Prediction of burden distribution and

electrical resistance in submerged arc

furnaces using discrete element method
modelling
by S.J. Baumgartner1,3, Q.G. Reynolds2,3, and G. Akdogan3

Affiliation:
1Samancor Chrome, Dikwena Chrome,
South Africa. Synopsis
2Mintek, South Africa. A computational model of a submerged arc furnace (SAF) used in the production of ferrochrome
3Stellenbosch University, South Africa. is presented. The model‘s intended use is to investigate the extent to which intrinsic and extrinsic
properties of the raw materials affect burden distribution and electrical resistance within the
furnace. The model is built on the discrete element method and calculates the mechanical
Correspondence to: interactions of particle distributions resulting from the flow motion of typical raw materials used
S.J. Baumgartner in the smelting of chromium ore. This model excludes the effects of thermodynamics, furnace
chemistry, and heat transfer. It illustrates how the consumption of materials (chromite pellets,
Email: flux, and reductant) is affected by changes in electrode length, reductant fractions, and reductant
shana.baumgartner@samancorcr.com sizing and density during the formation of a reductant bed. The resistance calculation algorithm
developed by Mintek was applied to construct networks developed from particle contacts, which
can quantitatively generate estimates of the electrode-to-electrode and electrode-to-bath electrical
Dates: conduction conditions.
Received: 18 Aug. 2023
Revised: 15 Jan. 2024 Keywords
Accepted: 22 Jan. 2024
Published: March 2024 submerged arc furnace, resistance, ferrochrome, segregation.

How to cite:
Baumgartner, S.J., Reynolds, Q.G., and
Akdogan, G. 2024
Prediction of burden distribution and
electrical resistance in submerged arc Introduction
furnaces using discrete element method
modelling. Ferrochrome is an essential source of elemental chromium for the production of stainless steel, to which
Journal of the Southern African Institute it imparts increased corrosion resistance, hardness, and toughness. South Africa is a primary producer of
of Mining and Metallurgy, vol. 124, ferrochrome and possesses 72–80% of the world’s total chromite ore reserves (Beukes, Dawson, and van Zyl,
no. 3. pp. 123–132
2010). The industry contributes approximately R42 billion in GDP per annum and employs over 200 000
people directly or indirectly (Africa Mining IQ, 2019). Recently, the ferrochrome industry has been placed
DOI ID: under severe strain due to increasing electricity tariffs and a decaying electrical infrastructure that poses a
http://dx.doi.org/10.17159/2411- threat to the sustainability of high-energy processes.
9717/3042/2024 Researchers such as Kajiwara, Jimbo, and Sakai, (2006) and Shi et al. (2016) have studied burden
distribution in blast furnaces used in the steel industry and developed prediction models that have
ORCID: been used in the understanding and optimization of blast furnace operations. However, these models
G. Akdogan failed to consider the influence of raw material properties on the charge, which would influence in the
http://orcid.org/0000-0003-1780-4075 burden distribution due to the changes in the properties of the raw materials used (Mio et al., 2020). In
addition, Mio et al. (2020) included the specifications of the equipment used to dictate burden behaviour
and developed these models using discrete element modelling (DEM). The study by Mio et al. (2020)
investigates the influence of particle segregation during the charge flow within a blast furnace. The authors
found that DEM can analyse in detail the behaviour of the charge particles, and can inform researchers
on the behaviour of the charge layers and time series changes that would be impossible to observe via
traditional experimental methods. In addition, the use of DEM in blast furnace studies clarifies control
mechanisms and the influence of design operations on the process (Mio et al., 2020). Mio et al. acknowledge
the influence of material properties on the dynamics of segregation and advise that these findings from
studies be compared against actual process phenomena.
Considering the similarities in burden flow between blast furnace and SAF operations, along with the
lack of studies specifically conducted on SAFs for ferrochrome production regarding material flow and
resulting segregation, it was decided to investigate the behaviour of the burden using DEM modelling. Mio
et al. (2020) effectively contrasted DEM simulations with real blast furnace performance and subsequently
implemented their findings in real blast furnace operations. In addition, the DEM simulations were
conducted to analyse the behaviour of the burden in a ferrochrome furnace. This study also considered the
electrical behaviour of the burden, taking into account various properties of the raw materials.

The Journal of the Southern African Institute of Mining and Metallurgy VOLUME 124 MARCH 2024 123
Prediction of burden distribution and electrical resistance in submerged arc furnaces
Model description
DEM governing equations and force models
Both the translational and rotational motion of granular particles
can be defined by Newton’s second law of motion in combination
with the force-displacement correlation at the points of contact
between particles (Wei et al., 2019; Zhao et al., 2015). An
appropriate time step of 1 x 10-4 seconds was adopted for the
simulations. The time step depends on the damping coefficient,
particle mass, and normal stiffness and is defined by Equation [1]:

[1] Figure 1—Visualization of the resistance network between conductive

particles (green spheres) and nonconductive particles (red spheres)
where ∆tc is the critical time step (s), Sn is the normal stiffness
(N/m), m represents the mass of the particle (kg), and Zn is a
dimensionless coefficient.

