DEM numerical technique from a largely scientific tool into a quantitative predictive
tool.
The Institute for Infrastructure & Environment (IEE), one of five research
institutes of the School of Engineering at the University of Edinburgh, is one of
Scotland's foremost centres for research in our Built and Natural Environment. The
IEEs academic and research staff and postgraduate students together form four
Research Groups covering an extensive range of topics related to the field of Civil &
Environmental Engineering.
E-mail: J.Morrissey@ed.ac.uk
AbSTRACT
The cohesive strength of a sticky industrial bulk solid is generally recognised to be
dependent on the prior consolidation stress exerted on the bulk solid. As a result of this
characteristic, the previous stress states of a bulk solid leading up to a handling scenario
need to be considered when evaluating the handling behaviour of bulk materials. Many of
the currently implemented DEM contact models that attempt to account for the adhesion that
develops within a granular material, such as the JKR model [1] or capillary force models
[2,3], fail to capture this stress history dependent behaviour and as such may not be
representative of a bulk solid in many handling and processing operations.
This paper describes the development of a new contact model in EDEM that accounts for
this stress history dependent frictional-adhesive behaviour. It is assumed that the adhesive
forces arising within the granular solid from the consolidation stress are responsible for the
handling problems related to bulk materials during production and storage, where high levels
of adhesion developing during material storage can lead to blockages near outlets during
discharge.
In this study a meso-scale approach is adopted here where the aim is to reproduce the
observed stress history dependent bulk behaviour. As a first attempt, a relatively simple bilinear spring model giving rise to plastic permanent deformation [4-10] was chosen for the
contact model. A single adhesive force parameter is defined as a function of the maximum
contact overlap for each contact which is tracked continuously throughout the simulation.
The DEM simulations were conducted using EDEM v2.3 particle simulation software, with
the contact model implemented through the use of the API feature [11,12]. Custom contact
properties were used to record the stress history for the simulation. The initial results show
that the contact model can capture the stress history dependent cohesive behaviour of bulk
materials.
REFERENCES
[1]
K.L. Johnson, K. Kendall, and A. Roberts, Surface energy and the contact of elastic
solids, Proceedings of the Royal Society of London. Series A, Mathematical and Physical
Sciences, vol. 324, 1971, p. 301313.
[2]
G. Lian and C. Thornton, A theoretical study of the liquid bridge forces between two
rigid spherical bodies, Journal of Colloid and Interface Science, vol. 161, 1993, pp. 138-147.
[3]
T. Groger, U. T z n, and D.M. Heyes, Modelling and measuring of cohesion in wet
granular materials, Powder Technology, vol. 133, 2003, pp. 203-215.
[4]
S. Luding, R. Tykhoniuk, and J. Tomas, Anisotropic Material Behavior in Dense,
Cohesive-Frictional Powders, Chemical Engineering & Technology, vol. 26, Dec. 2003, pp.
1229-1232.
[5]
S. Luding, Anisotropy in cohesive, frictional granular media, Journal of Physics:
Condensed Matter, vol. 17, Jun. 2005, p. S2623-S2640.
[6]
S. Luding, Shear flow modeling of cohesive and frictional fine powder, Powder
Technology, vol. 158, 2005, pp. 45-50.
[7]
S. Luding, K. Manetsberger, and J. Mullers, A discrete model for long time sintering,
Journal of the Mechanics and Physics of Solids, vol. 53, Feb. 2005, pp. 455-491.
[8]
S. Luding, Cohesive, frictional powders: contact models for tension, Granular
Matter, vol. 10, 2008, pp. 235-246.
[9]
J. Tomas, Fundamentals of cohesive powder consolidation and flow, Gran. Matt.,
vol. 6, 2004, pp. 75-86.
[10]
J. Tomas, Micromechanics of ultrafine particle adhesioncontact models, AIP
Conference Proceedings, American Institute of Physics, 2 Huntington Quadrangle, Suite 1
NO 1, Melville, NY, 11747-4502, USA,, 2009, p. 781.
[11]
DEM Solutions Ltd., EDEM 2.3 User Guide, Edinburgh, Scotland, UK. 2010.
