Escolar Documentos
Profissional Documentos
Cultura Documentos
1473-1486, 1994
Pergamon
0892--6875(94)00085-9
ABSTRACT
This paper reviews the cohmm flotation practice in the largest Chilean copper-moly
concentrators where this process unit has been increasingly chosen to replace multi stage
cleaning circuits in new and existing operations. Process flowsheets incorporating columns
are presented fi~r selected concentrators. Procedures for sealing up, attd process control
are also illustrated.
Keywords
Flotation columns, flotation circuits, processing costs
INTRODUCTION
The Chilean mineral industry has always been ready to accept novel and alternative process technology;
it suffices to mention the large impact that semi-autogenous (SAG) milling has had since the early 80's
and the on-going full scale testing of innovative hyperbaric-filtering technologies. In the last ten years
colunm flotation has been increasingly accepted in the process circuits of the Chilean concentrators [1]
and it may be expected that in the near future on going R&D will contribute to make this novel technology
suitable to more applications. Certainly the low price of copper in 1993 (75 cents/lb, for high grade
copper) has caused a short term general decline in re-investments and in the expenditure for pilot plant
testing and R&D but has stressed the need to contain production costs. At the moment economy of scale
and enhanced processing technology can contribute to better the economic performance of copper
operations. A number of concentrators are considering capacity expansion (Andina & Disputada) and
enhanced metallurgical recuperation projects have also been implemented (Andina, PARM) [2]. Column
flotation can contribute both to cost saving and to increased revenues by improving the metallurgical
performance of concentration circuits. Most of the column flotation experience in Chile is restricted to
Cu-Mo bulk concentration and Cu/Mo separation. Table 1 summarizes colunm tlotation installations to
date. In the copper industry the level of knowledge and experience is such that today columns can be
considered in flotation flowsheets with increased confidence.
1473
CHUQU1CAMATA/
CODELCO
ESCONDIDA/
RTZ - BHP
EL SALVADOR/
CODELCO
EL TENIENTE/
CODELCO
ANDINA/
CODELCO
Rencagua
il Region Antifag-gta
Escondida Antofagasta
La
concentrator
Chuqui concentrator
as above
cleaning
m 1
Cu/Mo sep.
cleaning in the
1
Chalcopyrite
Cu-Mo
bulk
(0.9% Cu ca.)
flotation
cleaning in
2
Chalcopyrite
Cu/Mo sep.
i (1.3% Cu ca.)
8
Calcosina and cleaning in the
Cu-Mo
bulk
covelina
(2.12% Cu ca.) flotation
Chalcopyrite
(I .13%Cu ca.)
VI
Region
Chalcopyrite
(I.25- 1.3% Cu)
V RegionLos Andes
El
Salvador I11 Region
concentrator
Andina
grindingflotation plant
(3000m.a.s.I.)
Coletonas
concentrator
Mo circuit
as above
Los Bronces-
Number
of column
units
bulk cleaning:
iI
3rd and sole
cleaner
cleaning in the
2
Cu-Mo
bulk
flotation
1
cleaning un
Cu/Mo sep.
as above
as above
as above
cleaning in the
Cu-Mo
bulk
flotation
as above
DISPUTADM
EXXON-CHILE
Application
Company
Mineral/s
Chilean
region & main >rocessed
(main species)
city
16
14.3x4.0x4.0
industrial
industrial
3.5
pilot
pilot
ind~hlal
13.6xl .83xl.83
0.81
0.71
13
pilot
229
179.4
industrial
pilot
pilot
Indtts;, ial
Indu.shial/
pilot plant
10x0.76 (30")
13.3x0.9x0.9
15x0.9
12
Ixl.08
6.5
13.85x3.71xl.76
13.8x2x6.5
0.65
12x0.91(36")
Total
section
area
[m2]
16
0.17
N. Buff. Total
Ix w
volume
[m]x[m] ~ unit
[m31
16
232
13.5x0.46 (18")
14.5 x2x8
Column size
HxLxW or
HxD [m.l
TABLE 1 Some industrial and pilot plant applications of flotation columns in Chile.
Cominco
and
Minnovex
USBM/CII
Turbo-Air
1992-
19901991-
+
+
as
above
+
1989+
19871991
1991-
f.cloth and
USBM/CII
as
above
19851987
Yeas
of start
up and
end
1992-
filt~r cloth
USBM/CII
Tmbo-Air
h,+.-ffece
level
USBM/CII
Turbo-Air +
Minnovex
Bias
+/-
~;ira cloth
Type og
control
strategy
Type of
bubble
gene~or
.>
>
U'3
.o
.-.d
4~
1475
Flowsheet
The concentrator is located at 800 m. above see level near Colina and one hour by car from Santiago.