Software implementation
LIGGGHTS-PUBLIC® (CFDEMresearch GmbH, 2022), which is
available as an open-source computational software package, was
used as the modelling tool for the present work. Reynolds et al.
(2021) describe the model as based on the time-step of soft sphere
algorithms and it can introduce a broad range of friction and
collision force models. An advantage of using LIGGGHTS-PUBLIC®
is that the software allows flexibility in the application thereof and Figure 2—Contact relationship between two particles
the implementation of other models as subcomponents.
To perform the electrical resistance calculations based on the to the graph generated for the contact networks; however, in this
particle assemblies resulting from the DEM modelling, Python case each edge is represented by the electrical resistance of each pair
(Python Software, 2022) was used to process the electrical of particles. Thereafter, Kirchoff ’s Circuit laws are used to convert
contact network algorithm. The construction and manipulation the resistance network into a matrix equation containing the voltage
methodology was implemented using the Network X package of each particle node, which is finally solved using linear algebra
(Network X, 2022). The SciPy package (SciPy, 2022) was used to methodology.
optimize and increase the efficiency of the matrix algorithm of the When particles touch each other, they have a shared surface,
linear system, and to speed up the resistance calculation, parallel which allows for the calculation of the electrical resistance between
computing was exploited through the mpi4py Python library the pairs of particles. This is accomplished by integrating the
(mpi4py, 2021). combination throughout the entirety of the particle pair's volume,
starting at the centre of each particle and extending to the point
The use of the resistance calculation algorithm where they make contact. Figure 2 depicts the concept by showing
The methodology employed in this study, to determine the the geometric configuration of spherical particles i and j. Each
resistance of the cases , is proprietary to Mintek and is known as the particle has a radius of ri and rj, respectively, and separation is
resistance calculation algorithm (RCA) (Reynolds et al., 2021). defined as rij < ri + rj. aij represents the shared radius between the
Utilizing the RCA, the resistance between the three electrodes two particle.
and the surrounding burden is estimated. At each time step, DEM The current density vectors are assumed to be parallel to the
simulations yield the particle assemblies, which are defined as the axis where all the points are in contact between the particle pair,
coordinates of the individual particles within the array and the thus the net resistance Ωij between the particle centres is defined by
number of distinct surfaces that some of the particles make contact Equation [2] below. The normal force is implicitly accounted for by
with. the rij; however, the normal force contributing to the interparticle
The algorithm takes into account pairings of particles that are resistance is not accounted for in the equation.
in mutual contact within the given system. In the current study,
the collection of paired particles was arranged to exclude any [2]
interaction with nonconductive particles, including flux and pellets
(sintered and pelletized chromite concentrate). The remaining sets Here the electrical conductivities of the two particles, i and j, are
of particle pairings in the contact network are utilized to graphically represented by σi (S/m) and σj (S/m), respectively (Reynolds et al.,
construct a contact network by constituting nodes as edges on the 2021).
graph. The resistance network between a pair of electrodes and Two electrical burden conduction models were used to
conductive and nonconductive particles is depicted conceptually in determine the inter-electrode resistances of the SAF DEM model
Figure 1. simulation outcomes for various model situations. One model
Thereafter, the graphed data is analysed for connections and examines the direct conduction between electrodes, as depicted in
contacts between the cathode and anode surfaces, and where there Figure 3a. The other model analyses conduction from the electrode
are legitimate relationships, the data is processed to determine the through the molten phases, slag, and alloy, and then back to the
resistance network. The resistance network is structurally identical neighbouring electrode, bypassing the burden, as shown in

124 MARCH 2024 VOLUME 124 The Journal of the Southern African Institute of Mining and Metallurgy
Prediction of burden distribution and electrical resistance in submerged arc furnaces

Figure 4—Illustrations of the different consumption cases

A comparative analysis was conducted to identify the best

Figure 3—Illustration of the different inter-electrode conduction models, realistic consumption scenario that closely mimics the typical
(a) electrode-to-electrode (b) electrode-bath-electrode
behaviour found in an industrial furnace. This investigation
entailed assessing the resistances between electrode and electrode,
Figure 3b. To measure the resistance of the two models, the as well as between the electrode and bath, for all three scenarios.
electrode surfaces were designated as the cathode, while any other Furthermore, the study examined the rates of consumption and
surface was selected as the anode within the resistance calculation their corresponding patterns for these three scenarios to determine
algorithm (Reynolds et al., 2021). the conditions that are most likely to imitate the consumption
patterns of raw materials within a SAF, with a specific emphasis
Particle consumption model
on the flow of burden. The model that best matches the observed
The particle consumption region in the DEM simulation can furnace consumption behaviour was chosen based on the
be found in one of three areas in the furnace: directly below the comparative results acquired during the post-processing of the data.
electrode tip, in the intermediate zone between the charge and the The post-processing results of the DEM electrode resistances
liquid phase (consisting of slag and alloy), or in the molten slag for the three cases of the particle consumption zones are shown in
phase. Figure 4 depicts this configuration. Figure 5, where the electrode-to-electrode resistance through the
Using the information given: burden and the electrode resistance through the molten bath,are
(1) I ndicates the region where particles are consumed below the shown.
electrode tip. The cumulative duration of all simulations (the total runtime)
(2) the area within the furnace where partial slag consumption was 2000 seconds, during which the initial conditions reached
occurs, situated in the intermediate region. a pseudo-steady dynamic state (or in cases 1 and 2, a state of
(3) I llustrates the entire area of consumption, encompassing fluctuating stability). Table I provides a thorough analysis of
the entirety of the molten phase, including slag and alloy the average resistance results for every model and occurrence
phases. throughout the pseudo-steady dynamic state.

