FLOTATION CONTROL & OPTIMISATION

A global leader in mineral and metallurgical innovation

FLOATSTAR OVERVIEW

Flotation is a complex process that is affected by a multitude of

factors. These factors may be inherent in the circuit design, or in
how the flotation plant is operated. The FloatStar suite of control
modules utilises advanced process control to overcome
design-related limitations and maximise circuit performance during
operation.
As far as possible, Minteks control modules are designed to be
implemented with little or no additional instrumentation. Minteks
modular approach ensures the design of a customised solution,
specific to the objectives of each plant.
Furthermore, this approach allows for phased implementations
which are useful both in assessing the benefit of the system, and in
allowing plant personnel to be introduced gradually to any changes
in operation (Change-Management).
The diagram below illustrates Minteks bottom-up approach to
flotation stabilisation and optimisation:

Management
Information
System
FloatStar Flow Optimiser
FloatStar Grade-Recovery Optimiser
FloatStar Reagent Optimiser

Optimisation

FloatStar Level Stabiliser

Stabilisation

FloatStar pH Controller
FloatStar Level Fault Detector

Process Monitoring

FloatStar Performance Monitor

Plant Control
FloatStar Dynamic Simulator

Simulators

Plant Instrumentation
Plant

STABILISATION
Level Stabiliser
The FloatStar Level Stabiliser has been implemented on flotation
circuits around the world. It has repeatedly been proven efficient in
rejection of disturbances, and in rapid tracking of setpoints.

Bank 1
Bank 2
Bank 3

The main difficulty in controlling flotation levels is that they form part
of a highly interconnected system. The control actions of one bank
will therefore influence other banks and, unless handled correctly,
disturbances will propagate through the circuit. This means that good
level stabilisation control cannot be achieved by using controllers
that only act locally. The top two figures on the right show how a
disturbance will propagate through the circuit.
The FloatStar Level Stabiliser solves this problem, and is designed to
provide the following benefits:
Bank 1

Level

Fast setpoint tracking by:

Considering all levels simultaneously to eliminate interaction.

Rejecting disturbances quickly and efficiently throughout the

entire circuit.
Reduced startup time:

Typically a reduction from over 3 hours to under 1 hour in the

time taken to stabilise the final tails grade.

Bank 2

Disturbance is
Propagated

Bank 3

Time

Improved overall recovery:

On numerous plants, processing a wide range of mineral types,

the FloatStar Level Stabiliser has consistently increased recovery
by between 0.5% and 1.3%.

On most minerals processing plants, the payback period is in

the order of 1 to 3 months.

Case study

Comparison of controller performance

The graph on the right shows a comparison of normal PID control to
FloatStar Level Stabilisation. Worth noting is how, under PID control,
the disturbances from the first banks are magnified further
downstream. The control has visibly improved while the FloatStar Level
Stabiliser was in operation. The general deviation from setpoint has
decreased markedly, and changes to setpoint are tracked rapidly
and without producing downstream disturbances.

Plant PID Control

Level (% - values shifted for clarity)

Performance testing, with a benefit analysis, has been performed on

concentrator plants of several different mineral types. On-Off tests
during normal operation, as well as benefit achieved during startups
were assessed.

700

FloatStar Level
Stabiliser

600
500
400
300
200
100
0
05:30:00 06:20:00 07:10:00 08:00:00 08:50:00
Time (hours)

The table below shows the Integral of Absolute Error (IAE) controller
performance measure. The deviation from setpoint under FloatStar
Level Stabiliser control is considerably lower than under normal plant
PID control.

