You are on page 1of 10

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/268881977

Evaluation of dry grinding using HPGR in closed circuit with an air classifier

Article in Minerals Engineering · November 2014


DOI: 10.1016/j.mineng.2014.10.023

CITATIONS READS

20 2,373

4 authors, including:

Alex Jankovic Steven Suthers


Hatch The Commonwealth Scientific and Industrial Research Organisation
78 PUBLICATIONS 877 CITATIONS 22 PUBLICATIONS 691 CITATIONS

SEE PROFILE SEE PROFILE

Walter Valery
The University of Queensland
111 PUBLICATIONS 875 CITATIONS

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Flotation View project

Sintering and Pelletising View project

All content following this page was uploaded by Steven Suthers on 20 September 2018.

The user has requested enhancement of the downloaded file.


Dear Author,

Please, note that changes made to the HTML content will be


added to the article before publication, but are not reflected
in this PDF.

Note also that this file should not be used for submitting
corrections.
Our reference: MINE 4510 P-authorquery-v11

AUTHOR QUERY FORM

Journal: MINE Please e-mail or fax your responses and any corrections to:

E-mail: corrections.esch@elsevier.sps.co.in

Article Number: 4510 Fax: +31 2048 52799

Dear Author,
Please check your proof carefully and mark all corrections at the appropriate place in the proof (e.g., by using on-screen annotation in the PDF
file) or compile them in a separate list. Note: if you opt to annotate the file with software other than Adobe Reader then please also highlight
the appropriate place in the PDF file. To ensure fast publication of your paper please return your corrections within 48 hours.
For correction or revision of any artwork, please consult http://www.elsevier.com/artworkinstructions.

Any queries or remarks that have arisen during the processing of your manuscript are listed below and highlighted by flags in the proof. Click
on the ‘Q’ link to go to the location in the proof.

Location in Query / Remark: click on the Q link to go


article Please insert your reply or correction at the corresponding line in the proof

Q1 Please confirm that given name(s) and surname(s) have been identified correctly.

Please check this box if you have no


corrections to make to the PDF file

Thank you for your assistance.


MINE 4510 No. of Pages 1, Model 5G
10 November 2014

Highlights

 HPGR was tested in closed circuit with an air classifier to produce a fine product.  This was benchmarked against HPGR with dry screening
followed by ball milling.  Logged power input was compared with power input calculated from Bond tests.  HPGR with air classification used
30% less power than HPGR with dry screening.  Bond test calculations conservatively indicated a power saving of around 20%.

1
MINE 4510 No. of Pages 7, Model 5G
10 November 2014

Minerals Engineering xxx (2014) xxx–xxx


1

Contents lists available at ScienceDirect

Minerals Engineering
journal homepage: www.elsevier.com/locate/mineng

5
6

3 Evaluation of dry grinding using HPGR in closed circuit with an air


4 classifier
7 Q1 Alex Jankovic a, Steve Suthers b,⇑, Thomas Wills a, Walter Valery a
8 a
Metso Process Technology and Innovation, PO Box 221, Kenmore, QLD 4069, Australia
9 b
CSIRO Process Science and Engineering, PO Box 883, Kenmore, QLD 4069, Australia

10
11
a r t i c l e i n f o a b s t r a c t
1
2 3
4
14 Article history: Comparative dry grinding tests were conducted for two grinding flowsheet options using commercial 25
15 Received 17 June 2014 aggregate as feed material: Option A, using a high pressure grinding roll (HPGR) in closed circuit with 26
16 Accepted 30 October 2014 air classification, and Option B, using HPGR in closed circuit with a 2.36 mm screen, followed by a 27
17 Available online xxxx
locked-cycle Bond test. Bond tests were also carried out on standard crushed feed passing 3.35 mm 28
and 2.36 mm for comparison. The feed for the tests was screened and crushed to pass 10 mm. The air 29
18 Keywords: classifier produced a fine product with an 80% passing size (P80) of around 50 lm. In order to maintain 30
19 HPGR
comparability, the Bond tests were carried out using a 75 lm closing screen, producing a final product 31
20 Air classification
21 Energy efficiency
with a P80 of 57 lm. In all tests, the power consumption of the HPGR, Bond mill and air classifier were 32
22 Ball mill recorded directly using a power meter, while the Bond mill power consumption was also calculated using 33
23 Bond’s third law and other published methods. 34
Testing determined that the specific energy consumption of the Option B circuit was 41.9% greater than 35
Option A when evaluated using power logging, or 26.2% greater when calculated using Bond’s law. Option 36
A required 20.8–29.5% less energy per tonne of ore processed than Option B, a conservative estimate due 37
to the finer grind size achieved. Further Bond tests showed that the work indices of standard crushed 38
3.35 mm and 2.36 mm feed were similar (15.0 and 15.3 kW h/t, respectively), while the HPGR crushed 39
2.36 mm feed produced a lower work index of 14.0 kW h/t. These results agree with observations by 40
other workers that HPGR renders a sample more amenable to comminution, most likely due to the intro- 41
duction of micro-cracks. 42
Ó 2014 Published by Elsevier Ltd. 43
44