[12]
DEM Solutions Ltd., EDEM 2.3 Programming Guide, Edinburgh, Scotland, UK. 2010.
Mohammadreza Ebrahimi
PhD Student, PARDEM, Institute of Infrastructure and Environment,
School of Engineering
Edinburgh, Scotland, United Kingdom
Abstract:
Pneumatic conveying is widely used in various industries for solid handling and
transportation. Generally, depending on particle properties, gas velocity and system
geometry various flow regimes may take place in pneumatic lines. In this study diverse flow
patterns in vertical and horizontal pneumatic conveying are simulated by using coupled
EDEM-FLUENT software. The fluid phase is simulated by using FLUENT to solve timeaveraged Navier-Stokes equation and solid phase is modelled as discrete elements by using
DEM software, EDEM. Two-way coupling through the full momentum exchange between gas
and solid phases is applied in simulation and Eulerian-Lagrangian method is selected to
have better insight to the particle level phenomena.
All operating conditions and particle properties have been extracted from Lim et al.(Lim,
Wang, & Yu, 2006) study to re-simulate their results. For horizontal pneumatic conveying
plug flow, stratified flow, moving dunes and homogeneous flow, and for vertical conveying
dispersed and plug flow are simulated to show the ability of commercial software to resimulate DEM-CFD code results. For vertical and horizontal pneumatic conveying effect of
gas velocity on the radial solid concentration profile is also investigated.
0.5 s
2s
8s
10 s
Fig1. Dispersed flow regime in vertical pneumatic conveying (gas velocity 24 m/s, 500
particles)
The results illustrate that coupled EDEM-FLUENT software can simulate the gas-solid phase
systems modelled by CFD-DEM code accurately and this software may open a promising
way for further development in two-phase modelling.
Reference:
Lim, E. W. C., Wang, C. H., & Yu, A. B. (2006). Discrete element simulation for pneumatic
conveying of granular material. Aiche Journal, 52(2), 496-509
Introduction
At present, the following methods that are used for dosing additives in preserved foods are
liquid dosing (additives dissolved in the canning liquid) and dry dosing (additives in
powdered or tablet-form). Dry dosing methods are more hygienic and precise than liquid
dosing, imbuing the end foodstuffs with greater quality and safety. However, dry dosing is
used less frequently than liquid dosing because of high labor costs and low productivity.
Thus, an automatic dosing device was designed to provide the use of food additives in the
solid phase, specifically in the form of tablets, avoiding the discharge of wastewater. As a
direct result, the water used in the process is free of any corrosive agent. This extends the
useful life of the canning line machinery. Also, this dosing method improve both productivity
and the safety-quality of the end product, compared with the other dry dosing methods.
This research aims to optimize the devices serializer mechanism that we have patented
(Device for supplying / packaged tablets dosing for the food industry; EP1595795). In this
mechanism, a driving force acting on a pair of blades that rotate in the opposite direction,
and determining each unit step to the tablets (Figure 1).
Primarily, we tried several geometries of the blades for the tablets (Figure 2).
Approach
The dosing system is shown below. The purpose is to find the angle and speed of blade
optimal.
Figure 3: (left) Prototype in industrial process, (middle) Catia Dossing prototype model,
(right) Test Dossing prototype
Material
Poissons
Ratio
Shear Modulus
Density
Salt
0,2500
1,e+04
2165,0000
Aluminium
0,3500
3,e+10
2700,0000
Polycarbonate
0,3700
8,e+08
1200,0000
Interaction
Coef.
Restitution
Static Friction
Rolling Friction
Salt-Salt
0,5000
0,4500
0,0500
Salt-Aluminium 0,5000
0,3000
0,0100
Salt-Polycarb.
0,5000
0,3300
0,0100
Figure 4: (left) Catia tablet model, (center and right) Several particle models created with
different radius and number of surfaces. The tablet model, as particle, is designed to use it
as template to conform with different numbers of surfaces and find a compromise between
simulation time and his approach to real model.