IAE

Primary
Primary
Rougher1 Rougher2

Primary
Rougher3

Primary
Rougher4

Primary
Rougher5

Primary
Rougher6

Primary
Rougher7

Primary
Rougher8

Primary
Rougher9

IAE OFF

13.42

14.84

25.59

56.13

93.76

129.64

139.89

195.76

116.64

IAE ON

9.00

7.58

8.38

10.32

6.29

6.60

8.10

8.63

8.34

Tailings Grade

2500

Mineral 2 (ppm)

The plant used for this case study processed two different mineral
types. Measurements were taken regularly while the plant was
starting up, once under PLC PID control and once under FloatStar
Level Stabiliser control. The graph on the right shows a
comparison of the tailings grade measurements during the
startup procedure. It is clear that the startup under FloatStar
Level Stabiliser control stabilises more rapidly, and with lower
tailings grades, than that using PID control.

Time Period
Considered

2000

1.4

1.6 hrs

1500

0.8
0.6

1000

0.4

500
0

The table below shows the quantitative economic evaluation

of the savings achieved by using the FloatStar Level Stabiliser
for circuit startup. A total savings of US$7500 was achieved on
each startup by using the FloatStar system.

1.2
Mineral 1 (ppm)

Benefit during startup

0.2
0

20

0
40
60 80 100 120 140
Time (minutes)

Mineral 1 FloatStar
Mineral 2 Floatstar

Mineral 1 PID
Mineral 2 PID

Mass flow of tailings stream: 1975 tons/hr

Time period of benefit/loss calculation: 1.6 hrs
Cost of mineral 1: US$ 350/oz

Cost of mineral 2: US$ 1528/ton

Mineral 1 lost by PID control compared to FloatStar control: 0.2ppm

Mineral 2 lost by PID control compared to FloatStar control 250 ppm

Mass of mineral 1 lost 527g (18.6oz)

Mass of mineral 2 lost: 0.66 tons

Cost of mineral 1 lost US$ 6500

Cost of mineral 2 lost: US$ 1000

Savings by using FloatStar Level Stabiliser during startup: US$ 7500

OPTIMISATION

Flow Optimiser
With the FloatStar Level Stabiliser providing circuit stability and rapid
setpoint tracking, it then becomes possible to consider flotation
optimisation.
The FloatStar Flow Optimiser controls circulating loads and mass
pull in the flotation circuit. It is well known that circuit performance
(grade and recovery) is strongly affected by these parameters. By
stabilising circulating load and mass pull, it is possible to ensure
consistent circuit performance.
The Flow Optimiser uses multivariable control techniques to
continually optimise the numerous variables that affect circuit
flows. Typically these are the level and air setpoints in different
sections of the plant.
Depending on the circuit configuration and available
instrumentation, the FloatStar Flow Optimiser can be used to control
any of the following parameters:

Mass pull rates.

Residence times.
Internal flowrates.

Controller

Required
Equipment

Mass Pull Control

Flowmeter on
Concentrate Line
Densitometer on
Concentrate Line

Residence
Time Controller

None

Flow Controller

Flowmeter

Mass Pull Control

In order to obtain a constant mass pull from a flotation train, all of
the banks that produce concentrate need to be set at the correct
level and air setpoints, as these affect the mass pull achieved. The
FloatStar Flow Optimiser makes use of multivariable techniques to
achieve the desired mass pull rates from these banks, in the manner
depicted by the image right:

AIR

By automatically adjusting the level and air setpoints on the final

stages, the mass pull can be accurately controlled and maintained
throughout changing ore conditions and circuit upsets, thereby
ensuring tonnage targets.

Controlled
Variable:

Mass Pull

Measured
Values:

Density & Flowrate

of Concentrate

Manipulated
Variables:

Level and Air Setpoints

AIR

Mass Pull Setpoint

Sump

Pump

Concentrate

Residence Time/Re-Circulating Load Controller

The mechanics of flotation shows that there is a direct impact on
recovery as the residence time changes.

AIR

AIR

AIR

AIR

LoadSP/

It is possible with the FloatStar Flow Optimiser to control the inferred

flow, and hence the residence time of a bank, to setpoint. The
residence time (and re-circulating load in the case of a closed circuit)
can be stabilised ensuring maximum efficiency of key flotation banks
thereby improving overall circuit performance.