It is fed through a 56 km. long pipeline with the product of the Los Bronces grinding plant at 3600 m.
above sea level (m.a.s.1.). The feed (37000 tpd; 80% -65 mesh or (212 am), 1.2% Cu) is mainly
chalcopyrite (Cu-34.6 %) but chalcocite (Cu-79.8%) is sometimes present. It is treated in two parallel
flotation lines with a common 3.8 x 8.2 m. (12.5 x 27 ft.) regrinding ball mill for the combined rougher
concentrates (Figure 1). The rougher stage consists of two parallel sections of nine 85 rrra (3000 cu. ft.)
Wemco cells. The rougher concentrate is ground to 80% -325 mesh (45/zm) in a ball n'fill circuit closed
by ten 0.50 m. (20 in.) diameter Vulco cyclones. The overflow joined to the scavenger concentrate is fed
to two parallel flotation colunms with cross section 16 m 2 and 14.5 m. high. The colunm Cu-Mo
concentrate is de-watered and filtered in two 48 m2 hyperbaric Andritz filters that produce a cake
humidity of 8%. The scavenger stage consists of two parallel sections of seven 42.5 m3 (1500 cu. ft.)
Wemco cells each.
_ =-=_-=
~ ~
'~ ~ ', z~-~ ' " t ~ s ~
~ o , ,
. t ~ ~ / . .
' -I
%ter~,~, I
[LasTrtlas'regrindingandl
I
flotation plant
[
~OU,~,..OT,,~O~,~L,
L _
"
L~
,
WATER
F W:-LOO
REAGENTS I I
i
~
I~l
~ WATER
1476
The two columns are equipped with two different external bubble generation systems. One unit uses the
USBM/CII Turbo-Air device. The other unit uses a bubble generator marketed by Minnovex-Canada. The
two bubble generation systems allow comparable metallurgical results. They are tested in parallel to
compare their wearing performances. The Minnovex bubble generator device made from steel lasted a
few days, a carbon-tungsten steel device has worked for several weeks without need of maintenance.
Water addition to sparger air for bubble generation is kept at minimum values and frother is not required.
The colunm feed and products flow rates are measured on-line. Wash water and air flows are also
measured. Pulp-froth interface level is calculated from multiple pressure measurements. The column
control system is used to keep the interface level within suitable limits by adjusting the tailing discharge
valve. However, on line measurements can be used to adjust the washing water addition as well. A
display in the control room allows visual control of the top of the column.
The new 36000 tpd Las Tortolas flotation circuit was conceived and designed on the basis of the industrial
and colunm pilot plant experiences in the previous Los Bronces-San Francisco concentrator (3600 m.
above sea level). The earlier San Francisco conventional circuit (Figure 2) was successively modified to
a configuration similar to that of Las Tortolas. The initial treatment capacity of San Francisco was 4800
tpd (1978), rising to 8400 tpd in 1981 and was 12000 tpd in 1988. The plant was put out of service and
finally substituted by Las Tortolas in 1991. One 0.45 m. (18") and one 0.90 m. (36") flotation columns
have been tested at San Francisco since 1985. Based on the finding from pilot testing in 1987 a full scale
3.91 x 1.76x 13.85 m. colunm was put in service as third cleaner (Figure 3). Later the cleaning section was
fully converted to columns - that is replacing three stages of cleaning with one unit (Figure 4). The
experiences are thoroughly reported by Villanueva and Vergara [3]. The flowsheet of Figure 1 used today
at Las Tortolas and sketched in Figure 5 is indeed a further evolution of the Los Bronces flowsheet in
Figure 4.
FEED
TAIL
I REGRINDING
iI
t
FINALCONCENTRATE
Fig.2 San Francisco flotation plant - earlier conventional circuit (1978-87)
Future developments
A flotation plant for the treatment of the bulk concentrate for copper/naolybdenuna separation is under
construction and will be operating early in 1994. The new flowsheet incorporates column flotation and
is shown in Figure 6.
Currently the expansion of the treatment capacity from 36000 to 50000 tpd is also being investigated.