45
46
47 1. Introduction shortages in countries such as Africa and Chile, the mining industry 59
is now considering dry grinding options more readily. 60
48 Traditionally, the majority of mineral processing operations Dry grinding has been utilised in the cement industry for many 61
49 conduct ore handling and comminution steps by a combination years for comminution of limestone, slag and clinker, firstly 62
50 of dry crushing moist ore, followed by further wet grinding and through ball and rod mills, and in recent decades using high pres- 63
51 classification stages to reach a target grind size. In the current min- sure grinding rolls (HPGR). Circuits closed with air classification 64
52 ing climate however, with the rising number of operations in areas have enabled the use of HPGR in the cement industry for reliable 65
53 with poor power handling or capacity and water scarcity, such as pre-mill grind or final product grind, reducing power and steel 66
54 in North Western Australia, the Andes and the Sahara Desert, usage cost, and increasing capacity, while operating to make high 67
55 exploration and consideration of dry grinding is growing. With quality product as fine as 25 lm (Van der Meer et al., 2012; 68
56 issues such as climate change, green house gas emissions and car- Aydoğan et al., 2006). 69
57 bon footprint gaining priority status in public, coupled with rising The current HPGR comprises two motor driven counter rotating 70
58 electricity costs in response to high fuel prices and electric power rolls, one fixed, and one acting against hydraulic cylinders con- 71
nected to pressurised nitrogen accumulators. Rock is choke-fed 72
to the roll gap, with nip and pre-breakage occurring for particles 73
larger than the gap by single particle comminution, and smaller 74
⇑ Corresponding author. Tel.: +61 (07)33274565. particles forming a compressed bed between the rolls enabling 75
E-mail addresses: alex.jankovic@metso.com (A. Jankovic), steve.suthers@csiro.au more efficient bed breakage mechanics (Hilden and Suthers, 2010). 76
(S. Suthers), thomas.wills@metso.com (T. Wills), walter.valery@metso.com
(W. Valery).

http://dx.doi.org/10.1016/j.mineng.2014.10.023
0892-6875/Ó 2014 Published by Elsevier Ltd.

Please cite this article in press as: Jankovic, A., et al. Evaluation of dry grinding using HPGR in closed circuit with an air classifier. Miner. Eng. (2014), http://
dx.doi.org/10.1016/j.mineng.2014.10.023
MINE 4510 No. of Pages 7, Model 5G
10 November 2014

