XXXI International Mineral Processing Congress 2024 Proceedings/Washington, DC/Sep 29–Oct 3 887
NMPC Application
As already discussed in Section “Case study simulation,”
step changes in the HPGR feed BSA do not represent a
realistic portrayal of the HPGR operation. As such, based
on HPGR feed BSA changes predicted by Eq. (21) and
presented in Figure 2b, Figure 4 depicts the behavior of
the product BSA (a) and throughput (b) over the 180-min-
ute simulation for case study #1. Although significant
variations occur in the feed BSA at two distinct moments
(60 and 120 minutes), it is worth noting that the NMPC
managed to keep both controlled variables close to the ref-
erence setpoints. However, results presented for the period
after 140 minutes of operation highlight the limitation
of the NMPC in dealing with a feed BSA close to 1650
cm2/g. In this situation, there was a significant increase in
the product Blaine specific surface area (Figure 4a), accom-
panied by a reduction in throughput (Figure 4b).
Variations in manipulated variables are then presented
in Figure 5, where it is possible to observe significant changes
in the values of operating pressure (a) and roll peripheral
Figure 2. Changes in the HPGR feed BSA demonstrating the dynamics described by Eq. (21) during a time interval of 50 min
(a) and proposed changes predicted by Eq. (21) in case studies #1 and #2 in a time interval of 180 min (b)
Figure 3. Response of the product BSA (a) and throughput (b) for different prediction and control horizons in the NMPC. PH
and CH are, respectively, the prediction and control horizons
NMPC Application
As already discussed in Section “Case study simulation,”
step changes in the HPGR feed BSA do not represent a
realistic portrayal of the HPGR operation. As such, based
on HPGR feed BSA changes predicted by Eq. (21) and
presented in Figure 2b, Figure 4 depicts the behavior of
the product BSA (a) and throughput (b) over the 180-min-
ute simulation for case study #1. Although significant
variations occur in the feed BSA at two distinct moments
(60 and 120 minutes), it is worth noting that the NMPC
managed to keep both controlled variables close to the ref-
erence setpoints. However, results presented for the period
after 140 minutes of operation highlight the limitation
of the NMPC in dealing with a feed BSA close to 1650
cm2/g. In this situation, there was a significant increase in
the product Blaine specific surface area (Figure 4a), accom-
panied by a reduction in throughput (Figure 4b).
Variations in manipulated variables are then presented
in Figure 5, where it is possible to observe significant changes
in the values of operating pressure (a) and roll peripheral
Figure 2. Changes in the HPGR feed BSA demonstrating the dynamics described by Eq. (21) during a time interval of 50 min
(a) and proposed changes predicted by Eq. (21) in case studies #1 and #2 in a time interval of 180 min (b)
Figure 3. Response of the product BSA (a) and throughput (b) for different prediction and control horizons in the NMPC. PH
and CH are, respectively, the prediction and control horizons