Controlled
Variable:
Measured
Values:
Manipulate
d Variables:

Controller

Residence
time

Residence Time
None - Values Inferred
Air and Level Setpoints
of Banks Supplying
Concentrate

Case Study of implemented Residence Time controller

Consider the flotation circuit shown in the figure right which is currently
employed by one of our clients. The aim is to control the residence
time of PC6 by adjusting the setpoints of the banks supplying
concentrate. The multivariable algorithm of the optimiser determines
changes to the air and level setpoints.

SP

PR1

PR2

PC1

SR1
PR3

PC2

Circuit Configuration

PR: Primary Rougher

SR: Secondary Rougher
PC: Primary Cleaner

SR2

PR4

PC3

PC6

PC4

PC5

SR3

Froth Depth (mm)

A special estimating filter based on available plant measurements

Control of Concentrate Flow.
Manipulation of Aeration Rate Setpoints.
is used to infer the flowrate through PC6 and hence estimate the
FloatStar Level Stabiliser
FloatStar Flow Optimiser
residence time. With this estimation
4.5
4
Control of Concentrate Flow.
the optimising algorithm is able to
3.5
Manipulation of Rougher Level-Setpoints.
3
adjust the level and aeration
FloatStar Level
FloatStar
Flow Optimiser
PID
2.5
Stabiliser
2
setpoints of the Roughers and
PR 1
800
1.5
PR 2
PR 3
Cleaners.
1
700
0.5

600

500
400

0 1 2 3 4 5

PR 4
SR 1
SR 2
6SR
7 38
PC 6

12 13 14 15 16 17 18 19 20 21 22 23 24 25

Time (hours)

PC 7

300

PC 8
PC 9
PC 10

200
100
0

PR: Primary Rougher

SR: Secondary Rougher
PC: Primary Cleaner

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25

Time (hours)

The graph right shows the inferred cleaner tailings flow over a
25-hour period, which is directly related to the residence time in the
bank. Under normal PID control, it can be seen that there is virtually
no control of the residence time.

Standard Deviation of Flow Error from Setpoint

PID

4.20

FloatStar

2.20

Flow Optimiser

1.44

Another example of effective circuit optimisation is the control of

a tailings flowrate in the cleaning stages. The level and air setpoints
of the banks supplying concentrate are manipulated in order to
obtain the desired tailings flowrate.

Controlled
Variable:

Tailings Flowrate

Measured Values:

Flow Rate, or Inferred Flow Rate

Manipulated
Variables:

Air & Level Setpoints of Banks

Supplying Concentrate

Additional features and benefits:

25
20
15
10
5
0

0 1 3 4 6 7 8 1011

14

Time (hours)

Flow Controller

PID

FloatStar
Level
Stabiliser
only

30

Inferred Flow (%)

However, when the Flow Optimiser is switched on, the tailings flow
is rapidly and tightly controlled to setpoint. In this case, the Flow
Optimiser manipulated both the aeration rates and level setpoints
of relevant Rougher and Cleaner banks.

The relative mass pull contributions from the different banks

can be specified.
Operation of the circuit is steered away from bottlenecks
(e.g. valves saturating).
Built-in safety loops prevent limits from being violated.

The FloatStar Flow Optimiser has been successfully implemented on

numerous flotation circuits with varying configurations and degrees
of instrumentation and automation. The fast level setpoint tracking
that can be achieved by the FloatStar Level Stabiliser further
enhances the performance of the optimisation strategy.

FloatStar
Flow
Optimiser
with
Level
Stabiliser

17

19

22

25

Grade-Recovery Optimiser

Typically, flotation circuits aim to produce a concentrate grade of

a specified quality, while recovering as much of the valuable mineral
as possible. The behaviour of the circuit (ultimately the final grade
and recovery) will be influenced by, amongst others, the:

90
80

Grade and Recovery (%)

With the increase in availability of online grade analysis, it is now

possible to manipulate the flotation circuit operating conditions to
optimise plant performance. Developed as a result of these
advancements, the FloatStar Grade-Recovery Optimiser provides
continuous, online optimisation of circuit operation throughout
fluctuating plant conditions.

