3
capabilities and scope of the VisioFroth ™ system, integrat-
ing it into every aspect of the plant, we have developed a
robust expert control system capable of automatically oper-
ating the entire plant. This system is based on a handful
of fundamental principles and finetuned by advanced data
analytics and machine learning.
ADVANCED DATA ANALYTICS
A Foundation to Develop Digital One Expert Control
and Operating System
Advanced statistical and mathematical data modelling
played critical roles, randomly or systematically conducted
on various data sources from dynamic lab flotation tests
to large datasets from PI System. It’s a foundation to be
able to develop expert flotation control and especially the
Digital One Enterprise Expert Control and Operating
System. Numerous data models generated using multi-
variant response surface modeling are all statistically sig-
nificant using different set of process variables, reagents,
air and MIBC, and quantitative VisioFroth ™ bubble char-
acteristics. The large commercial and industrial flotation
plant at Buick mill, as complex as it is, or as stochastic as
it appears, is thought, and believed to be, indeed follows
fundamental laws, very strongly and persistently, meaning
thermodynamic and kinetic not just statistical and math-
ematical. So, flotation is pseudo-stochastic, and we can fun-
damentally influence, guide, direct, improve and optimize
it from so many directions, of course specifically from using
VisoFroth bubbles, air and MIBC, modern and bulk flota-
tion reagents, and XRF online readings. Therefore, a com-
prehensive enterprise expert flotation control and operating
system like Digital One is fundamentally certain, can be
developed and found, vastly expandable and by nature can
be very stable and very robust.
The individual controls for the expert control system
were developed not just because they are the obvious ones,
but because rigorous analysis of available data indicated that
they would be appropriately effective ones. Similarly, the
appropriateness of a control remaining in the expert system
is determined by aggressively ensuring that it improves the
overall system performance.
The kinetics of the hybrid reagents were found to be
very clear during plant trial. The flotation kinetics in the
presence of the hybrid reagents were found to be very well
fit by first order kinetic models, with clear selectivity and
effect towards the appropriate minerals.
This is also a large part of why it is possible to run these
reagents in the basic mode with just on/off controls. The
behavior of the reagents is well-behaved over the dosages
which the basic mode runs them at, and their influence is
sufficient at those minimum dosages to impact the targeted
minerals significantly but appropriately.
Similarly, strong indicator variables such as bubble tex-
ture being used to indicate mineral loading is similarly sup-
ported by statistical methods.
Plant and circuit data analysis is made possible via the
PI data hub which automatically records all major process
parameters, including VisioFroth ™ bubble characteristics,
a large dataset from 19 cameras, during the operation of
Buick Mill.
HYBRID REAGENT SYSTEM
Analysis and modeling of a large campaign of dynamic lab
flotation tests on Buick ores invariably revealed first order
flotation kinetics, with all model parameters statistically sig-
nificant for each test. This provides the confidence of using
two-level flotation dosing tests to screen 11 commercially
proven Solvay reagents, otherwise requiring many dozens
of tests even by today’s most efficient computerized custom
Design of Experiments.
The hybrid reagent system consists of five additional
Solvay collectors and depressants selected from screening
flotation tests that support the bulk chemicals when pro-
cessing complex ores. These five reagents allow for the flota-
tion circuits to be finetuned as needed. The reagents roles
are, as follows:
1. Iron depressant: Primarily to suppress marcasite
also supports cyanide, starch, and lime in suppress-
ing pyrite as necessary.
2. Pb collector: To improve Pb recovery when it is
not floating sufficiently in the bulk concentrate,
which equivalently reduces Pb present in the Zn
concentrate.
3. Pb depressant: To improve Pb recovery in the Cu
circuit, where Cu is floated away from the Pb to
separate the two concentrates.
4. Cu collector: To improve Cu recovery in the bulk
concentrate which feeds into the Cu circuit and
planned to improve recovery of the Cu in the Cu
scavenger circuit.
5. Zn collector: To improve Zn recovery in the Zn
circuit.
All these reagents are quite potent, with typical dosing
rates being on the order of 2–20 cc/ton of mill feed. Due to
the sensitivity of the system to these reagents, it is relatively
easy to overdose any of them while using them manually.