Figure 5: (left) Blade configuration parameters, (right) Geometry model of dossing device in
EDEM
Figure 6: Factory designed to emulate the tablet fallen into the prototypes hooper
Results
Discussion
This automatic dossing device provides advantages in the food industry: improving
productivity, cost efficiency. According to achieved results in modelling and process
simulation using EDEM, related to real behaviour, this tool is appropriate for this process.
EDEM is an efficient and fast tool to optimize device parameters, like the hoopers angle or
blades shape.
In a future work, we hope using EDEM Results data to integrate with data mining
techniques. The objective will be take the best data to feedback it into the model (frictions,
blades velocity or angle) to minimize contacts forces between the tablets, which produce
mass loss in each. Another important task is the calibration of a vibration system to avoid
tablet jams. Moreover, it will let fix a non excessive angle and must respect the tablets mass.
Acknowledgements
This work is made possible thanks to support from the Regional Research Plan of the
Autonomous Community of La Rioja (Spain) through the project FOMENTA 2010/02, and
from the University of La Rioja through the project API10/15.
References
DEM Solutions, Ltd. (2010), EDEM 2.2 User Guide, Copyright 2009, Edinburgh,
Scotland, UK.
Alba, Fernando, Ordieres, Joaquin, Vergara, Eliseo, Martnez de Pisn, Francisco Javier
and Castejn Manuel (2005), European patent, EP 1 595 795 A1, Device for
supplying/dosing packaged tablets for the food industry.
Alba, Fernando, Ordieres, Joaquin, Vergara, Eliseo, Martnez de Pisn, Francisco Javier
and Castejn, Manuel (2007), Previous test patent, ES 2 277 503, Mejoras introducidas en
la patente de invencin n P200202907 por: Suministrador-dosificador de comprimidos a
envases para la industria alimentaria.
Kovthaman Murugaratnam
DPhil Student, Discrete Element Research Group, Department of
Engineering Science
Oxford, United Kingdom
Murugaratnams research In the Discrete Element Research Group focuses on the
Shot peening optimization using DEM and is funded by the Engineering and Physical
Sciences Research Council (EPSRC), DEM Solutions, and Rolls Royce.
The Department of Engineering Science at Oxford is the only unified department in
the UK which offers accredited courses in all the major branches of engineering. A
broad view of engineering, based on a scientific approach to the fundamentals, is
part of the tradition that started with our foundation in 1908 - one hundred years of
educating great engineers, and researching at the cutting edge
Abstract
Compressive residual stresses are beneficial in enhancing the fatigue life of metal
components. Shot peening (SP) is an industrial cold working process that is applied to
induce a field of compressive residual stresses and modify the mechanical properties of the
metal component. The SP process involves impacting a surface with tiny shots with forces
sufficient to create plastic deformation. The process is governed by a number of important
parameters, such as the shot size, angle of attack, and impact velocity and mass flow rate.
But the relation among the desired peening effect, particularly the residual stress distribution
of the treated surface and the peening parameters is still unknown and need to be
investigated. Modelling the process is very complex as it involves the interaction of a metallic
surface with large number of shots of very small diameter. Shot peening parameters are
customarily chosen on the basis of either empirical laws or past practice.
The objective of this work is to develop a discrete element model that can suitably simulate
the shot peening process so that parameters may be chosen on the basis of mechanical
considerations. A discrete element model with numerous randomly distributed steel shots
bombarding a steel component at various velocities is developed as an example. With this
model, the shot peening shot-shot interaction and shot-target interaction and particularly the
surface coverage, angle of impingement, shot size, impact velocity and the overall shot flow
can be studied in detail and with limited computational effort. A new technique to dynamically
change the coefficient of restitution for repeated impacts of shots in the same spot was
implemented.
Introduction
In recent years, sand filled geotextile bags (geobags) have been used as a means of long
term riverbank revetment stabilization. However, despite their deployment in a significant
number of locations, the failure modes of such structures are not well understood. Three
interactions influence the geobag performance, i.e., geobaggeobag, geobagwater flow
and geobagwater flowriver bank. The aim of the research reported here is to develop a
detailed understanding of the failure mechanisms in a geobag revetment using a DEM
model.