Figure 5—Resistance comparisons of the different particle consumption zones

Table I
Average resistance results for the different consumption zones
Consumption zone Electrode conduction model type
Electrode to electrode via burden average Electrode to electrode via molten bath (alloy and slag)
resistance (mΩ) average resistance (mΩ)
Below electrode Case (1) 2.06 3.06
Partial slag area Case (2) 1.67 1.72
Full slag area Case (3) 1.52 1.40

* (1)– (3) refers to Figure 4 above

The Journal of the Southern African Institute of Mining and Metallurgy VOLUME 124 MARCH 2024 125
Prediction of burden distribution and electrical resistance in submerged arc furnaces
If the model had considered the resistances related to the slag
and metal, the results would have indicated that the conduction
model between the electrodes was mostly affected by conduction
through the molten bath. In this situation, the resistance value
would have been significantly reduced in comparison to the
conduction model, as the current flows from one electrode to
another through the burden (Reynolds et al., 2021). However,
in the present study the conduction caused by the slag and metal
is ignored. The main objective of this research is to examine
the material segregation and resistance that may be completely
attributed to the burden. The expected result of the inter-electrode
conduction for both models is anticipated to be similar. This is
grounded on the consistency of the material composition across
the entire model, along with the measurements of the distance
between the electrode tip and the molten bath (1.50 m) and the
electrode pitch radius (2.25 m). The most closely related results for Figure 6—Simulation geometry of the base case model
this condition are found in the case of a complete slag consumption
zone (case 3), where the resistances from electrode to electrode via
the burden and through the molten bath are measured at 1.52 mΩ
and 1.40 mΩ, respectively.
To eliminate the inclusion of material thermodynamics and heat
transmission, it is beneficial to utilize a technique that effectively
accommodates the consumption zone, hence ensuring the validity
of the present study.

Results and discussion

Base case model
The geometry of the generic model was based on the design
parameters of a common mid-sized industrial SAF used in South
Africa, which are summarized in Table II.
The electrical conductivity of the reductant was derived from
studies by Surup et al. (2020) and Nhiwatiwa and Cromarty (2021).
Consult Table III for the particle properties and parameters, which
include the calibrated parameters.
Figure 6 is a graphical depiction of the simulation model's
geometry, illustrating the arrangement of the charge chutes and
electrode layout.
Figure 7 provides a 360 deg view of the base case furnace DEM Figure 7—Particle assembly of the base case shown at different angles (blue =
model profile. An essential element in the model is the placement of pellets; red = reductant; grey = flux)
the electrode tip, which presents challenges in the SAF environment
due to the several variables that impact the electrical performance external and internal factors, as well as complex interactions
of the electrodes and the alignment of the tip with the metal involving mechanical, electrical, and chemical components.
heel. The variables being considered encompass characteristics Yoneka et al., (1981) investigated the reduction burden of
such as burden resistance, slag composition, and tapping cycles, ferroalloys and highlighted the significance of maintaining an ideal
along with other pertinent parameters. Various endeavours have position for the electrode tip, particularly in larger furnaces. They
been made to employ data and methodologies, such as electrode state that the electrode tip has a crucial function in controlling and
hydraulic pressure, data analysis, modelling, and acoustics, to improving furnace operations. Hence, it was imperative in this
precisely ascertain the length and positioning of the electrode tip. study to determine the optimal parameters for the baseline scenario,
Nevertheless, this endeavour is additionally impeded by numerous which could then be compared to the actual furnace values of a
system that conforms to design criteria. The electrode tip position
Table II was evaluated by utilizing operational expertise and employing the
information from design specifications provided in confidential
Base case model geometric parameters
documentation. The electrode tip height, as measured from the
Parameter Symbol Value metal heel, was determined to be 1.5 m.
Furnace vessel radius Rf 6.00 m In its initial state the model volume is empty, and it is gradually
Electrode radius Re 0.75 m filled with particles introduced through the charge chutes. The
Electrode pitch circle radius Rpcd 2.25 m total number of particles introduced into the DEM simulation
Electrode tip height He 1.5 m
Maximum burden height Hf 4.8 m ranged between 180 000 and 200 000. These particles enter the
Reductant feed mass fraction xc 0.2 furnace and are then continuously generated, with the rate of
Charge chute radius Rcc 0.44 m supply being directly proportional to the rate at which the burden is
Reductant electrical conductivity sc 7500 S/m
consumed. The models were simulated for a duration of 650 seconds

