932 XXXI International Mineral Processing Congress 2024 Proceedings/Washington, DC/Sep 29–Oct 3
implement the Optimizer functionality of the Flotation
Optimization App. Working closely with the operation the
Value Drivers and operating limits were captured as a series
of rewards and constraints in brains.app. These guide the
operation of the Optimizer, providing operators every 30
minutes with new recommended setpoints for airflow and
froth depth levels–and for reagent dosage as soon as the
feed properties change. brains.app’s configurable UI guides
operators in the task of updating these setpoints, and pro-
vides them with the expected gains in recovery and grade
that the introduced changes will make in the circuit. It is
worth noting that the Optimizer can be configured to pro-
vide recommended setpoints more often, to as little as every
one minute.
With the Optimizer configured it was applied to his-
torical data, quantifying an opportunity gain in production.
The difference between the as-is and the optimised metrics
over the historical period is a measure of the value that the
App can capture for the operation. This improvement was
estimated to positively impact the overall iron recovery of
the plant from 0.8% (25th percentile) up to 4.8% (75th
percentile), depending on feed conditions and with respect
to a base recovery of 70.8%. This comes also at an increase
in the iron grade from 65% to 66.6% (median) directly
attributed to the Flotation Optimization App. The results
obtained for this case study are in line with the quantified
value that the Flotation Optimization App has delivered in
other sites and commodities from 1–3% recovery increases
across different commodities, including copper and gold.
This level of enhancement represents significant additional
revenue for miners, with value captured adding directly to
the miners bottomline.
CONCLUSIONS
Froth flotation is one of the most important means of
concentrating minerals in the industry. Annually, billions
of tonnes of ore are processed by flotation circuits world-
wide, making even the smallest improvements hugely
impactful. Metallurgists and control room operators have
the challenge of having to optimise the performance of a
multivariate process with limited information. This trans-
lates into circuits being operated with minimum changes
to their setpoints, even under changes in feed conditions.
In this paper, we have summarised the development and
implementation of an optimization solution that includes
the configuration of a Digital Process Model capable of
responding to feed characteristics and operating param-
eters in real-time. This Digital Process Model generates
hydrodynamic and plant performance Virtual Sensors that
allow metallurgists and control room operators to better
understand the state of the flotation circuit. The simula-
tion model is used to reproduce the circuit’s response to
Figure 4. Bubble diameter and surface area flux profile across Rougher banks (R1 and R2). Different operating profiles are
present in these parallel banks, which are made apparent for operators and metallurgists
implement the Optimizer functionality of the Flotation
Optimization App. Working closely with the operation the
Value Drivers and operating limits were captured as a series
of rewards and constraints in brains.app. These guide the
operation of the Optimizer, providing operators every 30
minutes with new recommended setpoints for airflow and
froth depth levels–and for reagent dosage as soon as the
feed properties change. brains.app’s configurable UI guides
operators in the task of updating these setpoints, and pro-
vides them with the expected gains in recovery and grade
that the introduced changes will make in the circuit. It is
worth noting that the Optimizer can be configured to pro-
vide recommended setpoints more often, to as little as every
one minute.
With the Optimizer configured it was applied to his-
torical data, quantifying an opportunity gain in production.
The difference between the as-is and the optimised metrics
over the historical period is a measure of the value that the
App can capture for the operation. This improvement was
estimated to positively impact the overall iron recovery of
the plant from 0.8% (25th percentile) up to 4.8% (75th
percentile), depending on feed conditions and with respect
to a base recovery of 70.8%. This comes also at an increase
in the iron grade from 65% to 66.6% (median) directly
attributed to the Flotation Optimization App. The results
obtained for this case study are in line with the quantified
value that the Flotation Optimization App has delivered in
other sites and commodities from 1–3% recovery increases
across different commodities, including copper and gold.
This level of enhancement represents significant additional
revenue for miners, with value captured adding directly to
the miners bottomline.
CONCLUSIONS
Froth flotation is one of the most important means of
concentrating minerals in the industry. Annually, billions
of tonnes of ore are processed by flotation circuits world-
wide, making even the smallest improvements hugely
impactful. Metallurgists and control room operators have
the challenge of having to optimise the performance of a
multivariate process with limited information. This trans-
lates into circuits being operated with minimum changes
to their setpoints, even under changes in feed conditions.
In this paper, we have summarised the development and
implementation of an optimization solution that includes
the configuration of a Digital Process Model capable of
responding to feed characteristics and operating param-
eters in real-time. This Digital Process Model generates
hydrodynamic and plant performance Virtual Sensors that
allow metallurgists and control room operators to better
understand the state of the flotation circuit. The simula-
tion model is used to reproduce the circuit’s response to
Figure 4. Bubble diameter and surface area flux profile across Rougher banks (R1 and R2). Different operating profiles are
present in these parallel banks, which are made apparent for operators and metallurgists