Approach
To enhance the fundamental knowledge of the performance of geobags in a revetment, work
has been carried out using 1:10 scale models of geobags, in a laboratory flume. In such
circumstances three interactions influence the geobag performance, i.e., geobaggeobag,
geobagwater flow and geobagwater flowriver bank. In the following EDEM has been
applied to simulate these interactions and replicate the laboratory observations using the 3D
discrete element method (DEM).
Firstly, for geobaggeobag interaction, the frictional resistance up to the point of geobag
sliding were unknown. To evaluate this, a dry frictional resistance test was carried out on a
wooden test rig consisting of two parts, i.e., a fixed part and the mobile part. The mobile part
of the test rig can move up to 0.10 m downward, and provides a simple representation of
riverbank toe scouring. EDEM has been applied to represent this behaviour and thus a
coefficient of friction has been obtained for the geobags.
Secondly, geobagwater flow interaction was studied using 600 model geobags in a
laboratory open channel. Different failure modes were observed at different water levels
through several experiment runs. The active hydrodynamic forces were unknown for these
failure modes. So, to mimic the laboratory observation, a one way coupling of the measured
water velocity field and geobags was run using EDEM. Thus the coefficient of drag and the
lift force applied by the flowing water to the geobags were obtained from the DEM
simulations.
Finally, the geobagwater flowriver bed interaction was studied through repetition of the
previous experiment on 0.10 m sand bed underneath the model geobag revetment. In
addition to the hydrodynamic forces, toe scouring was added as an additional parameter.
The measurements of bed changes were recorded from the laboratory and used in the
EDEM model as a moving support boundary. The same oneway coupled model as
previously applied was then used here. The findings provided clarification of geobag
movement due to combined application of the hydrodynamic forces and toe scouring.
Velocity (m/s)
Mobile
Fixed
Velocity (m/s)
Results
Given variations in bag size used in the experiment, tolerance limits in their initial placement,
ignorance of the bag permeability and its state of wetness, the hypothesis is that the initial
response of any layer geobags in the DEM model would indicate the critical location for bag
instability in the revetment.
Figure 1 represents the geobaggeobag interactions; the static coefficient of friction was
obtained for the desired setup as 0.55. The coefficient of drag and lift forces were acquired
for geobagwater flow interaction, these being 0.5 and 0.8 respectively (Figure 2). The same
coefficients could reproduce the laboratory observations for the geobagwater flow
interaction on a mobile sediment bed (Figure 3).
Discussions
In this study a coefficient of friction of 0.55 was found to give the best comparisons; this is
close to the published dry geotextilesand interaction, which gives a coefficient of friction of
0.57 to 0.70 [1, 2] and to the finding by Recio and Oumeraci [3], for geobag geobag
interaction under waves, which was 0.53.
A coefficient of drag of 0.5 and coefficient of lift of 0.8 show good agreement with laboratory
observations in higher water level conditions. In the first experiment, the movement of bags
started in the bottommost layer due to void flow; the DEM model did not predict this,
although it did reproduce the bag movement in the surface level layer and the one below
this.
For more practical interactions (i.e., the geobagwater flow interaction on mobile sediment
bed), the DEM model gave good representation of revetment failure modes in all of the
selected water level conditions, and provides a useful tool for characterizing incipient
revetment failure.
References
1.Garcin P., Faure Y.H., Gourc J.P. and Purwanto E. (1995), Behaviour of Geosynthetic
Clay Liner (GCL): Laboratory Tests, Proceedings 5th International Symposium on Landfill.
Calgary, 1, pp. 347-358.
2.NAUE GmbH &Co. KG. (2006), Advantages of Needle-punched Secutex and Terrafix
Nonwoven Geotextiles, NAUE GmbH &Co. KG, Germany.
3.Recio, J. & Oumeraci, H. (2009), Processes affecting the hydraulic stability of coastal
revetments made of geotextile sand containers, Coastal Engineering 56, 260284.
Acknowledgement
This study is being carried out under the Joint Research Institute (JRI) collaboration in Civil
and Environmental Engineering. Funding for this work from Heriot Watt University through a
James Watt Scholarship and additional support from DEM Solutions Limited and NAUE
GmbH & Co, are gratefully acknowledged.