126 MARCH 2024 VOLUME 124 The Journal of the Southern African Institute of Mining and Metallurgy
Prediction of burden distribution and electrical resistance in submerged arc furnaces
Table III
Base case material properties
Material type Parameter Value
Pellets Diameter, D (mm)a 10–15
Density, ρ (kg/m3)h 3948
Young's Modulus, G (GPa)d Actual 70.500
Young's Modulus, G (MPa) DEM 5.000
Poisson's ratio, Eh 0.250
Coefficient of pellet-pellet static friction, μsh 0.724
Coefficient of pellet-pellet rolling friction, μri 0.400
Coefficient of pellet-pellet restitution, eh 0.480
Coefficient of pellet-anthracite static friction, μs 0.800
Coefficient of pellet-anthracite restitution, e 0.397
Coefficient of pellet-flux static friction, μs 1.000
Coefficient of pellet-flux restitution, e 0.434
Anthracite Diameter, D (mm)a 15–30
Density, ρ (kg/m3)e 1050
Young's Modulus, G (GPa)b Actual 5.570
Young's Modulus, G (MPa) DEM 5.000
Poisson's ratio, Ee 0.220
Coefficient of anthracite-anthracite static friction, μse 0.870
Coefficient of anthracite-anthracite restitution, ee 0.390
Coefficient of anthracite-flux static friction, μs 0.400
Coefficient of anthracite-flux restitution, e 0.389
Number of spherical clumps (particle shape) 3.000
Flux Diameter, D (mm)a 20–40
Density, ρ (kg/m3)e 2604
Young's Modulus, G (GPa)c Actual 85.000
Young's Modulus, G (MPa) DEM 5.000
Poisson's ratio, Eg 0.170
Coefficient of flux-flux static friction, μse 1.000
Coefficient of flux-flux restitution, ee 0.370
Number of spherical clumps (particle shape) 6.000

aValues obtained from plant data

bAverage value calculated from Bukowska et al.,2022
c
Value extrapolated from Cong et al., 2023
dValue cited from Gustafsson., 2012
eValue cited from Wei et al., 2019
fValue cited from Moncada et al., 2022
gValue cited from Mena-Negrete et al., 2022
hValue cited from Barrios et al., 2013
iValue cited from Wei et al., 2019

until stable pseudo-steady dynamic conditions were achieved. consistent and stable electrode penetration to maintain optimum
Similarly, Reynolds et al. (2021) assumed that the electrodes exhibit metallurgical conditions for improved smelting. DEM simulations
symmetry. To make the computational model feasible, the size of were run where the electrode tip position relative to the metal heel
the region was scaled down by a factor of 5. was varied between static positions ranging between 500–2900 mm,
A summary of material properties determined and used in the while keeping all other parameters in the model constant as per the
base case model is given in Table III. base case model in Tables II and III. The comparison simulations
are provided in Figure 8.
Effect of electrode length on model behaviour A notable observation is the consistent distribution and
A pivotal control mechanism in ferrochrome SAF furnaces and separation of the burden with electrode tip positions 700 mm and
one that efficiencies depend upon is the electrode positions. In 2700 mm from the heel. While the position of the electrode may
industrial ferrochrome furnaces, the electrode position is controlled vary, the composition of the raw materials remains constant. This is
by automation and relates to the electrodes' secondary current and because the characteristics of the raw materials remain consistent
resistance. Therefore, studying and understanding the impact that regardless of the electrode tip position. Although the degree to
electrode tip position relative to the metal heel has on the burden which this regularity in the burden distribution is visually apparent
structure and the electrical behaviour is of importance to improve (Figure 8) may be limited, a more precise demonstration of this
electrode control on an industrial scale to increase alloy throughput, consistency has been conducted. This involved directly calculating
provide consistent and even power distribution, and allow for the mass fractions of the reductant and FeCr pellets at different

The Journal of the Southern African Institute of Mining and Metallurgy VOLUME 124 MARCH 2024 127
Prediction of burden distribution and electrical resistance in submerged arc furnaces

Figure 8—Model simulations of the electrode tip position at a) 700 mm b)

2700 mm relative to the metal heel

positions within the simulated areas. The resulting quantitative

data is visually presented in a two-dimensional format in Figure 9, Figure 10 — Total furnace resistance as a function of electrode tip position
providing a more precise representation of the level of consistency. (mm)
The red circles in the figure indicate the positions of the electrodes
and the feed ports. The representation consists of a composite of all
the planes through the burden . Visualizations of the particle assemblies for reductant fractions
Figure 10 demonstrates a clear trend: as the electrode tip* of –50% and +50% are shown in Figure 12. An increase in the
location increases, the resistance also increases. The connection
reductant fractions leads to the formation of a larger reductant bed,
between 1250 and 1900 mm demonstrates a variance in the trend,
which will result in better contact with the electrodes, and it reduces
as seen in Figures 10 and 11. However, it may be argued that the
the occurrence of the reductant depleted pockets.
values within this range are in the expected range of resistance
The changes in the burden structure are indicated by the
values. The anomaly is most likely caused by approximations made
resistance calculation results. With a richer reductant bed, the
in the model, leading to numerical noise. Specifically, there may
resistance is reduced as shown in Figures 13 and 14.
be a disproportionality between the compact model, designed for
Another interesting observation is the large range in which the
computing efficiency, and the effects of physical contact regions.
resistance of the furnace charge can fall when the reductant mass
Effect of change in reductant fractions on model behaviour fraction is set between 15% and 20%. This is most likely attributable
To investigate the impact of varying reductant concentrations on the to a more randomized and discontinuous stable reductant bed.
flow characteristics and electrical resistance in different scenarios,
simulations were conducted with reductant fractions ranging Effect of change in reductant sizing on model behaviour
from 10% to 40% of the total mass. All other parameters were held The effect of reductant particle size on the bed resistance is
constant, aligning with the base case model, as outlined in Tables II summarized in Figures 15 and 16. There is a notable correlation
and III. between electrical resistance and the sizing of reductant particles,
The fraction range chosen to simulate the reductant fractions with an increase in reductant size resulting in a decrease in furnace
was purely for experimentation purposes to observe and study resistance.
the behaviour of resistance of the reductant contribution in a Visualizations of the particle assembly are shown in Figure 17
comprehensive manner. for a sizing ranging from –25% to +25% of the base case reductant

