You are on page 1of 7

IV-111

IV-110

SAG
2006
DEPARTMENT OF MINING ENGINEERING
UNIVERSITY OF BRITISH COLUMBIA
Vancouver, B. C., Canada

HPGR VERSUS SAG MILL SELECTION FOR THE LOS BRONCES


GRINDING CIRCUIT EXPANSION
By Claudio Oestreicher1 and Conor F. Spollen2
1

Senior Project Engineering Manager, Anglo American Chile, Santiago,


Chile 2Senior Process Manager, Anglo American Chile, Santiago, Chile

ABSTRACT
The selection of the grinding circuit for the Los Bronces expansion
project considered the evaluation of High Pressure Grinding Rolls
(HPGR) for tertiary crushing ahead of ball milling. Conceptual
engineering for this circuit and for a semi autogenous - ball mill with
pebble crushing (SABC) circuit were developed for an 80 000 tpd
expansion.
HPGR pilot plant and bench scale testing was conducted for equipment
sizing and wear rate determinations. In addition, bench scale testing
with the HPGR crushed product was performed in order to determine ball
mill power requirements and the metallurgical response of the HPGR
treated flotation plant feed ore. This paper presents the capital and
operating costs estimates and the metallurgical finding for these two
options.
INTRODUCTION
The Rio Blanco-Los Bronces deposit, located in the high Andes of
central Chile, 65km northeast of Santiago is amongst the largest copper
resources in the world. The eastern portion of the deposit is owned by
the Andina Division of Codelco. To the west, Minera Sur Andes, a 100%
subsidiary of Anglo American plc, owns and operates the Los Bronces
open pit mine at elevations between 3400m and 4000m above sea level.
Crushing, grinding (through a SABC circuit) and thickening of the ore are

carried out at the mine site. The ground slurry is transported in a


pipeline to the concentrator and tailings dam located at Las Tortolas,
some 56km from the mine at an elevation of 800m above sea level. In
2005, 21 million tonnes of ore, grading 1.0% Cu, was processed at a
copper recovery of 88.3%. A feasibility study to increase the Los
Bronces production will be completed in 2007. The development of the
Los Bronces mine affords the opportunity to assess the use of High
Pressure Grinding Rolls as a viable alternative to a SAG mill for ball mill
feed preparation. The high pressure grinding rolls (HPGR) is a
relatively new comminution device that offers realistic potential in
dramatically reducing comminution energy requirements. As such, it is
viewed as an alternative to the existing, less energy efficient processes
such as semi-autogenous (SAG) and ball mills.
Initial laboratory testwork undertaken at Anglo Research indicated
potentially positive opportunities for the use of HPGR technology at Los
Bronces. Further testwork was undertaken on three samples from Los
Bronces at facilities of Polysius in Germany (in a semi-industrial HPGR)
to confirm equipment sizings and final circuit selection. Anglo American
conducted a conceptual engineering exercise in 2005/6 to determine
the relative economics of using HPGRs in the Los Bronces
development project. This paper summarises the results of this test
work, confirming the optimum HPGR circuit equipment sizing and
examining the economics of the optimum HPGR circuit compared with a
SABC circuit.
SAMPLE CHARACTERISATION
Three ore samples were collected for testing in an attempt to represent
variations in ore hardness. The ore types are described in table 1.

Sample
A

Table 1: Sample Description


Rock Type
Alteration
Comment
Quartz
Quartz-sericitic
"Hard" predicted to be
Monzonite
(weak) and
< 2300 t/h in existing
moderate to strong plant
potassic
Quartz
Weak to moderate "Intermediate"
Monzonite
sericitic - presence predicted to be
Ghost breccia of secondary
equivalent to 2450 t/h
quartz
in existing plant
principally
weak quartz
"Soft" predicted to be
Western
sericitic and weak
equivalent to >2800 t/h
breccia and
chloritisation
in existing plant
Ghost breccia

