Luis Bergh

7/30/2019 Luis Bergh

1/21

Rougher Flotation Operation Aided

by Industrial Simulator

Luis G. Bergh and Juan B. YianatosAutomation and Supervision Centre for Mineral Industry

Santa Mara University

Valparaso, Chile


2/21

30 Noviembre 2 Diciembre 2011

Content

Introduction: motivation, plant control

Develop control strategy based on Distributed models

Experimental data

Simulator design

Some results (sensitivity to level profile)

Application: Level profile selection criteria, case studies

Conclusions and further work

Air

LICLIC

LIC

Tailings

Concentrate

LIC

Feed


3/21


Typical copper flotation plant

From grinding

Rougher Circuit

Scavenger Circuit

Cleaning

Circuit

Regrinding


4/21


Rougher flotation circuit (El Teniente)

Objectives : Max recoverySubject to concentrate mass and grade

Feed characteristics:

Flow, density, solid percentage, pH,particle size distribution, grades, .

Air

LICLICLIC

Tailings

Concentrate

LIC

Feed

Manipulated variables:Level profile


5/21


A distributed model for each cell

Each cell is an independent unit with two different zones:collection and froth

Detailed phenomenological description

Considering the variation of collection flotation rates andresidence time, as well as froth transport characteristics andfroth recovery downstream the bank of cells

Air

Tailings

Concentrate

LIC

Feed

RF

RC

Feed

Concentrate

Tailings


6/21


Pulp and froth models RF

RC

Feed

Concentrate

Tailings

Adjusted mass balanceper size class

For feed, concentrate andtailings

Operating conditions: Jg, h,

Solids input: Tph, %, density

Cu and Mo grade andsolid mass (%) per sizeclass: fine, intermediateand coarse

Kinetics responses per size class

Bubble load, top of froth

Residence time distribution: E

Cell volume

Froth recovery, Collectionrecovery, kinetics constant persize class

Calibrated model


7/21


Simulation results

Rougher objectif: Max Cu Recovery subject to Cu concentrategrade is at least 5%

Find the appropriate froth depth profile (eg. 8-5-8-9 cm)

Air

LICLICLIC

Tailings

Concentrate

LIC

Feed


8/21



9/21


88

89

90

91

92

93

94

95

0 5 10 15

Recovery(%)

Change in froth depth (cm)

Bank 1 Bank 2 Bank 3

For a typical feed (800 tph, 0.98% copper grade, 40% solids

and 18% +100#) and a froth depth profile of 8-5-8-9 cm, a

94.6% recovery and 5% Cu concentrate grade were obtained.

Increasing frothdepth of one bank ata time from 8-5-8-9profile

Remember bank 1 isone cell, other banksare two cells

Air

L

IC

LI

C

L

IC

Tailings

Concentrate

LI

C

Feed


10/21


4

4,5

5

5,5

6

6,5

7

7,5

8

0 5 10 15

Concentrate

grade(%)

Change in froth depth (cm)

Bank 1 Bank 2 Bank 3

Also, the minimumfroth depth in firstbank is sometimesconstrained (absorbfeed flowratedisturbances, avoid

pulp overflow)

Higher impact of frothdepth changes onrecovery and grade arein first banks


11/21


Considering the effects and constraints, some simple logic rules

can be written:

1. Keep the froth depth in first bank as low as rejection of feedflow disturbances permit it, avoiding pulp overflow (e.g. 8 cm)

2. Keep the froth depth in the second bank as low as possible

(e.g. 5 cm)

3. Regulate the concentrate grade by increasing the froth depth

of the last bank while froth overflows

4. If the concentrate grade is below the target, then increase the

froth depth in the penultimate bank as before

5. Continue increasing the froth depth upstream until the target is

met.


12/21


Some comments:

The controller performance in the first bank and the collapse of froth in

later banks can both be detected by using cameras and image analysis.

An alternative form of setting the froth depth set points is to set the froth

discharge velocities from high to low along the banks of the circuit

(discharge velocity is a distributed property!!).

Next take a look at some simulations of a Rougher Circuit, to illustrate how

to aid operating decisions to adequately respond to some common

disturbances.


13/21


Since usually a coarser feed is produced along with

a larger tonnage, a simultaneous change in feed

PSD (from 18 to 25% +100#) and tonnage (from

700 to 800 tph) were simulated.R = 95,34 L = 5,03


14/21


Sin control: R = 93,7 L=6,4


15/21


Con controlR = 94,95

L = 4,99


16/21


The effect of decreasing the copper feed grade from 0.98

to 0.7% was evaluated. In this case, the following

variables were kept constant: feed tonnage 700 tph, solid

percentage 40%, 18% +100# in mineral feed size. R = 95,34 L = 5,03


17/21


Sin control: R = 94,8 L=3,8


18/21


Con control

R = 92,89L = 5,02


19/21


Conclusions

To advance in the holistic control of grinding flotation plants, a capability for adjusting

flotation plant operation to feed characteristics is needed.

To control the whole flotation plant, it is necessary that each circuit can be under

control, and specially the rougher circuit, since plant recovery depends heavily on this

operation.

Furthermore, the operation of cleaning and scavenger circuits will depend on thecharacteristics of the rougher concentrate.

In this sense, a simulator for rougher circuits has been developed and calibrated for a

specific flotation plant, by estimating parameters to fit collection and froth recovery

models, from a set of designed experimental data.

The simulator has been used to sensitize the effects of changing the froth depth profile

on targets.


20/21


Conclusions and further work

A set of logic rules has been proposed and successfully tested for maindisturbances coming into the plant.

Installed cameras can play an important role in on-line estimating when control

performance or froth quality may become constraints.

This information can be considered in decision systems to avoid operatingproblems and to enhance the evolution of the operation to the desired targets.

In a near future it is expected to validate these results in an industrial application,

as well as to develop similar models for regrinding, cleaning and scavenger circuits.

After then, a control strategy for the whole flotation plant can be developed.


21/21


Automation and Supervision Centre for Mineral Industry

Santa Mara University

Valparaso, Chile

Rougher Flotation Operation Aided byIndustrial Simulator

Luis Bergh

Documents

Transcript of Luis Bergh