1477
FEED
TAIL
RF.GRINDING
II
FINAL CONCENTRATE
Fig.3
FEED
TAIL
REGRINDING
FINAL CONCENTRATE
Fig.4 Evolution of San Francisco flotation plant - column cleaning circuit (1988-92)
1478
G. SCHENAand A. CASALI
FEED
'rAIL
REGRINDING
FINAL C O N C E N T R A T E
Fig.5 Further hypothetical evolution from San Francisco flotation plant to that in use at Las Tortolas.
u~Mo
CONC. Cu
CONC. Mo
Andina concen[rator
Flowsheet
The operation is located in Los Andes 120 Km from Santiago leading north. The grinding-flotation plant
is in a cave at 3000 m. a.s.1.. The concentrator can be reached by tbur-wheel drive in spring and summer
but in winter with severe climate conditions the access is possible only by snow-mobiles. The mill treats
C o l u m n flotation circuits
1479
33500 tpd of ore mineralised by chalcopyrite. Average feed grades vary between 1.25-1.3 % Cu. The
feed is treated in three parallel grinding circuits. Each grinding line has one 3.7 x 4.9 m. (12 x 16 ft.)
Marcy rod mill and the discharge is split in three secondary 3.8 x 4.9 m. (12.5 x 16 ft.) ball mills in
close circuit. The rod mill discharge feeds three batteries of cyclones each consisting of three 0.4 m.
cyclones. The underflow of a battery of three cyclones feeds one ball mill. The 80% - 270/am (60 mesh)
overflow of the nine cyclones goes to five rougher flotation lines each with nine cells. There are 4 parallel
lines with 90utokumpu-36 (1350 of.) cells and 1 line and with 9 -42.5 m 3 (1500 cf.) Wemco cells. The
combined rougher concentrate is reground in closed circuit with one 2.75 x 3.96 m. (9 x 13 ft.) and seven
0.4 m (15 in.) cyclones. The overflow, along with the re-ground cleaner-scavenger concentrate, feeds the
two parallel rectangular 6.5 x 2 m. columns 13.8 m. high. (Figure 7). The scavenger is a bank of 13
Outokumpu-36 cells. The re-grinding of the scavenger concentrate is in open circuit. The bulk concentrate
is treated in a copper moly separation circuit. Copper is depressed and molybdenite floated. The Me
concentrate has 49 % Me and less than 3 % Cu. The separation is carried out by roughing and six cleaning
stage in counter-current sections.
FEED
TAlL
REGRINDING
FINAL CONCENTRATE
Future developments
A major expansion, almost doubling the actual treatment capacity, has just received approval (November
1993); treatment capacity will increase from 32000 to 64000 tpd. Column flotation has been inclt, ded in
G. SCH]ENAand A. CASALI
1480
the new Cu-Mo bulk flotation plant and also in the Cu/Mo separation plant. Here roughing and cleaning
stages will use columns, and mechanical cells will be used in the scavenger section. A probable flowsheet
of the integrated process is shown in Figure 8.
TABLE 2 Tentative mass balance around the columns operating at Andina
solid tph
Cu %
F.ed
1400.00
1.28
Me %
solid %
pulp inch
solid s g
Iv ~
R.t.ill
0,02
036
2947.20
2,80
Rconcentrates
346,00
4.60
O,O9
28,98
967,24
2,90
watar
c..... ,,,,e
3,5
o,4o
solid tph
Cu %
Me %
solid %
pulp inch
solid s g
td % -
~,,..3
~0-~'q'
0,44
29.23
134,03
4,20
COLUMN
CLEANER
I ~ S C
4,96
0,55
19,42
2279,55
Cu %
Me %
solid %
pulp inch
solid a g
rougher
Yield
Cu recover'/
Me recovery
24,71
88.82
87.20
column
9,86
60.92
7,54
460,50
2.15
0.59
O.19
2145.52
3,00
scavenger
35.81
95.01
97.60
F "
3,60
h= T.il.
Iv
1349,60
0,19
T
IS
.tails
|
=
II
5,70
1,61
26,33
507,19
3.20
avenger feed
solid tph
0,004
39,87
1979,96
2.7
164.80 ~
,t'.,.d ....
