924 XXXI International Mineral Processing Congress 2024 Proceedings/Washington, DC/Sep 29–Oct 3
flotation laws, but also with Treated PW, which might be
due to the relative more pronounced effect on Cu flotation
and higher recovery. This overall improvement of perfor-
mance requires more intensive water treatment (Table 6).
DISCUSSION, LIMITS AND PERPECTIVES
Dissolution Model and Experimental Protocol
Dissolution model calibration highlighted some difficulties
regarding model parameters setting: several Cmax –k pairs
enabled suitable fitting of experimental results obtained
with distilled water as starting water. However, integrat-
ing these parameters into the plant model shows that the
latter is sensitive to them. It is therefore necessary to be
able to determine which pair will provide the most realistic
behavior. The dissolution model seems to show a limitation
of the experimental protocol, related to the need to dilute
with starting water, which leads to pseudo-equilibrium and
not a thermodynamical one. It might be an option to use
thermodynamic modelling to determine Cmax, and back-
calculate k from dissolution test fit with sole distilled water.
From an experimental aspect, dissolution test with process
water enabled to extend the concentration range, getting
closer to the equilibrium. The joint use of data sets related
to distilled and process water enabled to refine the calibra-
tion, at least for sulfates. However, in the context of a pro-
spective study, process water will not be available until the
plant is in operation. It would therefore seem worthwhile
completing the experimental protocol to obtain extended
experimental ranges with higher concentrations, by adapt-
ing the test protocol to reduce the dilution due to loss
compensation.
Water Characterization Requirements
As mentioned previously, a major difficulty in understand-
ing the impact of process water quality on plant opera-
tion lies in the characterization of the water itself (Le,
Schreitofer and Dahl 2020, Le et al., 2020, Le 2021).
This implies uncertainties on the experimental data, and
hence in the calibration of the models based on these data.
Among others, this regards some components which have
been left out of this paper, but which can have a negative
impact on flotation performances (fines, colloids). Another
point is the evaluation of residual reagent concentrations,
Table 6. Water treatment sizing
Scenario
DAF Capacity,
m3/h
Number of IER SO
4 Columns
IER SO
4 Adsorption
Cycle Duration, h
IER Ca Requirement,
Input/Output
PW 1018.5 7 53.6 190.0 /146.5 mg/L
Concentrated PW
Treated PW
830.2
1316.3
6
8
77.0
41.7
223.1 /178.2 mg/L
155.2 /115.4 mg/L
for which assumptions have been made here, the impor-
tance of which is also mentioned by Güner, Bulut and
Yenial (2019) who carried out similar tests (water reuse on
chalcopyrite flotation).
The modeling approach proposed here is based on
several experimental results related to different unit mod-
els: flotation, dissolution, water treatment. It seems man-
datory that the protocols used are homogeneous between
these experiments, but also that the same parameters are
monitored, in particular the residual concentrations of the
reagents involved in the various unit operations. For exam-
ple, a water treatment operation can lead to the addition of
dissolved compounds by overdosing the reagent, which can
have an impact on flotation. This is in line with the need
to take a global approach to the problem, which should be
integrated in experimental protocol definition: “each” com-
ponent that could have an impact on another operation
should be systematically measured, rather than carrying out
experiments “in silo,” focusing only on the studied opera-
tion parameters, at the risk of missing some interaction.
Another major difficulty lies in determining the flo-
tation laws. Numerous parameters can have an influence,
and only a few calibration points are generally available due
to the cost of the experiments or experimentation limits,
which can lead to model overfitting. This leads to the use of
simple laws, but as seen in this paper, they quickly become
unrealistic as soon as used beyond experimental ranges.
This goes in the direction of extending dissolution tests, to
obtain more concentrated waters that would also broaden
the experimental range for flotation tests, which appears
necessary to tackle short-loop scenarios. In parallel, there
is also a need to deepen fundamental understanding of
the effect of electrolytes on flotation (Manono, Corin and
Wiese 2018). Knowledge of the predominant mechanisms
according to the experimental ranges would give indications
for targeting the parameters to be retained in empiric laws,
and qualitatively, information on their impact (negative/
positive, linear or not, interactions between parameters).
