A dynamic model for chloride control in SX plants

J.M. Menacho, L.E. Gutiérrez and Y.I. Zivkovic

De Re Metallica Ingeniería
Av del Valle Nº 601, Of. 31
Santiago, Chile
drm@drm.cl

ABSTRACT

Chloride and other impurities carried over into the electrolyte during
electrowinning is the most frequent operational problem in SX plants. This problem is
faced in practice based on experience and operator skill, but up until now, there has been
no tool which allows the study of these situations to adopt optimal and opportune
solutions. This paper presents a phenomenological model to predict levels of entrainment
and mass balance for chloride and other impurities, under a variety of design and
operational conditions. The model performance and usefulness is illustrated with several
study cases.
 FRAMEWORK

The most frequent operational problem in copper SX plants is the carryover of

impurities from the PLS to the electrolyte solution. This occurs mostly via aqueous in
organic entrainment, and chloride is the most common contaminant, although iron,
manganese and nitrate are also problematic.

Chloride carryover has strong negative impacts, both technical and economical,
such as:

• Net loss of cathode production

• Pitting corrosion in permanent cathode sheets
• Enhanced anode corrosion
• Cobalt loss due to electrolyte bleeding
• Air pollution in the tank house due to chlorine gas evolution

High chloride carryover is related to high aqueous entrainment in the loaded

organic solution, variable from plant to plant and also variable over time due to weather
changes. Plant disturbances produce a sudden increase in entrainment, whose magnitude
depends on actual design and operational conditions.

Lack of critical devices or under-dimensioned units or inappropriate equipment

for specific applications are among the controlling design aspects.

On the other hand, organic contamination with polar or tenso-active chemical

components is a frequent operational problem. Such contamination can be exogenous or
internally generated by degradation of components in the organic phase. Another
operational cause of high entrainment is the operation out of safe ranges. A typical
example is the width of the organic band in the settlers, which often is below the
recommended value to “save extractant consumption”. This is an expensive save as
demonstrated through a cost/benefit balance.

This investigation originated from the lack of a predictive tool which allows a
priori analysis of different design and operational scenarios in SX plants. In order to
predict expected results from programmed changes in the plant, in terms of magnitude of
the entrainments and chloride level in the electrolyte.

The model here presented is useful in stages of design of a new plant as well as to
estimate maximum capacity of an existing plant or to evaluate and select alternative
devices to reduce entrainment in a given plant and also to decide what to do just when a
sharp increase in entrainment is occurring.
 THE DYNAMIC MODEL (1, 2)

Structure

Entrance and exit chloride streams as well as chloride inventory fluctuation in the
electrolyte are presented in Figure 1. LO and UO are loaded organic and unloaded
organic, respectively.

m1out = Chloride in
Electrolyte
in Bleed
m1 = Chloride out
m2 = Gasified
in LO
Chlorine
Electrolyte
out
m3 = Chloride in
m2in = Chloride in
Cathodes
Water
Make Up m4
out
= Chloride
in UO

Chloride Inventory Change

Figure 1 – Chloride Balance in the Electrolyte

A differential mass balance is:

d [Cl − ] 2 in 4 out
Ve = ∑ mk − ∑ mk (1a)
dt k =1 k =1

where:

Ve = volume of the electrolyte inventory, m3

[Cl–] = mean Cl– concentration in the electrolyte, ppm
t = time of operation, h
mkin = kth Cl– mass flow entering the electrolyte, g/h
mkout = kth Cl– mass flow exiting the electrolyte, g/h.

Transient characterization of entrances and exits allows one to estimate the Cl–
level in the electrolyte from equation 1a, as a function of time. Figures 2 and 3 compare
response of the model with actual data for three different industrial SX plants. Agreement
is quite acceptable in all cases.
 70
Chloride in Electrolyte, ppm ...

60

50

40

30

20
06-01-02

06-05-02

06-09-02

06-13-02

06-17-02

06-21-02

06-25-02

06-29-02

07-03-02

07-07-02

07-11-02
07-15-02

07-19-02

07-23-02

07-27-02

07-31-02

08-04-02
Date
Simulated Experimental

Figure 2 – Experimental and Simulated Cl– in Plant 1

90
Chloride Electrolyte, ppm..

80
70
60
50
40
30
20
12-15-01

01-04-02

01-24-02

02-13-02

03-05-02

03-25-02

04-14-02

05-04-02

05-24-02

06-13-02

07-03-02

Date

Simulated Experimental

Figure 3 – Experimental and Simulated Cl– in Plant 2

The model outlined above is predictive in character, because the solution of the
mass balance through equation 1a involves the use of physical models for the different
process stages, as shown below.
Mixer Model

A hydrodynamic model is included to predict droplet size distribution exiting the

mixing stage as a function of the following variables:

• Flowrates and O/A ratio

• Density and viscosity of the two phases
• Agitation speed
• Type and diameter of impeller
• Temperature

Settler Model

A hydrodynamic model is considered to describe settler behavior. A/O

entrainment at the organic weir is computed from the entrance drop size distribution.
Predictions change according to:

• Settler geometry
• Width of the organic and aqueous bands
• Flowrates and O/A ratio
• Velocity profiles of aqueous and organic bands
• Density and viscosity of both phases
• Temperature

The effect of motor agitation speed on aqueous entrainment at the E-1 exit is
presented in Figure 4. Three curves corresponding to different extractant concentrations
are presented. Entrainment increases as the motor speed increases.