2 A. Jankovic et al. / Minerals Engineering xxx (2014) xxx–xxx

77 Since the introduction of HPGR to the cement industry in 1985, both products were collected and weighed. The oversize was 138
78 studies and applications of HPGR to hard rock operations have stage-crushed using a laboratory jaw crusher to pass 10 mm. The 139
79 become increasingly common. Several notable hard rock installa- resulting 10 mm sample was homogenised and split into 32 por- 140
80 tions include Cerro Verde McMoRan Copper–Molybdenum in Peru, tions of 15 kg using a 16-bin rotary splitter. A representative sub- 141
81 Mogalakwena Anglo American Platinum in South Africa, Bodding- split was taken for size analysis using a 12-bin rotary splitter, and a 142
82 ton Newmont Copper–Gold in Australia, Grasberg McMoRan Cop- further sub-split was taken using an eight-jar rotary splitter to 143
83 per–Gold in Indonesia, CVRD Iron Ore in Brazil, Kudremukh Iron achieve a 500 g sample for size analysis. The P80 of the 10 mm 144
84 Ore in India, El Brocal Lead–Zinc and Copper in Peru, CAP Minería aggregate feed was 7.4 mm. The feed for the Bond tests was pre- 145
85 Iron Ore in Chile, and SNIM Iron Ore in Mauritania (Hilden and pared by stage crushing 30 kg of the 10 mm sample to pass 146
86 Suthers, 2010; International Mining, 2012; Weir Minerals, 2011). 3.35 mm. Half of the 3.35 mm material was split out and retained 147
87 One of the main benefits of HPGR use is comminution energy effi- for the 3.35 mm Bond tests, while the other half was screened at 148
88 ciency, with researchers reporting energy savings of about 10–50% 2.36 mm to produce the sample for the 2.36 mm Bond tests. 149
89 compared to grinding by conventional ball and rod milling or semi-
90 autogenous grinding with ball milling (Daniel, 2007; Rosario and 2.2. Air classifier setup 150
91 Hall, 2008; Hilden and Suthers, 2010). The reported energy savings
92 depend on the circuit arrangements, whether the grinding is being A KHD rotating wheel air classifier was utilised in this test work, 151
93 carried out wet or dry, the hardness of the ore, the amount of addi- comprising a feeder hopper and motor, product storage and dust 152
94 tional material handling operations and the methods used to cyclone as shown in Fig. 1. Energy consumption during testing 153
95 define the energy savings. Other benefits include reduced water was logged using a NanoVip Plus digital clamp-on power meter, 154
96 and grinding media use, further reducing operational costs connected via a serial interface to a PC running the associated log- 155
97 (Hilden and Suthers, 2010). ging software. The no-load consumption was determined from the 156
98 In dry grinding applications, closed circuit HPGR utilises screens average instantaneous power draw of the fan motor prior to and 157
99 or air classification for dry classification and recirculation. While after sample feeding. The classification power draw (or load power 158
100 dry screens are only applicable for coarse product sizes due to draw) is the power draw during feeding, and the net power draw is 159
101 the issue of agglomeration, air classification may be applied for the difference between load and no-load draw. 160
102 P80 sizes between 25 and 1500 lm. Companies such as KHD, who
103 deal in dry separation equipment, have produced several commer- 2.3. HPGR tests 161
104 cially used air-based classification systems including the V separa-
105 tor and SKS dynamic cage wheel separator. Van der Meer et al. Grinding tests were carried out using a fully instrumented 162
106 (2012) reported feasibility testing for dry grinding with HPGR of Krupp Polysius HPGR unit as shown in Fig. 2. The specifications 163
107 a North American magnetite ore, as an alternative to tertiary crush- of the unit are provided in Table 1. Power consumption, working 164
108 ing and ball milling for size reduction from 50 mm to 90 lm. They pressure and operating gap were logged by a computer with Lab- 165
109 reported that a wet process would require downstream sedimenta- view software, and a standard peripheral roll speed of 0.38 m/s 166
110 tion, thickening and filtration stages and would be too capital was used in all tests. The oil pressure was set at 4.5 Mpa and the 167
111 intensive, while a dry ball mill would present high operation costs. gas pressure was set at 1.5 MPa to give an oil and gas pressure ratio 168
112 Comparative grinding tests of two-stage closed HPGR with 7 mm of 3:1, while a nominal roll gap of 1.6 mm was used, which 169
113 and 1 mm screens and a final mill grind, and two-stage closed achieved the desired specific press force of about 4–5 N/mm2. Size 170
114 HPGR with a 7 mm screen and an air classifier cutting for final
115 product, showed a 46% energy reduction for the latter HPGR-only
116 circuit (Van der Meer et al., 2012).
117 Two HPGR flowsheet options were tested in the present study
118 using commercial aggregate material, screened and crushed to
119 pass 10 mm, as the feed. Option A comprised a HPGR closed with
120 an air classifier generating final product directly from 10 mm
121 feed, while Option B comprised the same HPGR closed with a
122 2.36 mm screen, producing feed for a Bond mill test closed with
123 a 75 lm screen to generate the final product. The power consump-
124 tions of the HPGR, Bond mill and air classifier were directly mea-
125 sured and recorded via inline power meters, while the power
126 consumption of the Bond mill and a scaled-up mill were also calcu-
127 lated using various methods for comparison. Additional Bond
128 index tests have been conducted on standard 3.35 mm Bond test
129 feed, 2.36 mm crushed feed, and HPGR crushed 2.36 mm feed to
130 aid in energy comparison analysis.