A1.Laboratory
revetment height.
A2.DEM model
FLOW
FLOW
FLOW
Outward movement of
upstream outer corner
adjacent to water surface.
B2.DEM model
Outward movement of
upstream outer corner
adjacent to water surface.
FLOW
Figure 2(A1 to B2): Visual validation of the DEM simulation against laboratory observations
A2
A1
Flow
A3
Flow
Velocity (m/s)
A4
Velocity (m/s)
A5
water level equal to 85% to 100% of the revetment height. water level up to 49% of the revetment height.
Flow
Flow
Laboratory
A1: Void flow and sliding in the bottom layer;
A2:Detail of failure mode; and
A3: End of the experiment showing ripple bed formation along with a few bags
displaced in the bottommost layer.
DEM Model
A4: Bottommost layer showing displacement;
A5: Details of bag displacement showing outward movement of the upstream outer
corner of the bag, similar to A1 and reversed from A2. By comparing with A3, DEM
can be seen to represent failure initiation.
B1
B2
Flow
B3
Flow
Velocity (m/s)
Velocity (m/s)
B5
B4
Flow
Flow
Laboratory
B1: Void flow causes uplifting in the bottom-most layer and at the same time sliding
due to overtopping is observed in the next to the surface water level layer.
Figure 3 (A1 to B5): Visual validation of the DEM simulation against laboratory observations
Introduction
This work presents how EDEM [1] can be used to improve the design of a new mixer for the
food and pharmaceutical industries. Although the existing mixing industrial prototype (figure
1) already provides adequate times and proportions suitable for further processing
(according to the tests carried out with different granulated materials), it is possible to further
improve its performance by means of numerical simulations. Due to the characteristics of the
materials to be mixed, EDEM is suitable to improve the design of the prototype by modifying,
for instance, the geometry and angle of the blades, the speed of rotation, etc. Thus, a more
efficient design can be obtained in an economical way. At this moment, the optimal
parameters of the simulations in order to achieve the real mixing process results have been
accomplished.
Approach
The first step was to get a virtual model with EDEM to reproduce the mixing process (figure
2).
Different types of simulations with different particle sizes, densities, friction coefficients, etc.
were analysed in order to find out the optimal parameters which replicate what actually
happens with the industrial mixing prototype (figures 3 and 4).
Thanks to the study conducted by Hassanpour et al [2], it can be considered that larger
particles with a density equivalent to the bulk density of powders are moving with the same
momentum than packets of fine particles. According to this, it was possible to reduce the
computing time by using larger particles, as Table 1 illustrates.
10
15
20
72 hours (real)
Results
Figure 4. Different periods of time during the mixing process (experimental and virtual with
EDEM).
35%
35%
30%
30%
25%
25%
20%
20%
15%
15%
10%
10%
5%
5%
0%
0
60
120
180
240
Prueba experimental n 1
300
360
420
Prueba experimental n 3
480
540
600
660
720
780
840
900
960
1020
1080
1140
1200
1260
1320
0%
1380 0 1440
Simulacin virtual n 1
601500
1201560
1620
180
Prueba experimental n 1
1680
240
1740
300
1800
360
420
Prueba experimental n 3
480
540
600
660
720
780
840
900
960
1020
1080
1140
1200
1260
1320
1380
1440
1500
1560
1620
1380
1440
1500
1560
1620
Simulacin virtual n 1
60%
70%
55%
65%
60%
50%
55%
45%
50%
40%
45%
35%
40%
30%
35%
25%
30%
20%
25%
20%
15%
0
60
120
180
Prueba experimental n 1
240
300
360
420
Prueba experimental n 3
480
540
600
Simulacin virtual n 1
660
720
780
840
900
960
1020
1080
1140
1200
1260
1320
1380
Prueba experimental n 3
480
540
600
Simulacin virtual n 1
660
720
780
840
900
960
1020
1080
1140
1200
1260
1320
Figure 5. Comparison between experimental method (industrial mixing prototype) and virtual
simulation (EDEM): mixing factor against mixing time at different zones in the mixing
chamber.