IV-112

IV-113

The samples were all crushed to 100%-50mm prior to further testing at


Polysius in Germany and at CIMM and SGS-Lakefield in Santiago. The
ore characterization test results are summarized in Table 2.
Table 2: Ore characterization test results
A
B
JK Drop Weight parameters
A
71
67.2
b
0.78
0.66
Axb
55.4
44.4
ta
0.5
0.43
Bond Low Energy Impact Wi, (kWh/mt)
8.97
8.83
SPI, minutes
55.6
81.4
SPI P64, m
225
179
Bond Ball Mill Wi, (kWh/mt) (CIMM)
13.3
14.9
Bond Ball Mill Wi, (kWh/mt) (Polysius)
13.6
15.1
Bond Rod Mill Wi, (kWh/mt)
9.4
11.2
LAB Mill Index (kWh/mt) @ 100%-90m
12.1
12.9
% Cu
0.42
0.74
% Chalcopyrite
0.30
0.39
% Chalcocite
0.33
0.51
% Covellite
0.00
0.11
% Pyrite
1.67
2.22

C
63.5
0.52
33
0.42
9.46
74.5
179
14.5
14.4
13.1
12.1
1.22
1.21
0.75
0.08
2.68

There were three open circuit tests undertaken at differing operating


pressures on the samples, nominally 30, 40 and 50 bar pressure and
one closed circuit/locked cycle test using a wet 6mm screen as depicted
in Figure 1 below. The pressure setting for the REGRO unit was 40 bar
for the closed circuit/locked cycle test. Steady state operation was
achieved in three cycles.
The results from the tests on each sample are contained in Tables 3, 4
and 5. All Bond ball mill work indices are reported per metric tonne.
The results refer to the centre products from the tests.
The HPGR product size distribution varied little with the ore type tested
and with the press force applied to the rolls, indicating that the HPGR
machine appears to be capable of producing a consistent product size
distribution relative to the varying ore hardness. The circulating load
obtained from the oversize to undersize ratio after 3 cycles was steady
at 70-80% for all material types.

TESTWORK
The tests were undertaken at the Polysius Research Centre, Germany
with the Polysius Semi-industrial Unit (REGRO) to provide a basis for:
sizing of industrial rolls to match the throughput requirements;
determining the optimum grinding force to achieve a certain product
fineness;
determining the absorbed energy at the required grinding force;
simulating industrial product size distributions achievable on full
scale industrial HPGRs
The REGRO has the following dimensions:
Diameter of rolls
: 0.71 m
Width of rolls
: 0.21 m
Speed of rolls
: 0.29 m/s
Top feed size
: 16 - 35 mm
The REGRO is equipped with studded roll surfaces. The REGRO HPGR
is highly instrumented and the data is collected electronically. Polysius
use the data obtained from test work to allow the sizing of industrial scale
machines.

Figure 1: Simplified closed circuit / locked cycle test circuit for


HPGR tests at 40bar pressure

IV-114

IV-115
Table 3: HPGR Test Results for Sample A
Feed

Pressure Setting bar

Open Circuit Tests

30
40
Characterization Data

Bond Ball Mill Wi


(CIMM)
Bond Ball Mill Wi
(Polysius)
Lab Mill Index (kWh.mt)
At 100% -90m
Specific Wear Rate g/t
HPGR Throughput Rate
tph
Fresh Feed P80 mm
HPGR Feed P80 mm
HPGR Discharge P80
mm
Screen Undersize P80
mm
HPGR Power
Consumption kW
Product Flake Thickness
mm
HPGR Grinding Force
kN
Specific Press Force
N/mm2
HPGR Specific Energy
kWh/t
HPGR Specific
Throughput ts/m3h
HPGR Specific Power
kWs/m2
Screen Undersize %
Circulating Load (o/s :
u/s)
Circuit Specific Energy
kWh/t

13.3

11.2

11.4

50

Closed
Circuit
Test
40

17% for sample A and 12% for samples B and C. It is assumed that
this reduction is as a result of micro-cracking or other compressive force
effects on the samples produced by the action of the HPGR. Earlier
Table 4: HPGR Test Results for Sample B
Feed

10.8

13.6

12.9

12.1

8.8

29.5
Test Results
12.2

10.8

11.6

10.9

20.8

20.8

20.8

17.5

7.6

6.4

5.2

6.5

20.8

1.8
13.0

16.6

12.4

16.0

23.2

22.1

22.5

21.2

397

551

690

563

2.66

3.70

4.63

3.78

1.07

1.53

1.85

1.47

281

251

267

251

44.7

57.2

73.7

55.1
55.9
0.79
2.63

There is a general and consistent reduction in ball mill work indices with
increasing pressure in the open circuit samples. In the closed circuit
locked cycle test the level of reduction in Bond ball mill work index is

