3836 XXXI International Mineral Processing Congress 2024 Proceedings/Washington, DC/Sep 29–Oct 3
the circuit. A summary of simulation results is presented in
Table 3.
Figure 3a shows the evolution of selection function
during the simulation for each component. As the mill
hold-up increases during simulation, growing from 60 t to
100 t, it is evident that material breakage rate decreases for
both ores as expected given Austin’s scale up equation for
the selection function (Eq. 2, where C
5 is a function of
the mill hold-up, and goes from 1.20 to 0.76). The curves
also show that CC ore presents breakage rates that are three
times higher than that of FN ore (Table 1), explaining its
higher size reduction and similar product P80 despite an
almost three times higher fresh feed P80.
Figure 3b shows the size distributions at steady-state
conditions. Even with the large difference in the fresh feed,
product and mill discharge size distributions for the com-
ponents are similar. Mill output P80 is nearly the same and
product P80 presents only a 0.01 mm difference. This effect
occurs due to the different grindability from each ore.
CONCLUSIONS
The unit models of single and multichamber ball mills, an
air classifier and a hydrocyclone have been implemented
in Dyssol platform with multicomponent capabilities in
regard to grindability and density.
In the case of the dry ball mill circuit, the simulations
showed that accumulation of either the heaviest or toughest
component was observed in the mill hold-up either due to
their higher likelihood to return to the mill from the classi-
fier or their longer residence times in the mill.
The wet grinding circuit simulation based on a real
plant showed that despite having different particle size dis-
tributions, due the difference on each ore specific breakage
rates, Fábrica Nova (FN) ore and Conceição ore (CC) pre-
sented similar product P80.
Finally, the simulations of the two selected ball mill cir-
cuits demonstrate Dyssol capabilities in dealing with multi-
component stream structure. The simulation platform was
able to run the time-dependent simulations without any
convergence issues for the selected simulation scenarios,
showing that it is a powerful tool for integrating an entire
concentration process flowsheet considering several levels
of multicomponent properties.
ACKNOWLEDGMENTS
The authors would like to acknowledge the financial support
from Brazilian Research Agencies CNPq (313222/2021-5
a) b)
0.1
1
10
100
0.1 1 10
dp (mm)
FN -Initial
CC -Initial
FN -Final
CC -Final
0
20
40
60
80
100
0.01 0.1 1 10 100
Sieve size (mm)
FN -Feed
CC -Feed
FN -Mill output
CC -Mill output
FN -Product
CC -Product
Figure 3. a) Evolution of selection function for both components and b) Size distributions for the multicomponent simulation
after steady-state conditions have been reached
Table 3. Wet grinding circuit simulation results
FN CC Overall
Fresh feed (t/h) 110 110 220
Feed composition (%)50 50 100
Fresh feed P80 (mm) 2.579 7.448 5.013
Hold-up composition (%)49.2 50.8 100
Circulating load (%)48.0 52.0 100
Product P80 (mm) 0.066 0.076 0.072
s
(1/min) Passing
(%)
the circuit. A summary of simulation results is presented in
Table 3.
Figure 3a shows the evolution of selection function
during the simulation for each component. As the mill
hold-up increases during simulation, growing from 60 t to
100 t, it is evident that material breakage rate decreases for
both ores as expected given Austin’s scale up equation for
the selection function (Eq. 2, where C
5 is a function of
the mill hold-up, and goes from 1.20 to 0.76). The curves
also show that CC ore presents breakage rates that are three
times higher than that of FN ore (Table 1), explaining its
higher size reduction and similar product P80 despite an
almost three times higher fresh feed P80.
Figure 3b shows the size distributions at steady-state
conditions. Even with the large difference in the fresh feed,
product and mill discharge size distributions for the com-
ponents are similar. Mill output P80 is nearly the same and
product P80 presents only a 0.01 mm difference. This effect
occurs due to the different grindability from each ore.
CONCLUSIONS
The unit models of single and multichamber ball mills, an
air classifier and a hydrocyclone have been implemented
in Dyssol platform with multicomponent capabilities in
regard to grindability and density.
In the case of the dry ball mill circuit, the simulations
showed that accumulation of either the heaviest or toughest
component was observed in the mill hold-up either due to
their higher likelihood to return to the mill from the classi-
fier or their longer residence times in the mill.
The wet grinding circuit simulation based on a real
plant showed that despite having different particle size dis-
tributions, due the difference on each ore specific breakage
rates, Fábrica Nova (FN) ore and Conceição ore (CC) pre-
sented similar product P80.
Finally, the simulations of the two selected ball mill cir-
cuits demonstrate Dyssol capabilities in dealing with multi-
component stream structure. The simulation platform was
able to run the time-dependent simulations without any
convergence issues for the selected simulation scenarios,
showing that it is a powerful tool for integrating an entire
concentration process flowsheet considering several levels
of multicomponent properties.
ACKNOWLEDGMENTS
The authors would like to acknowledge the financial support
from Brazilian Research Agencies CNPq (313222/2021-5
a) b)
0.1
1
10
100
0.1 1 10
dp (mm)
FN -Initial
CC -Initial
FN -Final
CC -Final
0
20
40
60
80
100
0.01 0.1 1 10 100
Sieve size (mm)
FN -Feed
CC -Feed
FN -Mill output
CC -Mill output
FN -Product
CC -Product
Figure 3. a) Evolution of selection function for both components and b) Size distributions for the multicomponent simulation
after steady-state conditions have been reached
Table 3. Wet grinding circuit simulation results
FN CC Overall
Fresh feed (t/h) 110 110 220
Feed composition (%)50 50 100
Fresh feed P80 (mm) 2.579 7.448 5.013
Hold-up composition (%)49.2 50.8 100
Circulating load (%)48.0 52.0 100
Product P80 (mm) 0.066 0.076 0.072
s
(1/min) Passing
(%)