3970 XXXI International Mineral Processing Congress 2024 Proceedings/Washington, DC/Sep 29–Oct 3
During the pilot testing of CAHM, additional HPGR
tests were conducted at Corem’s facility using their Weir
pilot HPGR equipment. A DEM simulation of the Corem
HPGR equipment was also carried out. Comparison of
these results with CAHM pilot test results is discussed in
the subsequent sections.
MONOROLL DEM MODELING
MonoRoll design features are based on over 170 full and
partial simulations using the Ab-T10 breakage algorithm
in the Ansys Rocky DEM software. Combinations of liner
geometry, roll diameter, roll mass, and operating condi-
tions have been evaluated and compared based on model-
predicted power draw, mill stability, and basic breakage
behavior.
Given the computational intensity of these simulations,
even with state-of-the-art graphic processing units (GPUs),
the project required a practical approach to manage simula-
tion times. Simulations were conducted for both the entire
mill domain (full mill), and discrete portions of mill length
(mill slice). For mill slice simulations, the hammer roll and
shell were cut into equal-length (100 mm) slices to reduce
the total number of particles in the simulation domain.
Based on full mill simulation results and particle trans-
port features, particles pass under the comminution zone
between the hammer roll and the shell eight times on aver-
age. Thus, a total of eight slice simulations were conducted,
feeding each mill slice with the product size distribution
from the previous one as illustrated in Figure 4.
DEM mill slice simulation results for feed top size
(F100) of 12,500 µm, F80 of 9,000 µm, and feed rates of
1.5 t/h and 1.8 t/h are presented in Table 1. The minimum
fragment size for these simulations was set in the Ansys
Rocky DEM software as 300 µm. This artificial constraint
limits the smallest particle that can break to approximately
600 µm based on the particle splitting approximately in
half plus generating fragments finer than the preset 300 µm
limit.
The choice of a minimum fragment size of 300 µm is
a deliberate one, representing a balance between computa-
tional efficiency and the ability to reasonably predict prod-
uct particle size distribution. It is fine enough to provide a
reasonable prediction of the product particle size distribu-
tion while also maintaining the simulation execution pace
at a manageable rate. This highlights the inherent trade-off
in DEM modeling between computational time and mak-
ing assumptions about performance below a certain particle
fragment size.
The modeling results have demonstrated a significant
potential for MonoRoll to achieve more energy-efficient
ore breakage compared to the pilot-scale ball mill. Notably,
DEM evaluations indicated that, even when operat-
ing without any ore in either machine, the MonoRoll
required only one-twelfth of the power consumed by the
Table 1. Breakage results of DEM 100 mm slice simulations
Feed Rate, t/h
Product After Each Breakage in Each Slice
Breakage, µm 1 2 3 4 5 6 7 8
1.8 P80 2800 2400 1500 1280 1020 840 710 640
P50 1600 1500 1300 940 790 670 600 560
1.5 P80 2500 2200 1400 1160 880 700 640 610
P50 1600 1500 1200 780 670 620 580 570
Figure 4. Schematic of DEM 100 mm MonoRoll slice simulations
During the pilot testing of CAHM, additional HPGR
tests were conducted at Corem’s facility using their Weir
pilot HPGR equipment. A DEM simulation of the Corem
HPGR equipment was also carried out. Comparison of
these results with CAHM pilot test results is discussed in
the subsequent sections.
MONOROLL DEM MODELING
MonoRoll design features are based on over 170 full and
partial simulations using the Ab-T10 breakage algorithm
in the Ansys Rocky DEM software. Combinations of liner
geometry, roll diameter, roll mass, and operating condi-
tions have been evaluated and compared based on model-
predicted power draw, mill stability, and basic breakage
behavior.
Given the computational intensity of these simulations,
even with state-of-the-art graphic processing units (GPUs),
the project required a practical approach to manage simula-
tion times. Simulations were conducted for both the entire
mill domain (full mill), and discrete portions of mill length
(mill slice). For mill slice simulations, the hammer roll and
shell were cut into equal-length (100 mm) slices to reduce
the total number of particles in the simulation domain.
Based on full mill simulation results and particle trans-
port features, particles pass under the comminution zone
between the hammer roll and the shell eight times on aver-
age. Thus, a total of eight slice simulations were conducted,
feeding each mill slice with the product size distribution
from the previous one as illustrated in Figure 4.
DEM mill slice simulation results for feed top size
(F100) of 12,500 µm, F80 of 9,000 µm, and feed rates of
1.5 t/h and 1.8 t/h are presented in Table 1. The minimum
fragment size for these simulations was set in the Ansys
Rocky DEM software as 300 µm. This artificial constraint
limits the smallest particle that can break to approximately
600 µm based on the particle splitting approximately in
half plus generating fragments finer than the preset 300 µm
limit.
The choice of a minimum fragment size of 300 µm is
a deliberate one, representing a balance between computa-
tional efficiency and the ability to reasonably predict prod-
uct particle size distribution. It is fine enough to provide a
reasonable prediction of the product particle size distribu-
tion while also maintaining the simulation execution pace
at a manageable rate. This highlights the inherent trade-off
in DEM modeling between computational time and mak-
ing assumptions about performance below a certain particle
fragment size.
The modeling results have demonstrated a significant
potential for MonoRoll to achieve more energy-efficient
ore breakage compared to the pilot-scale ball mill. Notably,
DEM evaluations indicated that, even when operat-
ing without any ore in either machine, the MonoRoll
required only one-twelfth of the power consumed by the
Table 1. Breakage results of DEM 100 mm slice simulations
Feed Rate, t/h
Product After Each Breakage in Each Slice
Breakage, µm 1 2 3 4 5 6 7 8
1.8 P80 2800 2400 1500 1280 1020 840 710 640
P50 1600 1500 1300 940 790 670 600 560
1.5 P80 2500 2200 1400 1160 880 700 640 610
P50 1600 1500 1200 780 670 620 580 570
Figure 4. Schematic of DEM 100 mm MonoRoll slice simulations