Figure 5 shows the effect of the height of the organic band on aqueous
entrainment registered at the E-1 exit. Simulations at different O/A ratios are included.
These were generated by changing only the aqueous flowrate. Simulations demonstrate
the convenience of using 30 cm to 35 cm to keep entrainment under control.

The effect of the interfacial tension on aqueous entrainment in the loaded organic
phase is illustrated in Figure 6. For an organic phase with 30 dynes/cm interfacial tension
– that is, a new organic phase – entrainments are below 1000 ppm and no significant
difference is observed whether operating at 3.6 cm/s or 5.1 cm/s linear velocity in the
organic phase. However, if the physical quality of the organic is reduced by degradation
or by contamination with polar components, such that the interfacial tension is reduced to
below 20 dynes/cm, then at 3.6 cm/s entrainment would increase from 200 ppm up to
2000 ppm, while at 5.1 cm/s it would increase from 450 ppm to about 4200 ppm. These
simulations emphasize that interfacial tension is a relevant control variable as a measure
of the physical quality of the organic and its impact on entrainment.
 4000
A/O Entrainment, ppm .. ....

3500
3000
2500
2000
1500
1000
500
0
1300 1400 1500 1600 1700 1800 1900
Agitation Motor Speed, rpm
25% Extractant 27% Extractant 29% Extractant

Figure 4 – Simulated Effect of Motor Stirring Speed on Aqueous Entrainment Measured

at the E-1 Exit at Different Extractant Concentrations.

7000
A/O Entrainment ppm :)..

6000
5000
4000
3000
2000
1000
0
5 10 15 20 25 30 35 40
O/A ratio changes Organic Band Width (cm)
according to Aqueous
Flowrate change 1.00 1.15 1.30

Figure 5 – Simulated Effect of the Organic Band Width on Aqueous Entrainment

Measured at the E-1 Exit at Different O/A Ratios
 12000
A/O Entrainment, ppm...

10000

8000

6000

4000

2000

0
10 15 20 25 30 35 40
Interfacial Tension, dynes/cm

3.6 cm/s 4.3 cm/s 5.1 cm/s

Figure 6 – Simulated Effect of Interfacial Tension on Aqueous Entrainment Measured at

the E-1 Exit at Different Linear Speeds of the Organic Phase

Figure 7 shows alternative lateral velocity profiles of the organic band in a 25 m

width by 18 m length. These were used in simulations for 1000 m3/h PLS and 28 cm free
organic height. Results indicate the strong effect of the organic channeling along the
settler on entrainment at the exit. In practice this can be produced by abnormal behavior
of picket fences.

12

10
Lateral Velocity, cm/s..

0
0 5 10 15 20 25
Distance, m
Profile 1, A/O Entrainment 4382 ppm Profile 2, A/O Entrainment 3771 ppm
Profile 3, A/O Entrainment 2200 ppm

Figure 7 – Lateral Organic Velocity Profiles Indicating Different Degrees of Channeling

Coalescer Model

Behavior of the coalescer is described through an axial dispersion model, with a

delay associated with entrainment retention by coalescence phenomena occurring over
the internal coalescer filling. The model is transient and predicts the efficiency of the
coalescer as a function of:

• Level of entrainment in the entrance

• Specific flowrate
• Mean residence time
• Operation time between successive back washing.

Behavior of coalescer model is illustrated in Figure 8 for 16 to 22 m3/h/m2

specific flowrate. Dots represent experimental values. Efficiency is around 80% along
the first two weeks, but then is quickly reduced down to 50% after two more weeks.

120

100
Efficiency, %...

80

60

40

20

0
0 5 10 15 20 25 30
Time, days
18 m/h 16, m/h 22, m/h 24, m/h Experimental

Figure 8 – Coalescer Efficiency between Successive Back Washing

Coalescence efficiency is quite sensitive to the actual level of entrainment in the

entrance organic stream as illustrated in Figure 9. Dots are again experimental values.
Mean efficiency is over 80% for entrainment in the feed at around 3500 ppm, but is
reduced to 20% at about 500 ppm entrainment in the feed.
 100
90
80
70
Efficiency, %...

60
50
40
30
20
10
0
0 5000 10000 15000 20000 25000 30000 35000

A/O Entrainment in Feed, ppm

Figure 9 – Simulated and Experimental Coalescer Efficiency as Function

of the Level of Entrainment in the Feed

Washer Model

Washer behavior is described by a hydrodynamic model, similar to that of the

production mixer-settler, but here the concept is of mixing efficiency between aqueous
entrainment and aqueous washing solution.