Discussion
The achieved results (figure 5) show that it is possible to obtain the appropriate simulation
parameters to model the behaviour of the real industrial mixer prototype [3,4]. EDEM is an
efficient and fast tool to optimize device parameters, and future prototypes.
The future research is focused on improve the performance of the industrial prototype based
on this EDEM model. The geometry and angle of the blades, the speed of rotation, the
chamber geometry, etc. will be modified to analyse their effect on the mixing factor and
mixing time.
Acknowledgements
This work is made possible thanks to support from the Regional Research Plan of the
Autonomous Community of La Rioja (Spain) through the project FOMENTA 2010/02, and
from the University of La Rioja through the project API10/15.
References
DEM Solutions, Ltd. (2010), EDEM 2.3 User Guide, Copyright 2010, Edinburgh,
Scotland, UK.
Hassanpour, Ali; Tan, Hongsing; Bayly, Andrew; Gopalkrishnan, Prasad; Ng, Boonho and
Ghadiri, Mojtaba (2010). Analysis of particle motion in a paddle mixer using Discrete
Element Method (DEM). Powder Technology, available online 20 August 2010.
Garca, Andrs (2008). Sistema Industrial para el acondicionamiento de aditivo alimentario.
Industrial Engineering Final Project. La Rioja University.
Garca, Andrs (2010). Anlisis del proceso de mezcla de prototipo industrial. Mtodo
experimental y simulacin virtual. Diploma of Advanced Studies. PhD courses in Project
Management. La Rioja University.
Stefan Rakitsch
Research Associate, Institute for Materials Handling Material Flow
Logistics, Mechanical Engineering
Munich, Germany
The Institute for Materials Handling Material Flow Logistics of the Technische
Universitt Mnchen is one of Germanys leading institutes in materials handling with
an experience of over 30 years in scientific research on screw conveyors.
The Technische Universitt Mnchen is one of the most research-focused
universities in Germany and Europe. The Faculty of Mechanical Engineering offers
degrees in mechanical engineering, Energy and Process Engineering, Product
Development and Design, Automotive and Combustion Engine Technology,
Aerospace Engineering, Mechanical Engineering and Management, Mechatronics
and Information Technology, Medical Technology and Nuclear Technology.
Stefan Rakitsch, Institute for Materials Handling Material Flow Logistics, Technische
Universitt Mnchen, Garching b. Mnchen, Germany;
W. A. Gnthner, Institute for Materials Handling Material Flow Logistics, Technische
Universitt Mnchen, Garching b. Mnchen, Germany;
Introduction
The research project aims at the analysis of inclined screw conveyors. Determining element
of the project is the forecast of the conveying character inside the Screw Conveyor and
therewith the designation of the achievable Volume Flows and the needed Drive Power as
functions of the geometry, operating and bulk material parameters. In order to achieve this
data for numerousness different parameter combinations are gained and statistically
evaluated. The project is funded by the Deutsche Forschungsgemeinschaft (DFG).
The advantages, such as the simple and robust assembly, low equipment and maintenance
costs, low susceptance to failure, and in particular the dust-proof design, often lead to the
use of screw conveyors for example in the field of bulk handling. They are used for the
vertical transport of bulk material from the hold as well as for the inclined transport on the
boom there. Other applications for inclined screw conveyors are found in the silo discharge
in cement plants. But the requirements for reliability, performance and economy but also in
terms of energy efficiency and environmental protection for conveyors for bulk materials
have risen significantly in recent years. The key parameters that need to be determined in
the sizing of screw conveyors are the achievable volume flow respectively the required
geometry and operating conditions to achieve the required flow rate and the necessary
power requirement. These targets must be determinable for the user as simple and practical,
yet safe and reliable, as possible. As there are no calculation rules existing, the project aims
in finding calculation methods for strongly inclined screw conveyors by the use of regression
analyses. A test rig is used to get the required data to develop the calculation methods. But
for e.g. geometry parameters it is not or only with considerable financial effort possible to
vary them. For this reason the decision was made to use DEM-Simulation in the project. This
paper deals with the question how a simulation model for the screw conveyors can be
prepared.