131 2. Materials and methods

132 Experimental procedures include feed preparation, calibration


133 tests for HPGR and air classifier equipment, and comparative grind-
134 ing tests.

135 2.1. Feed preparation

136 The 500 kg of dry aggregate sample received was pre-screened


137 using a ‘‘Russel’’ vibrating sieve with a 10 mm aperture size, and Fig. 1. KHD rotating wheel air classifier. Photo: CSIRO.

Please cite this article in press as: Jankovic, A., et al. Evaluation of dry grinding using HPGR in closed circuit with an air classifier. Miner. Eng. (2014), http://
dx.doi.org/10.1016/j.mineng.2014.10.023
MINE 4510 No. of Pages 7, Model 5G
10 November 2014

A. Jankovic et al. / Minerals Engineering xxx (2014) xxx–xxx 3

the next cycle was produced by adding fresh feed equivalent to 187
the amount of removed screen undersize to the screen oversize, 188
followed by homogenising by three passes through a 12-bin rotary 189
splitter. Six locked cycles were performed, with all 2.36 mm 190
products combined, homogenised and split to produce a 10 kg 191
sample of 2.36 mm for Bond testing and size analysis. 192

2.6. Ball mill grindability tests 193

Bond ball mill locked cycle tests were carried out at a closing 194
screen size of 75 lm using a standard (305 mm diameter) labora- 195
tory-scale Bond mill with a standard Bond ball charge. Mill energy 196
consumption was logged to computer using the NanoVip clamp-on 197
power meter described earlier. No-load power draw was deter- 198
mined by running the mill empty for 1 h while logging and then 199

Fig. 2. Krupp Polysius 250  100 mm roll HPGR. Photo: CSIRO.


averaging the instantaneous power recorded in the 5 min prior to 200
the test. The power draw of the loaded mill was determined by 201
averaging the power draw recorded during testing, with the net 202
Table 1 power draw being the difference between the load and no-load 203
Laboratory HPGR specifications. power draw. 204
Item Description The mill energy input per revolution was calculated from Eq. 205

Manufacturer Krupp Polysius


(1): 206
207
Year of manufacture 1991 Enet  t
Roll diameter 250 mm ER ¼ ð1Þ 209
Roll width 100 mm
R
No. of hydraulic rams 4 where ER is the mill energy input per rev (J/rev), Enet is the net 210
Hydraulic ram diameter 63.0 mm
power draw (W), t is the grinding time (s), and R is the number of 211
Machine constant, k 1.263
Rolls surface Tungsten carbide studded mill revolutions. 212
Drive motors 2  30 kW According to Daniel (2007), the energy input of a standard Bond 213
Maximum gap 12 mm mill is about 93 J/rev and the specific grinding energy for the labo- 214
Peripheral roll speed 0.38 m/s nominal, variable up to 3.6 m/s ratory-scale mill can be calculated using Eq. (2): 215
Throughput About 3 t/h nominal, up to 8 t/h at max speed 216
ER 25:83
W lab ¼ ¼ ð2Þ
171 analysis was carried out on all feed and product samples by split- 3:6Gp Gp 218
172 ting to 300 g and wet screening over a 38 lm screen, with the over- where Wlab is the specific grinding energy of the laboratory-scale 219
173 size being dry screened using a standard screen series from mill (kW h/t) and Gp is the amount of total product produced per 220
174 9500 lm to 38 lm. mill revolution (g/rev). 221
The specific grinding energy of a 2.44 m interior diameter wet 222
175 2.4. Option A – HPGR in closed circuit with air classifier overflow ball mill is usually calculated using the Bond work index 223
and Bond’s third law equation (Bond, 1961): 224
176 For each grinding cycle, 20 kg of sample was passed through the 225
177 HPGR followed by air classification. Fresh feed equivalent to the W¼ 10W i ðP0:5
80  F 0:5
80 Þ ð3Þ 227
178 quantity of removed fine product was added to the coarse product where F80 is the 80% passing size of the feed (lm), P80 is the 80% 228
179 of the air classifier, homogenised by three passes through a 12-bin passing size of the product (lm), W is the specific grinding energy 229
180 rotary splitter, and utilised as feed for the next cycle. Six locked (kW h/t), and Wi is the Bond work index (kW h/t). 230
181 cycles were completed at the determined press force, with feed Comparable results to Eq. (3) can be obtained directly from the 231
182 and product samples taken for size analysis. Bond test if a value of 60 J/rev is used for the Bond mill energy 232
input per revolution (Daniel, 2007): 233
183 2.5. Option B – HPGR in closed circuit with 2.36 mm screen plus ball 234
ER 16:67
184 milling W¼ ¼ ð4Þ
3:6G G 236