Automatic control of these reagents thus greatly increases
their effectiveness and reduces costs as well.
capabilities and scope of the VisioFroth ™ system, integrat-
ing it into every aspect of the plant, we have developed a
robust expert control system capable of automatically oper-
ating the entire plant. This system is based on a handful
of fundamental principles and finetuned by advanced data
analytics and machine learning.
ADVANCED DATA ANALYTICS
A Foundation to Develop Digital One Expert Control
and Operating System
Advanced statistical and mathematical data modelling
played critical roles, randomly or systematically conducted
on various data sources from dynamic lab flotation tests
to large datasets from PI System. It’s a foundation to be
able to develop expert flotation control and especially the
Digital One Enterprise Expert Control and Operating
System. Numerous data models generated using multi-
variant response surface modeling are all statistically sig-
nificant using different set of process variables, reagents,
air and MIBC, and quantitative VisioFroth ™ bubble char-
acteristics. The large commercial and industrial flotation
plant at Buick mill, as complex as it is, or as stochastic as
it appears, is thought, and believed to be, indeed follows
fundamental laws, very strongly and persistently, meaning
thermodynamic and kinetic not just statistical and math-
ematical. So, flotation is pseudo-stochastic, and we can fun-
damentally influence, guide, direct, improve and optimize
it from so many directions, of course specifically from using
VisoFroth bubbles, air and MIBC, modern and bulk flota-
tion reagents, and XRF online readings. Therefore, a com-
prehensive enterprise expert flotation control and operating
system like Digital One is fundamentally certain, can be
developed and found, vastly expandable and by nature can
be very stable and very robust.
The individual controls for the expert control system
were developed not just because they are the obvious ones,
but because rigorous analysis of available data indicated that
they would be appropriately effective ones. Similarly, the
appropriateness of a control remaining in the expert system
is determined by aggressively ensuring that it improves the
overall system performance.
The kinetics of the hybrid reagents were found to be
very clear during plant trial. The flotation kinetics in the
presence of the hybrid reagents were found to be very well
fit by first order kinetic models, with clear selectivity and
effect towards the appropriate minerals.
This is also a large part of why it is possible to run these
reagents in the basic mode with just on/off controls. The
behavior of the reagents is well-behaved over the dosages
which the basic mode runs them at, and their influence is
sufficient at those minimum dosages to impact the targeted
minerals significantly but appropriately.
Similarly, strong indicator variables such as bubble tex-
ture being used to indicate mineral loading is similarly sup-
ported by statistical methods.
Plant and circuit data analysis is made possible via the
PI data hub which automatically records all major process
parameters, including VisioFroth ™ bubble characteristics,
a large dataset from 19 cameras, during the operation of
Buick Mill.
HYBRID REAGENT SYSTEM
Analysis and modeling of a large campaign of dynamic lab
flotation tests on Buick ores invariably revealed first order
flotation kinetics, with all model parameters statistically sig-
nificant for each test. This provides the confidence of using
two-level flotation dosing tests to screen 11 commercially
proven Solvay reagents, otherwise requiring many dozens
of tests even by today’s most efficient computerized custom
Design of Experiments.
The hybrid reagent system consists of five additional
Solvay collectors and depressants selected from screening
flotation tests that support the bulk chemicals when pro-
cessing complex ores. These five reagents allow for the flota-
tion circuits to be finetuned as needed. The reagents roles
are, as follows:
1. Iron depressant: Primarily to suppress marcasite
also supports cyanide, starch, and lime in suppress-
ing pyrite as necessary.
2. Pb collector: To improve Pb recovery when it is
not floating sufficiently in the bulk concentrate,
which equivalently reduces Pb present in the Zn
concentrate.
3. Pb depressant: To improve Pb recovery in the Cu
circuit, where Cu is floated away from the Pb to
separate the two concentrates.
4. Cu collector: To improve Cu recovery in the bulk
concentrate which feeds into the Cu circuit and
planned to improve recovery of the Cu in the Cu
scavenger circuit.
5. Zn collector: To improve Zn recovery in the Zn
circuit.
All these reagents are quite potent, with typical dosing
rates being on the order of 2–20 cc/ton of mill feed. Due to
the sensitivity of the system to these reagents, it is relatively
easy to overdose any of them while using them manually.
Automatic control of these reagents thus greatly increases
their effectiveness and reduces costs as well.