Simulation Model
Of course the screw conveyor is to be of fundamental importance in the simulation. In a first
step of abstraction, therefore, the geometry of the screw conveyor is reduced to the
necessary geometric and functional components [1]. In the case of the investigated screw
conveyor the interaction of the bulk material with the screw and the inner wall of the tube is
primarily of interest. To reduce the number of particles, therefore the function of periodic
boundaries of the simulation program is used. That means that only a short section of the
conveyor is really simulated (in this case 4 pitches). If a particle reaches the end of the
conveyor it is removed and relocated at the beginning of the conveyor with the same
characteristics (position in the cross section, speed, stresses, ...) again. In this way a quasiinfinitely long conveyor is built.
Figure 1:
The dimensions of the conveyor are chosen similar to the existing test rig for the first
simulations and validations. This allows the simulation model to be verified with data from
the test rig. The CAD models of the tube (here with intermediate bearing) and the associated
screw are shown in Figure 1. All geometry models were loaded into the simulation model
over the CAD data interface.
Finally the operating parameters of the screw conveyor, which are simulated, have to be
defined. The rotation speed n, the inclination , the screw diameter D and the filling level
are to be varied. The different rotation speeds can thereby directly be defined as dynamic
properties of the screw. To set the different inclinations of the screw conveyor the vector of
gravity is varied accordingly. This has the advantage that the rest of the simulation model
can be maintained unchanged. To set the respective filling level, first the theoretical filling
level of the conveyor with static screw is calculated. After filling the conveyor with much
more particles as needed and let the particles come to rest, the particles above the
calculated level are cut away. The simulated values of the described operating parameters
are listed in Table 1.
Table 1:
Rotating
Speed n
Inclination
Filling Level
Screw
Diameter D
[1/
s]
[]
[-]
[m]
30
0.2
45
0.4
60
0.6
0,20
0.26
0.40
Simulated Particles
The simulations are performed with PET-Pellets as bulk material. These pellets are also
used in the real test rig of the institute and are thus known in the properties and behaviour.
The particles are cylindrical with an elliptical basic shape and a volume of about 25 mm.
The CAD-Model and the model of the particle used in the simulation are shown in Figure 2.
The simulation model of the particle consists of the shell of the CAD imported template, on
those properties such as volume, weight, inertia, etc. are based, and nine spheres, which
represent the boundary of the particle in contacts.
Figure 2:
As the mass of the particle is proportional to the numerical time step, the simulation time
would not be practical if using full-scale particles. The second step in the abstraction of the
simulation is therefore to increase the particles so that the realism of the simulation is not
significantly reduced. Therefore the particle properties of the simulation model must however
be adjusted so that the behaviour of particles is still consistent on the real bulk behaviour.
For materials handling problems the inner and outer friction are of particular importance
here [2] and are therefore calibrated together with the bulk density. In preliminary simulations
a particle model with a similar geometry, whose volume is increased by a factor of 20,
carried out to acceptable computing times. To calibrate the particles tests to determine bulk
properties are reproduced in the simulation. The simulated parameters are modified
iteratively as long as the behaviour of the simulation model is equivalent to the real bulk
behaviour with sufficient accuracy. As tests the determination of the bulk density, angle of
repose and wall friction are performed. The experimental setup and the simulation models
for the calibration are based on the recommendations given in FEM 2481 [3]. The results of
these tests of the real PET-Pellets are shown in Table 2. The simulation models to realise
the calibration are shown in Figure 3. In each case, the simulations with the final results are
shown. These are also listed in Table 2.