Figure 9—Reductant and pellet mass fractions at (a) electrode tip position at 700 mm and (b) electrode tip position at 2700 mm

128 MARCH 2024 VOLUME 124 The Journal of the Southern African Institute of Mining and Metallurgy
Prediction of burden distribution and electrical resistance in submerged arc furnaces
size of 30 mm (base case = 1). A reduction in the reductant sizing segregation of the reductant particles towards the contours of the
results in a more even distribution of reductant particles with electrodes and the delta region between the electrodes (the area to
distinct pellet voids interrupting the continuous electrical paths the left of the visualizations) is dominant and the electrical paths
between the reductant particles, thereby imparting a high resistance formed between the reductant particles between the electrodes and
to the bed. Comparing the case with +25% larger reductant sizing, across the bed are concentrated, leading to a lower bed resistance.

Figure 11—Total resistance as a function of simulation time for different electrode tip positions

Figure 12—Model simulations of the two cases of reductant fraction Figure 13 – Total furnace resistance as a function of reductant fractions

Figure 14—Total resistance as a function of simulation time for different reduction fractions

The Journal of the Southern African Institute of Mining and Metallurgy VOLUME 124 MARCH 2024 129
Prediction of burden distribution and electrical resistance in submerged arc furnaces
The above observation is significant for industrial furnace Effect of change in reductant density on model behaviour
operations as the reductant sizing is a parameter that can be An investigation was conducted to examine the influence of density
manipulated to suit the required resistance of a reductant bed for on the segregation of particles and the resulting resistance.
optimum efficiencies. The findings indicate a significant correlation between density
and resistance, with resistance values ranging from 0.90 mΩ to
2.67 mΩ with increasing density. This suggests that changes in
reductant density have a impact on furnace resistance, as illustrated
in Figure 18.
The observed phenomenon can be ascribed to the extent of
segregation, which intensifies with increasing density. as reductant
particles of lower density tend to aggregate prominently on the
surface of the furnace simulation, as illustrated in Figure 19c.
Concurrently, it is noteworthy that particles of lower density
contribute to a more uniform dispersion of particles close to the
electrode and beneath it, as depicted in Figure 19a. In contrast,
reductant particles of higher density form a less concentrated zone,
leading to a greater degree of segregation among FeCr pellets,
as evident in Figure 19d. Specifically, higher-density reductant
particles exhibit increased concentration, featuring well-defined
voids amid FeCr pellets and flux particles, as portrayed in Figure
19b.
Figure 15 – Total resistance as a function of simulation time for different These findings align with the conclusions drawn by Gray
reductant sizings and Ancey (2015), whose investigations revealed a prevalence of
regions rich in the lowest density particles near the free surface,
with a notably limited presence of particles from other species.

Figure 17—Computational visualizations of furnace burden at different Figure 18—Comparison of resistance changes with change in reductant
reductant sizes density

Figure 16—Computational visualizations of furnace burden at different reductant sizes

130 MARCH 2024 VOLUME 124 The Journal of the Southern African Institute of Mining and Metallurgy
Prediction of burden distribution and electrical resistance in submerged arc furnaces
Chrome Ltd. The authors acknowledge the Centre for High
Performance Computing (CHPC), South Africa, for providing
computational resources.
The authors would like to thank the following institutions,
companies, and individuals for their contributions and support:
➤ Samancor Chrome for providing the authors with operational
data
➤ Stefan Swanepoel and Markus Erwee from Samancor Chrome
for their technical guidance and support.