Figure 10 shows the effect of water solution flowrate in washing efficiency EW

defined as

 Cl − Mass in Washer Exit 

EW = 1 − −  (2a)
 Cl Mass in Washer Entrance 

In this example efficiency is assumed constant at 80% for aqueous flowrates

higher than 10 m3/h.

It is well known that washer efficiency increases as the acid content in the feed is
also increased. This can be attributed to changes in interfacial tension as illustrated in
Figure 11, which predicts an increase in washing efficiency from 70% to 80% when acid
in aqueous feed increases from 20 g/L to 30 g/L H2SO4.
 85

80
Washing Efficiency, %...

75

70

65

60

55

50
0 2 4 6 8 10 12
3
Washing Water Flowrate, m /h

Figure 10 – Washing Efficiency as Function of Fresh Aqueous Solution Feed

90
Washing Efficiency, %...

85

80

75

70

65

60
15 20 25 30 35 40
Acid in Washing Water, gpl

Figure 11 – Washing Efficiency as Function of Acidity in Washing Water

Chlorine Evolution Model

Chlorine gas evolution is represented by a kinetic electrochemical model derived

from the Butler-Volmer equation. A mixed control is assumed with a reversible charge
transfer stage and diffusion to the anode. The magnitude of chlorine evolution is
predicted as a function of:

• Chloride concentration in electrolyte

• Redox solution potential
• Cell electrochemical parameters

Figure 12 illustrates the effect of chloride concentration in the electrolyte on

chlorine evolution for 500 and 800 mV vs SCE redox potential. The effect of redox
potential is decreased as Cl– concentration is decreased.
Chlorine Gas ON-OFF Difference in

5
EW, ppm

0
25 30 35 40 45 50
Chloride in Electrolyte, ppm
500 mV/SCE 600 mV/SCE 700 mV/SCE 800 mV/SCE

Figure 12 – Evolution of Chlorine as Function of Chloride Concentration in Electrolyte

Complete Model

Study Case Nº 1: Analysis of a Programmed Change

A 2E-2S-1W circuit is employed in a SX plant and the average entrainment at the

E-1 exit is about 250 ppm. Entrainment is estimated to increase to 600 ppm due to a
given increase in flowrate. What is the expected Cl– level in the electrolyte? Which
device is convenient to introduce in order to avoid an increase in aqueous entrainment?
 Response of the model is shown graphically in Figure 13. If nothing is done, the
current 49 ppm Cl– in the electrolyte would increase to 115 ppm within two months after
the change. Such increase can be avoided by introducing a coalescer battery or a second
washing stage or both. In this specific case coalescers were recommended as the lowest
cost technical solution.

130
.
Chloride in Electrolyte, ppm..

110

90

70

50

30

10
0 10 20 30 40 50 60 70 80
Time, days
Coalescer Added Washer + Coalescer Added
Nothing Washer Added

Figure 13 – Simulated Transient Plant Response Facing a Programmed Entrainment

Increase from 250 ppm up to 600 ppm at the E-1 Exit

Study Case Nº 2: Control of High Chloride Episode

In the same plant of the previous case, the PLS solution contains 30 ppm
suspended solids instead the usual 18 ppm over a period of two weeks. Consequently,
entrainment increases from 250 ppm to 600 ppm. What operational actions are to be
taken? The model response is illustrated in Figure 14. The following alternatives are
analyzed:

• To continue as before
• To reduce flowrates
• To intensify chlorine gas evolution
• To increase electrolyte bleeding
• To increase organic band width at E-1.

In this example gasification enhancement is the most effective, followed by

increasing organic band width, then reducing the flowrate, and finally, increasing the
electrolyte bleed. In practice, a combination of these actions led to rapid normalization of
the plant.
 90
80
Chloride in Electrolyte, ppm...)

70
60
50
40
30
20
10
0
0 10 20 30 40 50 60 70
Time, days

Reference Situation Electrolyte Bleed Increase

Flowrate Reduction Organic Band Width Increase
Chlorine Gasification

Figure 14 – Simulated Effect of Different Operational Actions to Control

Chloride Level in the Electrolyte

CONCLUSION

A robust tool to control chloride and other impurities in EW electrolytes has been
developed. This tool allows the analysis of alternative design conditions, as well as
optimization in the use of existing equipment and control of transient episodes.

REFERENCES

1. J. Menacho, L. Gutiérrez and Y. Zivkovic, Chloride Balance in SX Plants, Taller

de Hidrometalurgia del Cobre 2002, Depto. de Ingeniería de Minas, Universidad
de Chile, Antofagasta, 23 – 25 October 2002.

2. J. Menacho, L. Gutiérrez and Y. Zivkovic, A Transient Model to Control

Chloride in SX Plants, Minerales, IIMCh, Vol. 57 (248), Dec. 2002, pp. 7–12.