100

70
50
40
30
20

Concentrate grade is
sacrificed

10
0

Level setpoints.
Aeration rates.
Bank residence times.
Reagent addition.

Choosing the correct values for these variables will ensure that the
best recovery will be obtained for a specified grade.

30

In most cases, the Grade-Recovery Optimiser is configured to prioritise

recovery while attempting to ensure a consistent product grade. In
the simulated scenario (top graph on the right), at one point the
recovery has dropped below the minimum. The Grade-Recovery
Optimiser then adjusts the circuit operation to increase the recovery.
As a consequence, the concentrate grade decreases.

20

4 5 6
Time (hours)

10

Rougher 1+2 On-Off Data

Grade (%)

25

15
10
5
0
60

The graph on the right (actual plant data) illustrates the following
benefits of the Grade-Recovery Optimiser:

Recovery has dropped

below the minimum

60

70

80

90

100

Recovery (%)

Reduction in the variation of the grade (y-axis).

Increased recovery (x-axis).

Off

Since it is in touch with the entire circuit, it is able to constantly and

consistently optimise the full spectrum of variables to ensure peak
circuit performance at all times.

LIC

On

LIC

LIC

LIC

Gt

Gf

LIC

LIC

LIC

LIC

Gc

Gt

The figure to the right shows how the Grade-Recovery Optimiser

would be installed on the Rougher section of a typical circuit.
Grade Recovery Optimiser

Case study

Control Rougher concentrate grade to setpoint.

Maximise Rougher recovery.

100
90

140

80

120

70

100

60

80

50
40

60

30

40

20

20
0

10
Grade

6
Time (hours)

Grade SV

10

0
12

Grade and Recovery (%)

The objectives of the optimiser on this section of the circuit were as

follows:

Level (% - value shifted for clarity)

The graph on the right shows the typical effect of the optimiser on
the level setpoints of the circuit. The levels being manipulated are
the three lines of four rougher banks each. These banks were
self-aspirating, and hence the aeration rate was not available for
manipulation.

Recovery Minimum Recovery

160

A benefit analysis was conducted on a copper plant using the

Grade-Recovery Optimiser. On and Off data were collected over
several months, and the difference in performance was assessed.
Statistical techniques were used to verify the results. The values were
found to be valid with a 95% statistical confidence.
The table below shows the results of this analysis:

Normal PID Control

(mean)

Grade-Recovery
Optimiser (mean)

Difference

Recovery

86.36%

88.35%

1.99% improvement

Feed Grade

1.632%

1.634%

Statistically equal

Concentrate Grade

37.51%

37.32%

Statistically equal

Tailings Grade

0.226%

0.197%

0.029% decrease

Reagents play an important role in froth flotation. They strongly

affect the performance of the circuit and contribute considerably
to the circuit operating costs. The selection of the reagent suite is
a complex task, with multiple, often counterintuitive, rules and
calculations.
The Reagent Optimiser aims to automate this process, to provide
consistent round the clock adjustments to the reagent suite tailored
to the plant operating conditions.
The figure on the right is a graphical representation of a fuzzy logic
calculation that forms part of the Reagent Optimiser. In this case,
the plant concentrate and tailings grade readings are used to
make decisions regarding adjustments to one of the plant reagents.
Features:

Incorporates fuzzy logic.

Uses plant feedback.
Nonlinear, multi-input, multi-output controller.
Functions cooperatively with operators and metallurgists.
Customisable for additional calculations and logic.
Detection of, and reaction to, invalid grade measurements.
Automatic adjustment to feed tonnage changes.
Plant-specific switch-on logic.
Complements the FloatStar Grade-Recovery Optimiser.