References
Adema, A., Yang, Y., and Boom, R. 2009. Coupled DEM-CFD modeling of the
ironmaking blast furnace. Proceedings of the Seventh International Conference
on CFD in the Minerals and Process Industries, CSIRO, Melbourne, Australia,
9-11 December 2009. 6 pp. https://www.cfd.com.au/cfd_conf09/PDFs/138ADE.
pdf
Africa Mining IQ. 2019. Chrome mining in South Africa. https://www.projectsiq.
co.za/chrome-mining-in-south-africa.htm [accessed 22 February 2020],
Figure 19—Computational visualizations of furnace burden at a low and high Amberger, S., Freidl, M., Goniva, C., Pirker, S., and Kloss, C. 2012.
reductant density Approximation of objects by spheres for multisphere simulations in DEM.
Proceedings of the European Congress on Computational Methods in Applied
Sciences and Engineering, Vienna, 10-14 September. Vienna University of
A discernible trend emerges towards the base, wherein the Technology.
concentration of the lightest species decreases, while that of the Barrios, G.K.P., De Carvalho, R.M., Kwade, A., and Tavares, L.M. 2013. Contact
highest-density particles undergoes a corresponding increase – a parameter estimation for DEM simulation of iron ore pellet handling. Powder
correlation congruent with the empirical observations made in the Technology, no. 248. pp. 84–93. https://doi.org/10.1016/j.powtec.2013.01.063
present study. Baumgartner, S.J., Reynolds, Q.G., Bergmann, C., Akdogan, G., and Coetzee,
C.J. 2022. Toward discrete element modelling of material flow in submerged arc
furnaces. Proceedings of APCOM 2021. pp. 205–214.
Conclusion
BenSaïda, A. 2023. Shapiro-Wilk and Shapiro-Francia normality tests.
This study aimed to develop models for understanding solid MATLAB Central File Exchange. https://it.mathworks.com/matlabcentral/
particle flow behaviour in a submerged arc furnace, with a focus on fileexchange/13964-shapiro-wilk-and-shapiro-francia-normality-
tests?tab=reviews [accessed 10 November 2023].
establishing relationships between material properties and electrical
Beukes, J.P., Dawson, N.F., and van Zyl, P.G. 2010. Theoretical and practical
resistance. The discrete element method (DEM) was employed aspects of Cr(VI) in the South African ferrochrome industry. Journal of the
for investigating furnace burden properties, with accurate results Southern African Institute of Mining and Metallurgy, vol. 110, no. 12.
dependent on appropriate parameter values. Calibration involved pp. 743–750.
direct measurements, cold flow experiments, and bulk calibration Blender Software. 2023. OpenCV. https://www.blender.org/ [accessed 15 August
methodology, using the angle of repose as a primary foundation. 2023].
Following the establishment of key parameters, a scaled-down Bukowska, M., Kasza, P., Moska, R., and Jureczka, J. 2022. The Young’s Modulus
and Poisson’s Ratio of hard coals in laboratory tests. Energies, vol. 15. p. 2477.
DEM model was developed to simulate a submerged arc furnace https://doi.org/10.3390/en15072477
(SAF) used for smelting ferrochrome (FeCr) ores. This model acts CFDEMresearch GmbH. 2022. LIGGGHTS open source discrete element particle
as a foundational reference for conducting parametric studies. The simulation code. https://www.cfdem.com/liggghts-open-source-discrete-
study revealed a complex correlation between electrical properties element-method-particle-simulation-code [accessed 30 January 2023].
and raw material attributes, establishing a clear relationship Chakrabarty, A., Basu, S., Nag, S., Ghosh, U., and Patra, M. 2021. Model
between reductant size and furnace resistance. A 27% increase in study of centre coke charging in blast furnace through DEM simulations. ISIJ
International, vol. 61. pp. 782–791. https://doi.org/10.2355/isijinternational.
resistance was observed with an increase in sizing, aligning with ISIJINT-2020-498
prior research findings. Altering the reductant charge fraction led Coaplen, J., Stronge, W.J., and Ravani, B. 2004. Work equivalent composite
to a 76% change in resistance. The study also explored the impact of coefficient of restitution. International Journal of Impact Engineering, vol. 30,
particle morphologies on segregation and resistance, aligning with no. 6. pp. 581–591. https://doi.org/10.1016/j.ijimpeng.2003.10.038
prior observations. Increased reductant density resulted in a 66% Coetzee, C.J. 2016. Calibration of the discrete element method. Powder Technology,
resistance increase. no. 297. pp. 50–70. https://doi.org/10.1016/j.powtec.2016.04.003
Cong, R., Yang, R., Li, G., Huang, Z., Gong, Y., Jing, M., and Lu, M. 2023.
Geomechanical properties of thinly interbedded rocks based on micro- and
Future work macro-scale measurements. Rock Mechanics and Rock Engineering, vol. 56.
The current research will be enhanced by investigating several pp. 1–19. https://doi.org/10.1007/s00603-023-03360-w
factors. Firstly, the effects of intrinsic material properties, such as Dhainaut, M. 2004. Simulation of the electric field in a submerged arc furnace.
Young's modulus, the coefficient of friction, and the restitution Proceedings of INFACON X, Cape Town, South Africa, 1-4 February 2004.
pp. 605–613. https://www.pyrometallurgy.co.za/InfaconX/079.pdf
coefficient, will be examined. Secondly, the study will explore the
Dijs, H.M. 1979. A laboratory investigation of reducing agents for use in the electric
impact of furnace geometry. This includes analysing the pitch circle smelting industry. National Institute for Metallurgy, Randburg, South Africa.
diameter, electrode diameter, and furnace diameter. These aspects idem, P.A. 2008. Electrical resistivity of coke beds. Doctoral thesis, Norwegian
will be evaluated for their influence on particle segregation in SAFs University ofScience and Technology. http://ntnu.diva-portal.org/smash/get/
and their subsequent impact on electrical resistance relationships. diva2:132857/FULLTEXT01
Eidem, P.A., Tangstad, M., and Bakken, J.A. 2007. Measurement of material
Acknowledgements resistivity and contact resistance of metallurgical coke. Proceedings of INFACON
XI: Innovations in the Ferroalloys Industry, New Delhi, India, 18-21 February
This paper is published by permission of Mintek and Samancor 2007. Indian Ferro Alloy Producers’ Association. pp. 561–571.