185 For each grinding cycle, 30 kg of sample was passed through the where W is the specific grinding energy (kW h/t) and G is the Bond 237
186 HPGR followed by dry screening over a 2.36 mm screen. Feed for test grindability index (g/rev). 238

Table 2
Option A grinding cycle results (feed P80 = 7395 lm; final product P80 = 50 lm).

Parameter Cycle 1 Cycle 2 Cycle 3 Cycle 4 Cycle 5 Cycle 6


2
Specific press force (N/mm ) 4.9 4.9 3.9 3.6 3.6 3.5
Circulating loada [(A + B)/F] (%) 325 551 746 735 733 683
Circulation ratiob [B/F] (%) 0 381 609 635 632 589
[HPGR] net specific energy wrt final product (kW h/t) 6.3 8.6 9.1 7.9 7.9 7.5
[Air classifier] net specific energy wrt final product (kW h/t) 4.3 7.5 5.9 6.8 6.2 n/a
[HPGR] average net specific energy wrt final product (kW h/t) 7.9 ± 1.0
[Air classifier] average net specific energy wrt final product (kW h/t) 6.1 ± 1.2
Overall net specific energy wrt final product (kW h/t) 14.0 ± 2.2
a
Circulating load = (fresh feed + recycle)/product.
b
Circulation ratio = recycle/product.

Please cite this article in press as: Jankovic, A., et al. Evaluation of dry grinding using HPGR in closed circuit with an air classifier. Miner. Eng. (2014), http://
dx.doi.org/10.1016/j.mineng.2014.10.023
MINE 4510 No. of Pages 7, Model 5G
10 November 2014

4 A. Jankovic et al. / Minerals Engineering xxx (2014) xxx–xxx

100 100

90 90
Recovery to underflow (wt.%)

Cumulative wt.% passing


80 80

70 70
60
60
50
50
40
40
30
30
20
20
10
10
0
0 10 100 1000 10000
10 100 1000 10000 Particle size (µm)
Particle size (µm) Feed Option A product Option B product

Fig. 3. Air classifier partition/efficiency curve from the Option A test. Fig. 5. Option A and B final product and fresh feed comparative particle size
distributions.

100
fines to the coarse fraction. The imperfection I, i.e. separation effi- 253
90
ciency of the air classifier, was calculated to be 0.29 using Eq. (5): 254
Cumulative wt.% passing

80 255
70 d75  d25
I¼ ð5Þ 257
60 2d50
50
where d25, d50 and d75 are the particle sizes that have a 25%, 50% and 258
40
75% chance of reporting to the undersize fraction, respectively. In 259
30 the Option A test, the values of d25, d50 and d75 were determined 260
20 to be 34 lm, 49 lm and 62 lm, respectively. For comparison, the 261
10 imperfection of hydrocyclones ranges from about 0.2 to 0.6 with 262
0 an average of about 0.3 (Gupta and Yan, 2006), so the separation 263
10 100 1000 10000
efficiency of the air classifier in the Option A test was similar to that 264
Particle size (µm)
of an average hydrocyclone. 265
Actual Feed Calculated feed
The air classifier feed distribution (actual and calculated) is 266
Fig. 4. Air classifier feed distribution (actual and calculated) from the Option A test. shown in Fig. 4. There is a difference between the actual and calcu- 267
lated feed size distributions, most likely due to some mechanical 268
attrition by the air classifier rotor – a known effect of centrifugal 269

