XXXI International Mineral Processing Congress 2024 Proceedings/Washington, DC/Sep 29–Oct 3 2563
already in the optimum range and due to technical limita-
tions, it was left unchanged.
The start simplex consists of a total of five components,
four of which close to the optimal parameter settings,
which were found on a pilot scale. The remaining point was
set outside the pre-identified optimal range (third point in
Table 10). The selection of this point was made carefully
to ensure that the optimization is carried out in the right
direction. The results of the first simplex experiment con-
firmed this, as this specific point had the lowest GRP index.
This point was also associated with a financial loss of sev-
eral thousand euros. This underlines the importance of this
workflow to avoid industrial experiments that could have a
negative impact on the process when moving to industrial
scale.
This point is replaced by a reflection point that has an
improved GRP index. The next reflection, expansion or
contraction points are formed iteratively and this process
continues until the termination criterion (standard devia-
tion of GRP values less than or equal to 5%) is reached.
Figures 9 to 15 show the optimization path of the simplex
through the formation of reflection and expansion points
until it reaches the termination criterion and converges.
By reaching the optimum, the scheelite grade was
increased by 16%, while at the same time the yield was
increased by 3% as shown in Figure 16.
Table 9. Comparison between the results of the benchmark process and the after 4 iterations achieved optimal parameter
settings for the full and the simplified simplex
pH
Rougher
pH
Cleaner
Collector
Dosage,
g/t
Frother
Dosage,
g/t
Quebracho
Dosage,
g/t
CS
Dosage,
g/t
Scheelite
Grade,
%
Scheelite
Recovery,
%
Calcite
Recovery,
%
Benchmark 9.8 9 60 20 16 0 4.38 84.36 25.84
Full simplex 10 9.24 116 49 21 315 8.11 75.95 7.06
Simplified simplex 9.79 9.54 90 33 24 325 7.11 74.09 10.01
Table 10. Parameter settings of the start simplex of the industrial trials
Iteration 1 – Start-Simplex Results
Quebracho
Dosage,
g/t
Collector
Dosage,
g/t
CS Dosage
Rougher Stage,
g/t
CS Dosage
Cleaner Stage,
g/t
Scheelit
Grade,
%
Scheelit
Recovery,
%GRP-Index
16 60 0 0 29.40 85.46 1.63
20 80 0 250 37.63 86.31 1.84
5 40 250 0 25.78 83.46 1.53
22 100 125 125 38.99 87.15 1.88
8 110 180 70 26.84 86.92 1.59
Figure 9. Two-dimensional scatter plot of the starting simplex of the first iteration as a function of the dosage of colloidal
silica in the rougher in cleaner stage
already in the optimum range and due to technical limita-
tions, it was left unchanged.
The start simplex consists of a total of five components,
four of which close to the optimal parameter settings,
which were found on a pilot scale. The remaining point was
set outside the pre-identified optimal range (third point in
Table 10). The selection of this point was made carefully
to ensure that the optimization is carried out in the right
direction. The results of the first simplex experiment con-
firmed this, as this specific point had the lowest GRP index.
This point was also associated with a financial loss of sev-
eral thousand euros. This underlines the importance of this
workflow to avoid industrial experiments that could have a
negative impact on the process when moving to industrial
scale.
This point is replaced by a reflection point that has an
improved GRP index. The next reflection, expansion or
contraction points are formed iteratively and this process
continues until the termination criterion (standard devia-
tion of GRP values less than or equal to 5%) is reached.
Figures 9 to 15 show the optimization path of the simplex
through the formation of reflection and expansion points
until it reaches the termination criterion and converges.
By reaching the optimum, the scheelite grade was
increased by 16%, while at the same time the yield was
increased by 3% as shown in Figure 16.
Table 9. Comparison between the results of the benchmark process and the after 4 iterations achieved optimal parameter
settings for the full and the simplified simplex
pH
Rougher
pH
Cleaner
Collector
Dosage,
g/t
Frother
Dosage,
g/t
Quebracho
Dosage,
g/t
CS
Dosage,
g/t
Scheelite
Grade,
%
Scheelite
Recovery,
%
Calcite
Recovery,
%
Benchmark 9.8 9 60 20 16 0 4.38 84.36 25.84
Full simplex 10 9.24 116 49 21 315 8.11 75.95 7.06
Simplified simplex 9.79 9.54 90 33 24 325 7.11 74.09 10.01
Table 10. Parameter settings of the start simplex of the industrial trials
Iteration 1 – Start-Simplex Results
Quebracho
Dosage,
g/t
Collector
Dosage,
g/t
CS Dosage
Rougher Stage,
g/t
CS Dosage
Cleaner Stage,
g/t
Scheelit
Grade,
%
Scheelit
Recovery,
%GRP-Index
16 60 0 0 29.40 85.46 1.63
20 80 0 250 37.63 86.31 1.84
5 40 250 0 25.78 83.46 1.53
22 100 125 125 38.99 87.15 1.88
8 110 180 70 26.84 86.92 1.59
Figure 9. Two-dimensional scatter plot of the starting simplex of the first iteration as a function of the dosage of colloidal
silica in the rougher in cleaner stage