The Journal of the Southern African Institute of Mining and Metallurgy VOLUME 124 MARCH 2024 131
Prediction of burden distribution and electrical resistance in submerged arc furnaces
Eidem, P.A., Tangstad, M., and Bakken, J.A. 2008. Determination of electrical OpenCV Python. 2023. OpenCV. http://opencv-python · PyPI [accessed 15 August
resistivity of dry coke beds. Metallurgical and Materials Transactions B, vol. 39. 2023]
pp. 7–15. https://doi.org/10.1007/s11663-007-9105-7 PyMeshlab. 2023. OpenCV. https://www.meshlab.net/ [accessed 15 August 2023]
Eidem, P.A., Runde, M., Tangstad, M., Bakken, J.A., Zhou, Z.Y., and Yu, Python Software Foundation. 2022. Python. https://www.python.org/ [accessed
A.B. 2009. Effect of contact resistance on bulk resistivity of dry coke beds. 15 January 2023].
Metallurgical and Materials Transactions B, vol. 40. pp. 388–396. doi: 10.1007/
s11663-009-9235-1 Reynolds, Q., Baumgartner, S., Erwee, M., Swanepoel, S., Geldehuys, I.,
and Akdogan, G. 2021. Computational modelling of electrical conduction
Eidem, P.A., Tangstad, M., Bakken, J.A., and Ishak, R. 2010. Influence of coke in ferroalloy furnace burdens. INFACON XVI: Proceedings of the 16th
particle size on the electrical resistivity of coke beds. INFACON XII: Proceedings International Ferro-Alloys Congres, Trondheim, Norway, 27-29 September 2021.
of the Twelfth International Ferroalloys Congress, Helsinki, Finland, 6-9 June http://dx.doi.org/10.2139/ssrn.3926638
2010. Outotek Oyj. pp. 349–358.
Reynolds, Q.G. and Bisaka, K. 2013. Computational modelling of submerged-
Gray, J. and Ancey, C. 2015. Particle-size and -density segregation in granular free- arc furnace burden behaviour. Infacon XIII: Proceedings of the Thirteenth
surface flows. Journal of Fluid Mechanics, no. 779. pp. 622–668. doi: 10.1017/ International Ferroalloys Congress, Almaty, Kazakhstan, 9-12 June 2013. https://
jfm.2015.438 www.pyrometallurgy.co.za/InfaconXIII/0821-Reynolds.pdf
Gustafsson, G. 2012. Mechanical characterization and modelling of iron ore Ringdalen, E., Rocha, M., Ramos Neto, J., and Malvik, T. 2013. The high
pellets. PhD dissertation, Luleå Tekniska Universitet. https://urn.kb.se/ carbon ferrochromium process, reduction mechanisms. SINTEF Materials and
resolve?urn=urn:nbn:se:ltu:diva-17284 Chemistry, Trondheim, Norway.
Al-Hashemi, H.M.B. and Al-Amoudi, O.S.B. 2018. A review on the angle of repose Sakia, R. 1992. The box-cox transformation technique: A review. The Statistician,
of granular materials. Powder Technology, vol. 330. pp. 397–417. https://doi. vol. 41. doi: 10.2307/2348250
org/10.1016/j.powtec.2018.02.003 Scikit Image. 2023. Scikit: Image processing in Python. http://scikit-image [accessed
Hockaday, S.A.C. and Bisaka, K. 2010. Some aspects of the production of 15 August 2023].
ferrochrome in pilot AC arc furnaces at Mintek. INFACON XII: Proceedings of SciPy. 2022. SciPy. https://www.scipy.org/ [accessed 15 January 2023].
the Twelfth International Ferroalloys Congress, Helsinki, Finland, 6-9 June 2010.
Outotek Oyj. pp. 367–376. Shi, P.Y., Zhou, P., Fu, D., and Zhou, C.Q. 2016. Mathematical model for burden
distribution in blast furnace. Ironmaking & Steelmaking, vol. 43. pp. 74–81. doi:
Jingchun, L. 2001. Discussion of ideal smelting model of submerged-arc furnace. 10.1016/j.apm.2015.02.054
Infacon IX: Proceedings of the Ninth International Ferroalloys Congress, Quebec
City, Canada, 3-6 June 2001. The Ferroalloys Association, Wahington, DC. Sithole, N.A., Rambuda, N., Steenkamp, J.D., Hayman, D.A., and Hockaday,
pp. 121–130. C. 2018. Silicomanganese production at Mogale Alloys. Journal of the Southern
African Institute of Mining and Metallurgy, vol. 118, no. 11.
Kajiwara, T., Jimbo, T., and Sakai, T. 2006. Development of a simulation model for pp. 1205–1216. http://www.scielo.org.za/scielo.php?script=sci_
burden distribution at blast furnace top. Transactions ISIJ, vol. 23. arttext&pid=S2225-62532018001100013&lng=en&nrm=iso
pp. 1045–1052.
Surup, G.R., Pedersen, T.A., Chaldien, A., Beukes, J.P., and Tangstad, M.
Kapure, G., Kari, C., Mohan Rao, M., and Raju, K.S. 2007. Use of chemical 2020. Electrical resistivity of carbonaceous bed material at high temperature.
energy in submerged arc furnace to produce ferrochrome: prospects and Processes, vol. 8. 933. https://doi.org/10.3390/pr8080933
limitations. Proceedings of INFACON XI: Innovations in the Ferroalloys
Industry, New Delhi, India, 18-21 February 2007. Indian Ferro Alloy Producers’ Tripathi, A., Kumar, V., Agarwal, A., Tripathi, A., Basu, S., Chakrabarty, A.,
Association. pp. 165–170. and Nag, S. 2021. Quantitative DEM simulation of pellet and sinter particles
using rolling friction estimated from image analysis. Powder Technology, vol.