This would provide a trade-off between model efficiency
and simplicity, with more predictive laws conscribed to the
strict minimum, which is not always decidable when there
are few experimental results compared to the number of
parameters.
flotation laws, but also with Treated PW, which might be
due to the relative more pronounced effect on Cu flotation
and higher recovery. This overall improvement of perfor-
mance requires more intensive water treatment (Table 6).
DISCUSSION, LIMITS AND PERPECTIVES
Dissolution Model and Experimental Protocol
Dissolution model calibration highlighted some difficulties
regarding model parameters setting: several Cmax –k pairs
enabled suitable fitting of experimental results obtained
with distilled water as starting water. However, integrat-
ing these parameters into the plant model shows that the
latter is sensitive to them. It is therefore necessary to be
able to determine which pair will provide the most realistic
behavior. The dissolution model seems to show a limitation
of the experimental protocol, related to the need to dilute
with starting water, which leads to pseudo-equilibrium and
not a thermodynamical one. It might be an option to use
thermodynamic modelling to determine Cmax, and back-
calculate k from dissolution test fit with sole distilled water.
From an experimental aspect, dissolution test with process
water enabled to extend the concentration range, getting
closer to the equilibrium. The joint use of data sets related
to distilled and process water enabled to refine the calibra-
tion, at least for sulfates. However, in the context of a pro-
spective study, process water will not be available until the
plant is in operation. It would therefore seem worthwhile
completing the experimental protocol to obtain extended
experimental ranges with higher concentrations, by adapt-
ing the test protocol to reduce the dilution due to loss
compensation.
Water Characterization Requirements
As mentioned previously, a major difficulty in understand-
ing the impact of process water quality on plant opera-
tion lies in the characterization of the water itself (Le,
Schreitofer and Dahl 2020, Le et al., 2020, Le 2021).
This implies uncertainties on the experimental data, and
hence in the calibration of the models based on these data.
Among others, this regards some components which have
been left out of this paper, but which can have a negative
impact on flotation performances (fines, colloids). Another
point is the evaluation of residual reagent concentrations,
Table 6. Water treatment sizing
Scenario
DAF Capacity,
m3/h
Number of IER SO
4 Columns
IER SO
4 Adsorption
Cycle Duration, h
IER Ca Requirement,
Input/Output
PW 1018.5 7 53.6 190.0 /146.5 mg/L
Concentrated PW
Treated PW
830.2
1316.3
6
8
77.0
41.7
223.1 /178.2 mg/L
155.2 /115.4 mg/L
for which assumptions have been made here, the impor-
tance of which is also mentioned by Güner, Bulut and
Yenial (2019) who carried out similar tests (water reuse on
chalcopyrite flotation).
The modeling approach proposed here is based on
several experimental results related to different unit mod-
els: flotation, dissolution, water treatment. It seems man-
datory that the protocols used are homogeneous between
these experiments, but also that the same parameters are
monitored, in particular the residual concentrations of the
reagents involved in the various unit operations. For exam-
ple, a water treatment operation can lead to the addition of
dissolved compounds by overdosing the reagent, which can
have an impact on flotation. This is in line with the need
to take a global approach to the problem, which should be
integrated in experimental protocol definition: “each” com-
ponent that could have an impact on another operation
should be systematically measured, rather than carrying out
experiments “in silo,” focusing only on the studied opera-
tion parameters, at the risk of missing some interaction.
Another major difficulty lies in determining the flo-
tation laws. Numerous parameters can have an influence,
and only a few calibration points are generally available due
to the cost of the experiments or experimentation limits,
which can lead to model overfitting. This leads to the use of
simple laws, but as seen in this paper, they quickly become
unrealistic as soon as used beyond experimental ranges.
This goes in the direction of extending dissolution tests, to
obtain more concentrated waters that would also broaden
the experimental range for flotation tests, which appears
necessary to tackle short-loop scenarios. In parallel, there
is also a need to deepen fundamental understanding of
the effect of electrolytes on flotation (Manono, Corin and
Wiese 2018). Knowledge of the predominant mechanisms
according to the experimental ranges would give indications
for targeting the parameters to be retained in empiric laws,
and qualitatively, information on their impact (negative/
positive, linear or not, interactions between parameters).
This would provide a trade-off between model efficiency
and simplicity, with more predictive laws conscribed to the
strict minimum, which is not always decidable when there
are few experimental results compared to the number of
parameters.