Table 2:
Bulk Density
Angle of
Repose
Angle of Slip
PET-Pellets
Simulation
[kg/m
]
790
793
[]
35,8
35,7
[]
21,0
20,9
Figure 3:
Simulation Models for Calibration: Bulk Density, Angle of Slip, Angle of
Repose (left to right)
Data Evaluation
As targets, the average axial velocity of the bulk material vax and the torque M measured on
the screw are evaluated. The evaluation is done with the evaluation algorithms of the
simulation program. The axial velocity of the bulk material is exported as the average
velocity of all particles in the section Screw Conveyor, which represents the complete
conveyor, in axial direction per timestep. From the axial velocity of the bulk material the
coefficient of velocity can be calculated as a function of the pitch S and the rotation
speed n as shown below. The coefficient of velocity is a dimensionless coefficient, which
represents the influence of the geometrical and operational conditions on the achievable
volume flow and is used to compare the performance of different screw conveyors.
vax
S n
(1)
For the torque the axial component of the total torque of the screw per timestep is red-out. It
is the basis for calculating the coefficient of power . This is again a dimensionless
coefficient for the required power of the screw conveyor and is as a function of the filling
level , the screw diameter D, the shaft diameter d, the bulk density , the conveying
length L, the pitch S, the inclination and the coefficient of velocity .
8 M
S sin
2
D d g L
D
2
Validation of Simulation
(2)
To validate the whole simulation model the simulated conveyor is compared to results of
tests from the test rig. The screw diameter of the test rig amounts to 0.260 m, shaft diameter
to 0.076 m and the pitch to 0.230 m. For the validation the parameters shown in Table 3 are
chosen. The results of the comparison for the coefficient of velocity and coefficient of power
are also shown there. It can be seen that the simulation represents the conveyance in
inclined screw conveyors sufficiently accurate as the deviation is absolutely always smaller
than 5%.
Table 3:
Sim
Test
[]
[-]
[1/s]
[-]
30
0,6
0,2
45
0,2
60
60
0,4
7
9
5
Sim
Test
[-]
Deviat
ion
[%]
[-]
[-]
1,010
0,703
1,015
0,730
-0,5
-3,8
0,691
0,713
-3,0
0,635
0,617
3,0
3,791
13,06
1
22,41
9
6,085
3,880
12,69
1
22,83
6
6,256
Deviat
ion
[%]
-2,3
2,9
-1,8
-2,7
Summary
In the course of the project DEM-Simulation is used to simulate screw conveyors with
parameters, which are not possible to set at a test rig for example different screw diameters.
Therefore it is necessary to abstract and calibrate the real model to get a suitable DEM
model. On the one hand the geometry has to be simplified as much as possible. In this case
only the screw helix and the tube are depicted in the simulation. On the other hand the
simulated particles, which are blown up to shorten the simulation time, have to be adjusted
to get realistic results. This calibration is done by simulating three tests, which are commonly
used to get the bulk properties. In an iterating process the particle parameters in the
simulation are adjusted till the simulated properties are sufficient identical to the real
properties. After having the geometry and the particles the kind of data evaluation must be
defined. Therefore two parameters are selected which are also analysable in the test rig. In
having comparable results the last step in preparing the simulation is now to validate the
simulation model with results of the test rig. This was also done successfully so that a
validated simulation model of the screw conveyor is existing now.
Table of Symbols
Symb
ol
D
Unit
[m]
Name
Screw Diameter
Symb
ol
vax
Unit
[m/s]
IV
[m/s]
Volume Flow
[]
Name
Axial velocity of
Material
Inclination
[m]
Conveying Length
[-]
Coefficient of Velocity
[Nm]
Torque
Coefficient of Power
[m]
Pitch
[m]
Shaft diameter
[-]
[kg/m
]
[-]
Bulk Density
Filling Level
[1/s]
Rotation Speed of
Screw
References
[1]
Katterfeld, A.; Krause, F.: Funktionsanalyse eines Rohrkettenfrderers mit Hilfe der
Diskrete Elemente Methode (DEM); In: Tagungsband Fachtagung Schttgutfrdertechnik
2004, Technische Universitt Mnchen, Garching bei Mnchen, 2004.
[2]
Grger, T.; Katterfeld, A.: Kalibrierung von DEM-Simulationsmodellen fr die
Schttgutfrdertechnik; In: Tagungsband Fachtagung Schttgutfrdertechnik 2005, Ottovon-Guericke-Universitt Magdeburg, Magdeburg, 2005.
[3]
FEM 2.481:1997-07: Specific characteristics of bulk products as applicable to
pneumatic conveyors, Fdration Europenne de la Manutention.