Krogerus, H., Lintumaa, T., and Jokinen, P. 2006. Laboratory investigations of 380. pp. 288–302. doi: 10.1016/j.powtec.2020.11.024
the electrical resistivity of cokes and smelting charge for optimizing operation
Wei, H., Tang, X., Ge, Y., Li, M., Saxen, H., and Yu, Y. 2019. Numerical and
in large ferrochrome furnaces. Proceesings of Southern African Pyrometallurgy
experimental studies of the effect of iron ore particle shape on repose angle and
2006. Jones, R.T. (ed.). South African Institute of Mining and Metallurgy,
porosity of a heap. Powder Technology, vol 353. pp. 526–534.
Johannesburg. pp. 309–328.
Wensrich, C.M. and Katterfeld, A. 2012. Rolling friction as a technique for
Mena-Negrete, J., Valdiviezo-Mijangos, O.C., Nicolás-López, R., and
modelling particle shape in DEM. Powder Technology, vol. 217. pp. 409–417.
Coconi-Morales, E. 2022. Characterization of elastic moduli with anisotropic
https://doi.org/10.1016/j.powtec.2011.10.057.
rock physics templates considering mineralogy, fluid, porosity, and pore-
structure: A case study in Volve field, North Sea. Journal of Applied Geophysics, Wójcik, A., Klapa, P., Mitka, B., and Sladek, J. 2018. The use of the
vol. 206. 104815. doi: 10.22201/igeof.00167169p.2021.60.4.2118 photogrammetric method for measurement of the repose angle of granular
materials. Measurement, vol. 115. pp. 19–26.
Mitra, T. and Saxén, H. 2015. Discrete element simulation of charging and mixed
layer formation in the ironmaking blast furnace. Computational Particle Wolf, P.R., DeWitt, B.A., and Wilkinson, B.E. 2014. Elements of Photogrammetry
Mechanics, vol. 3. https://doi.org/10.1007/s40571-015-0084-1 with Applications in GIS. 4th edn. McGraw-Hill Education, New York. https://
www.accessengineeringlibrary.com/content/book/9780071761123.
Mio, H. and Narita, Y. 2018. Simulation of particle behavior in charging process
of blast furnace by discrete element method. Nippon Steel & Sumitomo Metal Yamagishi, K., Endo, K., and Saga, J. 1974. A comprehensive analysis of the
Technical Report no. 120. December 2018. https://www.nipponsteel.com/en/ furnace interior for high-carbon ferrochromium. Infacon I: Proceedings of the
tech/report/nssmc/pdf/120-11.pdf First International Congress on Ferro-Alloys, Johannesburg, South Africa, 22-26
April 1974. https://www.pyro.co.za/InfaconI/143-Yamagishi.pdf
Mio, H., Narita, Y., Nakano, K., and Nomura, S. 2020. Validation of the burden
distribution of the 1/3-scale of a blast furnace simulated by the discrete element Yoneka, S., Harada, K., Kojima, K., and Nakagawa, K. 1981. Considerations of
method. Processes, vol. 8, no. 1. https://doi.org/10.3390/pr8010006 electric furnace dimension& dynamic operation based on research. Journal
francais de l'électrothermie, vol. 1. pp. 2 9–34.
Moncada, M., Betancourt, F., Rodriguez, C., and Toledo, P. 2022. Effect of
particle shape on parameter calibration for a discrete element model for mining Yu, Y. and Saxén, H. 2012. Effect of DEM parameters on the simulated inter-particle
applications. Minerals, vol, 13, no. 1. 40. https://doi.org/10.3390/min13010040 percolation of pellets into coke during burden descent in the blast furnace. ISIJ
International, vol. 52. pp. 788–796.
mpi4py. 2022. Python bindings for MPI. https://pypi.org/project/mpi4py/ [accessed
15 January 2023]. Yu, Y. 2013. The experimental and discrete element simulation studies of bell-less
charging system on blast furnace. PhD thesis, Abo Akademi University,
Ndiaye, B.C., Gao, Z., Fall, M., and Zhang, Y. 2021. Effect of the rolling friction Finland.
on the heap formation of dry and wet coarse discs. Applied Sciences, vol. 11.
6043. https://doi.org/10.3390/app11136043 Zhao, T., Dai, F., Xu, N.W., Liu, Y., and Xu, Y. 2015. A composite particle model for
non-spherical particles in DEM simulations. Granuar Matter, vol. 17.
NetworkX. 2022. NetworkX: Network analysis in Python. https://networkx.org/ pp. 763–774.
[accessed 15 January 2023].
Zheng, C., Yao, Q., Li, X., Hudson-Edwards, K., Shan, C., Shang, X., and Li, Y.
Nhiwatiwa, J. and Cromarty, R. 2022. Evaluation of Coke Resistivity for the 2022. Experimental investigation of mechanical characteristics of coal samples
Manganese Alloy Market. Materials, vol. 15. 2897. https://doi.org/10.3390/ at different drying temperatures. Drying Technology, vol. 40, no. 16. https://
ma15082897 www.tandfonline.com/doi/abs/10.1080/07373937.2022.2060999Zhou, Z.Y., Yu,
Olsen, S.E., Tangstad, M., and Lindstad, T. 2007. Production of Manganese A.B., Ng, C., Standish, N., and Xie, Z. 2006. Determination of contact forces in
Ferroalloys. Tapir Academic Press, Trondheim, Norway. ball mills. Powder Technology, vol. 167. pp. 69–78. u

132 MARCH 2024 VOLUME 124 The Journal of the Southern African Institute of Mining and Metallurgy