Pressure Setting bar

30
40
Characterization Data

Bond Ball Mill Wi


(CIMM)
Bond Ball Mill Wi
(Polysius)
Lab Mill Index (kWh/mt)
At 100% -90m
Specific Wear Rate g/t
HPGR Throughput Rate
tph
Fresh Feed P80 mm
HPGR Feed P80 mm
HPGR Discharge P80
mm
Screen Undersize P80
mm
HPGR Power
Consumption kW
Product Flake
Thickness mm
HPGR Grinding Force
kN
Specific Press Force
N/mm2
HPGR Specific Energy
kWh/t
HPGR Specific
Throughput ts/m3h
HPGR Specific Power
kWs/m2
Screen Undersize %
Circulating Load (o/s :
u/s)
Circuit Specific Energy
kWh/t

14.7

50

Closed
Circuit
Tests
40

12.6

13.1

Open Circuit Tests

13.7

13.6

15.1

14.4

12.9

10.4

23.7
Test Results
13.3

11.8

11.9

11.0

20.0

20.0

20.0

15.0

7.0

7.0

5.7

6.8

20.0

2.1
13.6

18.7

21.6

16.3

24.9

23.5

22.3

23.7

412

589

707

561

2.76

3.95

4.74

3.76

1.02

1.59

1.82

1.49

307

272

275

253

46.9

64.4

74.5

56.3
54.6
0.83
2.73

IV-116

IV-117

tests undertaken at Anglo Research (Smit, 2005) showed encouraging


flotation response of the reconstituted laboratory HPGR /ball mill
samples from a single Los Bronces ore sample (Figure 2 next page).
Table 5: HPGR Test Results for Sample C
Feed
Pressure Setting bar

30
40
Characterization Data

Bond Ball Mill Wi


(CIMM)
Bond Ball Mill Wi
(Polysius)
Lab Mill Index (kWh/mt)
At 100% -90m
Specific Wear Rate g/t
HPGR Throughput Rate
tph
Fresh Feed P80 mm
HPGR Feed P80 mm
HPGR Discharge P80
mm
Screen Undersize P80
mm
HPGR Power
Consumption kW
Product Flake Thickness
mm
HPGR Grinding Force
kN
Specific Press Force
N/mm2
HPGR Specific Energy
kWh/t
HPGR Specific
Throughput ts/m3h
HPGR Specific Power
kWs/m2
Screen Undersize %
Circulating Load (o/s :
u/s)
Circuit Specific Energy
kWh/t

14.7

50

Closed
Circuit
Test
40

12.5

12.8

Open Circuit Tests

13.6

12.7

14.4

13.3

12.1

9.9

Based on Figure 2, rougher flotation tests were undertaken on the


centre products from the semi-industrial closed circuit tests. The
objective was to assess if there was a change in flotation performance if
prepared in a conventional laboratory crusher-ball mill as opposed to
HPGR-ball mill preparation. Flotation kinetics were assessed using the
Klimpel flotation rate equation. Results are summarized in table 6.

16.0
Test Results
11.5

11.7

11.2

10.9

13.0

13.0

13.0

13.0

Figure 2: Rougher flotation response


Table 6: Summary of flotation results

13.0
6.0

5.2

5.7

Feed

8.2
2.3

11.1

18.7

18.3

16.3

21.9

23.5

21.4

23.7

389

589

660

561

2.61

3.95

4.43

3.76

0.97

1.59

1.64

1.49

266

272

258

253

38.3

64.4

63.2

56.2
57.8
0.73
2.58

Pressure Setting
bar

Open Circuit Tests


30

Closed
Circuit
Test

40

50

40

0.53
8.04
91.2
95.9
3.3

0.58
8.9
90.6
98.7
1.7

0.52
8.12
90.4
95.7
3

0.76
8.94
81.5
86.4
2.5

0.78
8.45
77.4
83.0
2.2

0.72
8.51
81.9
87.1
2.4

1.25
11.5
91.4
96.7
2.7

1.23
10.7
90.4
97.6
1.9

1.20
11.7
91.7
97.9
2.5

Sample A
Head, %Cu
Concentrate, %Cu
Cu Recovery, %
R, %
k, min-1
Head, %Cu
Concentrate, %Cu
Cu Recovery, %
R, %
k, min-1
Head, %Cu
Concentrate, %Cu
Cu Recovery, %
R, %
k, min-1