Benefits:

Consistent changes to reagents.

Efficient usage of reagents through continuous analysis of
circuit requirements, thus decreasing reagent usage.
Maximises recovery by preventing unnecessary losses due to
reagent under-dosing.

The Reagent Optimiser can be combined as a comprehensive

solution with the Flow and/or Grade-Recovery Optimisers to ensure
optimal plant performance.

Reagent

Reagent Optimiser

Tai

ling

sG

rad

Co

ade

e Gr

trat
ncen

PROCESS MONITORING

Performance Monitor
Controllers must not only perform well in the conditions under which
they were installed and tuned, but must also be robust through
fluctuating plant conditions.
The FloatStar Performance Monitor keeps track of controller
performance at all times, and produces useful diagnostic information
to analyse when and why controllers are not performing at their peak.
Both plant personnel and control engineers will find this information
useful in ensuring that their controllers are always performing well.
Currently the Performance Monitor:

Provides a comparison of the performance of the flotation

circuit when the controller is on and off.
Compares the current performance to a defined Optimal
Performance.
Can be configured to reject specific invalid operational
periods (e.g. very low flows, saturated actuators).
Determines periods of poor performance.
Provides diagnostic information for determining faults.

Level Fault Detector

Examples of Spikes

The Level Fault Detector has several fault-detection routines, and a

general statistics and performance-monitoring package. The fault
detection routines include a frozen signal detector, a signal-spike
detector and an overflow detector. The statistics module provides
information on a moving window of data, including minimum value,
maximum value, average, standard deviation and instrument
resolution.

Measurement

The FloatStar Level Fault Detector is an online module that detects

level-related faults in sumps and flotation banks. It provides useful
feedback regarding flotation and general level-instrument
performance.

Time

A graphical interface also forms part of the fault detector package.

This interface has been designed to give the users intuitive feedback
regarding faults on the plant, without overloading them with too
much information or alarms.

Example of Frozen Signals

Measurement

The figures on the right are examples of spiking and frozen signals.

Time

ADDITIONAL STABILISATION

pH Control
The control of pH is required in a vast number of areas in the mining
industry. A key area is in flotation reagent control systems. There are
a number of difficulties in controlling pH:

Non-linearity of the pH curve;

Silting up of the base (typically CaOH) feed -valve.
Compensating for changing feed conditions.

Setpoint
pH Controller
CaOH

pH

F
Feed

The FloatStar pH controller is designed to handle all of these

control difficulties and consists of:

An advanced multi-variable algorithm, including feedforward

and feedback compensation.
Non-linear compensation.
Fault detection algorithms.

Mixing Tank

The above techniques have been combined into a pH controller

that is simple to set up, and provides robust control in an industrial
environment.

SIMULATORS

Dynamic Flotation Simulator

The FloatStar Dynamic Simulator is a tool for modelling flotation circuits.
It incorporates a mechanistic model that takes into account the
dynamics of both true flotation and particle entrainment.
The simulator has two modes of operation: True dynamic mode and
steady-state mode. In dynamic mode the simulator mimics an actual
plant, with dynamic responses to changes in plant operating
conditions. The steady-state mode is more useful for testing different
plant designs and configurations.
The dynamic mode allows the user to:

Change flowrates, valve positions and other dynamic plant

properties during run-time.
Predict the process response due to changes in head grade,
flowrates, level setpoints and air setpoints.
Prototype control and optimiser systems for an actual plant.
Train operators using SCADA-like frontends.

In the steady-state mode the user can:

Change steady-state variables such as flotation bank areas

and circulating loads.
Trial different plant designs.
Optimise steady-state variables such as residence times.

Actual Plant

Simulator

Website: www.mintek.co.za/floatstar.html
Contact Details:
e-mail: [email protected] Tel: + 27 (0)11 709 4379

A global leader in mineral and metallurgical innovation