XXXI International Mineral Processing Congress 2024 Proceedings/Washington, DC/Sep 29–Oct 3 1039
which are not considered in the models, particularly when
throughput is constrained by events occurring upstream or
downstream of the comminution circuit. Despite these lim-
itations, the predictions generally exhibit a reasonable align-
ment with the observed trends for the variables accounted
for in the model. The average absolute error, indicating the
deviation between predicted and actual values, was found
to be 3.3%. Figure 4 depicts the comparison of the SAG
power predicted over fourteen months period and then
compared that with the actual data. The figure shows IES
was capable of predicting SAG mill power as well as mill
load with an average error of 1.5%. The combined cyclone
overflow P80 and IES predicted P80 is shown in Figure 5
The average absolute error was 4.3%.
A more refined validation of the flowsheet was con-
ducted utilizing data from September 2022. This step aimed
to assess the flowsheet’s predictability using a dataset distinct
from the one employed in the tuning process. Despite the
less-than-ideal frequency of daily simulations, constrained
by day-to-day operational and maintenance issues within
the operation, the flowsheet demonstrated commendable
capability in forecasting throughput and grind size trends
observed during that month refer to Figure 6 and Figure 7.
Remarkably, the flowsheet exhibited a monthly average
error of merely 0.5% for throughput and 2.2% for grind
size, indicating its robust predictability. These results affirm
that the flowsheet can be reliably employed for both plan-
ning and optimization.
-20%
0%
20%
40%
60%
80%
100%
12000
12500
13000
13500
14000
14500
Error Actual IES
Figure 4. Actual and IES predicted plant SAG Mill power from Jan 2021 to Mar 2022
-20%
0%
20%
40%
60%
80%
100%
150
160
170
180
190
200
210
220
230
Error Series1 Series2
Figure 5. Actual and IES predicted combined cyclone overflow from Jan 2021 to Mar 2022
Error
(%)
SAG
Mill
Power
(kW)
Error
(%)
Grind
Size
(
)
which are not considered in the models, particularly when
throughput is constrained by events occurring upstream or
downstream of the comminution circuit. Despite these lim-
itations, the predictions generally exhibit a reasonable align-
ment with the observed trends for the variables accounted
for in the model. The average absolute error, indicating the
deviation between predicted and actual values, was found
to be 3.3%. Figure 4 depicts the comparison of the SAG
power predicted over fourteen months period and then
compared that with the actual data. The figure shows IES
was capable of predicting SAG mill power as well as mill
load with an average error of 1.5%. The combined cyclone
overflow P80 and IES predicted P80 is shown in Figure 5
The average absolute error was 4.3%.
A more refined validation of the flowsheet was con-
ducted utilizing data from September 2022. This step aimed
to assess the flowsheet’s predictability using a dataset distinct
from the one employed in the tuning process. Despite the
less-than-ideal frequency of daily simulations, constrained
by day-to-day operational and maintenance issues within
the operation, the flowsheet demonstrated commendable
capability in forecasting throughput and grind size trends
observed during that month refer to Figure 6 and Figure 7.
Remarkably, the flowsheet exhibited a monthly average
error of merely 0.5% for throughput and 2.2% for grind
size, indicating its robust predictability. These results affirm
that the flowsheet can be reliably employed for both plan-
ning and optimization.
-20%
0%
20%
40%
60%
80%
100%
12000
12500
13000
13500
14000
14500
Error Actual IES
Figure 4. Actual and IES predicted plant SAG Mill power from Jan 2021 to Mar 2022
-20%
0%
20%
40%
60%
80%
100%
150
160
170
180
190
200
210
220
230
Error Series1 Series2
Figure 5. Actual and IES predicted combined cyclone overflow from Jan 2021 to Mar 2022
Error
(%)
SAG
Mill
Power
(kW)
Error
(%)
Grind
Size
(
)