0.42
8.02
91.4
95.9
3.5
Sample B
0.74
0.75
10.05
8.81
79.5
78.7
84.1
83.3
2.8
2.6
Sample C
1.22
1.26
11.6
12.1
90.4
91.4
98.6
95.3
1.9
3.7

IV-119

IV-118
The results indicate that there is insufficient information to quantify an
eventual potential benefit in flotation, but it is concluded that there is no
negative effect of using the HPGR on the rougher grade-recovery
relationship as a result of using this technology.
CONCEPTUAL DESIGN FOR THE LOS BRONCES HPGR CIRCUIT
The preferred circuit under consideration for the use of HPGRs is in
Figure 3.

The conceptual design criteria to process 80 000 tpd additional


throughput are:
Secondary crushing circuit
Overall availability
Circuit feed size
SG of ore
Product size

=
=
=
=

85%
80%-150mm
2.8
98%-50mm

High Pressure Grinding Rolls circuit


Overall availability
=
Design screen aperture
=

93%
8mm

Ball Mill circuit


Overall availability
Design Bond Ball Mill Work Index
Design product size

96%
15.2 kWh/st
80%-180m

=
=
=

Simulations were carried out by Polysius on the HPGR circuit using a


proprietary simulator (Klymowsky and Lui, 1997) using the test work
results and scaling them to the design requirements. The main factors
that were different from the test work and the design criteria were
throughput rates, feed particle size distributions to the circuit and ore
hardness.
Initial ball mill sizing were also determined through simulations using
JKSimMet V5.1, and are based on surveys carried out at Los Bronces.
The ball mill dimensions were independently checked using reduced
recovery simulations (Barratt, 1989) for segregation of finished material
(simulated cyclone overflow) and feeding simulated cyclone underflow
(scalped product) to the ball mills. Whereas the proportion of fines (or
finished product in the screen undersize) might be viewed as being
higher than expected, it is interesting to note that the projected
operating work indices for secondary ball milling when using the
reduced recovery simulations are in the same range as those which
have been estimated by the JK SimMet model.
The resultant equipment details are:

Figure 3: Conceptual HPGR Circuit

Secondary crushing circuit


Number of screens
Screen aperture size
Screen area (per screen)
Number of crushers
Crusher type
Closed Size Setting

=
=
=
=
=
=

3
50 mm
30.1 m2
3
Standard cone
30 mm

IV-120

IV-121

Installed Power (total)


Specific energy consumption
High Pressure Grinding Rolls circuit
Screen undersize size distribution
Circulating load (% of fresh feed)
Intermediate stockpile capacity
Number of HPGR machines
Roll diameter
Roll length
Power Required (per HPGR)
Installed Power (per HPGR)
Specific energy consumption

=
=

=
=
=
=
=
=
=
=
=

2 238 kW
0.4 kWh/t

80%-2.115mm
152%
30000 t
2
2.4m
1.75m
4760 kW
5600 kW
2.66 kWh/t

OPERATING COSTS
The operating cost for the HPGR-ball mill circuit was estimated based
on the conceptual design, information gathered from the test work
performed and actual operational data from Los Bronces. The annual
operating costs are estimated to be 1.48 US$/t.
The operating costs were also compiled, on the same basis, for a SABC
circuit similar to the existing circuit at Los Bronces (Figure 5). The
operating costs are estimated to be 1.85 US$/t.
From Mine

Primary
Crusher

Ball Mill circuit


Number of ball mills
Ball mill diameter
Ball mill length
Power required (per mill)
Installed Power (per mill)
Specific energy consumption

=
=
=
=
=
=

2
7.53m
11.25m
13648 kW
13800 kW
7.87 kWh/t

Coarse
Ore
Stockpile

SAG
Mill
Screen

To thickening
and flotation

Pebble
Crusher

Hydrocyclones

Figure 4 is an example layout developed for the HPGR-ball mill circuit.


Ball Mill

Figure 5: SABC Circuit


The energy consumption per tonne of the HPGR-ball mill circuit is
calculated as 13.02 kWh/t (includes all ancillaries) compared with a
calculated value of 16.21 kWh/t for the SABC circuit
There is an estimated operating cost difference of 0.37 US$/t in favour
of the HPGR circuit, consisting of

Figure 4: Layout of HPGR Circuit

Energy savings - estimated to be lower by 0.22 US$/t, reflecting the


lower energy consumption per tonne of the HPGR-ball mill circuit at
13.02 kWh/t (including all ancillaries) compared with 16.21 kWh/t for
the SABC circuit.