I
I
t
1084
O.19013283
295,60
O,17
0,02
0,01
28,43
3890.00
2,70
14.o8
'
b,~[SCAVENGER =did% I~ ~
lY
overall
3.60
86,06
67,33
FEEl
TAIL
>
CONCCu
l,
BULKCu-Mo]
CONC.Mo
1481
From rougher
Wash water
Wa#hwatcr
WaJh water
To rougher
Mo Cone,
Air
Air
1482
Canc. Cu
Escondida concentrator
Tile processing plant was design for a nominal capacity of 35000 tpd. The initial reserve exploited should
have an average copper grade 2.12 % and total reserves 1.6 %. The projected life is 52 years. However
the average feed grade for the first years is 2.85 % Cu. Concentrate production is 760000 tpy with
320000 tpy of copper (42% Cu). Moly plant is not in operation and its activation is not expected in the
short run due to low Mo grade of the concentrate currently produced. Grinding is carried out in two
identical lines. Each line includes a 8.5 x 4.3 m (28 x 14 fi) SAG mill and a battery of 10 Krebs D-26
cyclones. Tile underflow is re-ground in a 5.5 x 7.5 m (18 x 24 ft) ball mill and re-classified. The cyclone
overflow is pumped to a rougher flotation bank. Flotation llowsheet is shown ill Figure 11; the two
parallel rougher tlotatiort banks include 40 (2 x 20) 44 m3 (1500 cf.) Dorr Oliver cells and are designed
to provide 22 minutes retention time with a pull~ solid content of 30%. The rougher COllcentrate of each
1483
line is classified in two parallel batteries of ten Krebs D-15 cyclones and the underflow re-ground in 4.3
x 8.1 (14 x 26.5 ft) ball mills. The overflow is fed to the cleaning columns. A total of eight 4 x 4 x 14.3
columns provide a cleaning retention time of 20 minutes. The column tails are re-treated in a scavenger
bank of 20 44 m 3 Dorr Oliver flotation cells; the tailings are de-watered and finally rejected and the
concentrate re-classified in the D-15 cycloning batteries. Column concentrates are de-watered in two 52
m diameter thickeners to 65 % solids and delivered by pipeline to the Coloso filtering-storage buildings.
The expansion to a nominal capacity of 55000 tpd by installing a third processing line is under evaluation.
FEED
TAIL
~GR~D~G
FINAL CONCENTRATE
),
COLUMN F L O T A T I O N P R A C T I C E
Scaling up
Scaling up procedures are largely based on experiences in pilot plants. Full scale equipment is scaled-up
by the use of carrying capacities measured in pilot plants and matching ranges of superficial velocities.
However the height of all the columns operating in Chile is between 12-14 m.
Table 3 reports typical values used in copper cleaning. The data were collected from bibliograplly [4,5,6]
and enquiries during plant visits.
Bubble generalors and energy requirement
The Minnovex, the Cominco and the Turbo-Air USBM/CI[ external bubble generators are in use in Chile.
As an example, in a typical copper cleaning operation two 16 m 2 cohmms consume 1200-1500 m3/h of
air at standard - i.e. atmospheric- conditions. The compressed air at 480 kPa (70 psi) of the air network
is ted to the cohmm at 290 kPa (45 psi). Each column has 16 inlet pipes. As a rule of thumb 50 m3/h of
air in standard conditions per square meter of colunm section are necessary. Power requirement for the
application is in the order of 0.1 kWh/m 3 of air in standard conditions, i.e. 1 bar. No applications of the
Microcel bubble generation systems are reported in Chile. With such a system the energy requirement is
higher because about 0.3 kWh per m 3 of feed pulp are probably needed by the re-circulating puml~.
1484
G. SCHENAand A. CASALI
Power consumption is essentially application specific: increasing with decrease in the mineral size and
depending on the recovery. There are experimental indications that air flow rate permits movements along
the recovery-grade curve of the column.
TABLE 3 Typical values of key operating variables and geometry for tlotation columns.
min.- max.
Carrying capacity
(tph. m-Z solid in conc.)
Carrying capacity
(tph .m-2 Cu in conc.)
Treatment capacity
(tph. m-2 solid feed )
Bias : (T-F)/F
Bias ratio : T/F
Superficial velocities
(cm.s-l) :
washing water
air or gas
bias
pulp
1.5-2.5
0.5-1.2
5-10
.10-.20
1.1-1.2
(T-F)/F =T/F-1
flowrate/section area
0.2-0.5
2.0-3.0
0-0.3
1.0-1.5
7-9
0.6-1.5
14-18
19-22
Geometrical ratios:
Overall height / overall width o r 2-3.5
diameter ratio
height / width ratio
10-15
estimated figures
Column flotationcircuits
1485
density in the pulp and froth zones for the calculation of the interface depth. Floating bobs or bubblingpressure measurement systems are also used. Sonar technology is in use at Escondida,
In many cases positive bias assures better metallurgical performance. The higher levels of control are
based on the bias control. This is achieved either by maintaining a constant positive difference between
tailing flowrate and feed flowrate or by ensuring a constant ratio ( > 1) between the two flowrates.