Materials costs - estimated to be lower by 0.15 US$/t reflecting a


5000 hour average life for roller tyres and savings in grinding media
and liners.

IV-122

IV-123

In general, energy and media/liner savings for the HPGR-ball mill


process are approximately split 50:50 between the HPGR versus SAG
mill and the more efficient ball mill operation.

Laboratory rougher flotation tests were also completed and there was
no negative effect of using the HPGR on the rougher grade-recovery
relationship.

CAPITAL COSTS

Simulations show that two HPGRs and two ball mills would be capable
of grinding 80 000 tpd under the design conditions with a preferred
screening size between the HPGR and ball mill circuits of 8mm. The
total circuit energy consumption is estimated at 13.2 kWh/t.

Capital costs for the two alternatives were estimated with a 1:4
probability of not over-run by more than 10% the calculated values. The
estimates for the completed grinding plants were completed based on
budgetary quotations for the main equipment, material take offs and
information gather from the design and data banks. The capital costs (in
US$) are presented in table 7. The main capital cost difference of
$38.5M relates to the additional intermediate storage stockpile of 30 000
tpd and associated conveyors between the HPGR and ball mill circuits.
Table 7: Capital cost estimates (millions of US$)
HPGR
SABC
DIFFERENCE
Equipment Purchases
119.1
118.0
1.1
Civil Works & Structures
80.9
60.0
20.9
Equipment Erection
9.5
4.8
4.7
Piping
2.3
2.1
0.2
Electrical &
5.5
6.3
(0.8)
Instrumentation
Tie-ins
8.4
6.4
2.0
Construction Indirect Costs
41.0
30.6
10.4
TOTAL COST
266.7
228.2
38.5

The HPGR circuit is estimated to have operating costs of 1.48 US$/t


compared with an equivalent SABC circuit of 1.85 US$/t. This
represents a potential differential operating cost saving of US$10.8M
per annum.
The HPGR option has a higher capital cost of US$38.5M in comparison
with the SABC circuit.
The comparative economics for the HPGR-ball mill circuit with the
SABC circuit indicate a 19.6% increase in IRR and increase in
NPV(10%) of 18.6 US$M.
Based on the results of this study, Anglo American is proceeding with
the feasibility stage of the development of Los Bronces with the two
grinding circuit options. This approach will permit further refinement of
cost estimates as well as transferring some further understanding of the
HPGRs and their possible performance.

CONCLUSIONS

ACKNOWLEDGEMENTS

Following positive indications from laboratory test work undertaken at


Anglo Research on one sample from Los Bronces, further test work was
completed on three samples of differing hardness (hard, Intermediate
and soft) from Los Bronces in a semi-industrial High Pressure Grinding
Rolls to confirm equipment sizing and final circuit selection. The key
results from these tests were:
The degree of size reduction was very similar to that of other copper
ores
The HPGR product size varied little with increasing press force
applied to the rolls
The circuit specific energy consumption was 2.58 to 2.73 KWh/t of
fresh feed in the closed circuit/locked cycle test
The circulating load obtained from the oversize to undersize ratio
after 3 cycles was steady at 70-80% for all material types.
There appears to be a general and consistent reduction in ball mill
work indices with increasing pressure

The authors would like to acknowledge Anglo American Chile, SKM


Minmetal, Polysius, CIMM, SGS-Lakefield and DJB Consultants for their
contributions to the study and this paper.
REFERENCES
1)Barratt, D.J., 1989. An Update of testing, scale-up and sizing
equipment for autogenous and semi-autogenous grinding circuits in
SAG1989, University of British Columbia, 25-46.
2)Klymowsky, I.B. and Liu, J., 1997. Modelling of the comminution in a
roller press in Proceedings of the XX IMPC, Aachen, Vol. 2, 141-154.
3)Patzelt, N., Klymowsky, R.I.B., Knecht, J., Burchardt, E., 2005.
HPGRs for Hard Rock Applications Proceedings of the Randol
Innovative Metallurgy Forum 2005, Perth
4)Smit, I., 2005. "Bench scale ore characterisation using the High
Pressure Grinding Roll" Proceedings of the Randol Innovative
Metallurgy Forum 2005, Perth

You might also like