WHh water
F,,d
kml
Co.~
Air
Tail
SET
LEVEL
a)
b)
c)
d)
Fig. 12 Control strategies : a) interface level with discharge flowrate - fix washing water; b) bias with
flowrate difference and level with washing water addition; c) bias with flowrate ratio and level with
washing water; d) bias with water addition and level with discharge flowrate
Legend : FC: flow controller; FDC: flow difference controller; FRC: flow ratio controller; FI: flow
indicator; FT: flow transmitter; LT: level transmitter; LC: level controller; BC: bias controller
In this class of strategies different possibilities exist. One method is to set a bias value as the difference
in the measured feed and tail flowrates and maintain the (positive) setting adjusting the bottom tailing
valve. An independent control loop operates the washing water addition valve in order to keep a constant
interface level ( Figure 12 - b). The method can be improved by setting the bias as a ratio greater than
one between the measured tailing and feed flowrates (Figure 12 - c). This second strategy is more suitable
for variable feed flowrates.
Another strategy in use controls the interface level that is kept constant by operating the tailing valve and
the bias by adjusting the washing water addition (Figure 12 - d). Therefore the measures of feed and tail
flow rates are used to modulate the water addition in order to keep a given bias setting; the tailing valve
operates to maintain the interface level. The tailing discharge system is regulated by a pinch valve.
All the control strategies reported are used in Chile; however in a number of plants the strategies based
on bias originally implemented were later modified and the simpler control of level is now adopted. It
seelns that to date available instrumentation does not offer the precision sufficient for successful colunm
control. Optimal control is not yet implemented, but is planned.
Research at tile Department of Chemical Engineering of the Technical University of Valparaiso is devoted
to column flotation variable identification and control. Alternative methods for the precise measurement
of the level interface position based on temperature and electrical conductivity profiles obtained along the
colunm are being investigated. The research group works close association with eolunm operators at El
Teniente. These novel methods will permit the estimation ofinterface levels and bias with a precision
superior to that allowed currently by tile use of available instrumentation. This will permit better
1486
G. SCHENAand A. CASALI
performance of the existing control systems and a step forward for the development of optimal control
strategy algorithms that require an improved knowledge of the relationships between variables and
metallurgical results.
CONCLUSIONS
Colunm flotation teclmology has significantly contributed to the simplification of the copper flotation
circuits, allowing a decrease in the number of units and the number of stages, while maintaining or
bettering the overall circuit metallurgical performance. They have also improved the economics of the
concentrators, column technology being less costly than conventional flotation. However, to date colunms
have yet to demonstrate their capability to fully substitute for conventional cells in all applications. Indeed
in modern copper circuits cells must still be used in the rougher section where coarse material is treated
and in tl'te scavenging section of the cleaning flowsheets. There are also important indirect benefits related
to the use of columns such as those related to the automatic control of the circuit. However, the lack of
fundamental knowledge of this novel technology does not allow scaling up of plants with confidence on
the basis of laboratory tests and it is wise to undertake pilot plant tests. Although Chile has much expertise
with sulphide copper ores there is very little experience with other minerals. However experiences
continuously arising from new industrial applications and R&D will contribute to a better understanding
of the technology.
ACKNOWLEDGEMENTS
This research was financially supported by CEC-IMPEXFLOTCOL Brite-Euram project.
The authors are grateful for the assistance to Mr. B. Soto and Mr. A. Cristiansen (Las Tortolas),
Mr. P. Molinet and Mr. G.Von Borries (Andina), Mr. D. Riveros (El Teniente), Mr. C. Barahona (C.M.
Disputada de Las Condes), Mr. J. Valdivieso (Chuquicamata), Prof. J. Yianatos (Universidad T6cnica
Federico Santa Maria).
ACRONYMS
CODELCO
USBM
CII
REFERENCES
I.
2.
3.
4.
5.
6.
Toro, 1t., Flotacion colmnn.'lr en Chile VI S3,mposium ARMCO, Vifia del Mar, Chile, (1990).
Anon, P.A.R.M. Proyecto aumento de recuperacion met.'durgica division And ina - Codelco Chile
Vill.'mueva, F. & Vergara, M.A., Desarrollo de la tlotacion columnar en el concentrador de los
Bronces - CMD. Minerales, 54, 191.
Yianatos, J.B. & Finch, J.A., Condiciones tipicas de disedno y operaciones de colunmas de
flotacion industrial. Mineria Chilena, Informe expecial, No. 74, (May 1987).
Yianatos, J.B. & Murdock, D.J., Nuevos avances en la tecnolo~ia de columnas de llotacion.
Mineria Chilena, N. 125, (Nov. 1991).
Bergh, L.G. & Yianatos, J.B., Control alternatives for llotation columns Minerals Engineering,
6, 6, (1993)