239 3. Results and discussion fan air classifiers (Ito et al., 1993). 270

240 3.1. Option A – HPGR in closed circuit with air classifier 3.2. Option B – HPGR in closed circuit with 2.36 mm screen plus ball 271
milling 272
241 The results of the Option A test (locked cycle HPGR closed with
242 air classifier) are summarised in Table 2. The air classifier power The results of the Option B test (locked cycle HPGR closed with 273
243 consumption was not recorded in the final cycle due to equipment a 2.36 mm screen and Bond test closed with a 75 lm screen) are 274
244 issues. The average net specific energy consumption was summarised in Table 3. The average net specific energy consump- 275
245 6.1 ± 1.2 kW h/t for the air classifier, 7.9 ± 1.0 kW h/t for the HPGR, tion for the circuit was 2.8 ± 0.4 kW h/t for the HPGR, 276
246 and 14.0 ± 2.2 kW h/t overall. The circulating load of the last two 17.1 ± 1.8 kW h/t for the ball mill, and 19.9 ± 2.3 kW h/t overall. 277
247 cycles was approximately 700% and produced a product with a The circulating load for the last two HPGR cycles was approxi- 278
248 P80 of 50 lm. Approximately 5% flake was generated by the HPGR; mately 150%. Approximately 5% flake was generated by the HPGR, 279
249 however, it was of low competency, decomposing when fed to but it was of low competency and decomposed when fed to classi- 280
250 classification. fication. The final product from the Bond test had a P80 of 57 lm. 281
251 The air classifier partition curve for the Option A test is shown Comparative particle size distributions for Option A and B products 282
252 in Fig. 3. The partition curve has not been corrected for bypass of and fresh feed are shown in Fig. 5. 283

Table 3
Option B grinding cycle results (feed P80 = 7395 lm; final product P80 = 57 lm).

Parameter Cycle 1 Cycle 2 Cycle 3 Cycle 4 Cycle 5 Cycle 6


2
Specific press force (Nmm ) 5.7 4.6 4.4 5.4 3.9 3.5
[HPGR + 2.36 mm screen] circulating loada(%) 140 142 144 138 149 152
[HPGR + 2.36 mm screen] circulation ratiob (%) 0 40 42 42 42 49
[HPGR] net specific energy wrt final product (kW h/t) 3.3 2.6 2.8 3.2 2.4 2.2
[Bond test] net specific energy wrt final product (kW h/t) 14.6 18.6 16.2 15.6 17.9 17.1
[HPGR] average net specific energy wrt final product (kW h/t) 2.8 ± 0.4
[Bond test] average net specific energy wrt final product (kW h/t) 17.1 ± 1.8
Overall net specific energy wrt final product (kW h/t) 19.9 ± 2.3
a
Circulating load = (fresh feed + recycle)/product.
b
Circulation ratio = recycle/product.

Please cite this article in press as: Jankovic, A., et al. Evaluation of dry grinding using HPGR in closed circuit with an air classifier. Miner. Eng. (2014), http://
dx.doi.org/10.1016/j.mineng.2014.10.023
MINE 4510 No. of Pages 7, Model 5G
10 November 2014

A. Jankovic et al. / Minerals Engineering xxx (2014) xxx–xxx 5

100 results for each test are shown in Table 4. The specific energy input 303
90 values shown in Table 4 were determined using four different 304
Cumulative wt.% passing

80 methods, namely, direct power logging, Eq. (2) (assuming 93 J/ 305


70 rev as the energy input for the standard Bond mill), Eq. (3) (Bond’s 306
60 third law) and Eq. (4) (assuming 60 J/rev as the energy input for the 307
50 standard Bond mill). 308
40 It can be seen in Table 4 that the specific energy inputs calcu- 309

30 lated by assuming 93 J/rev are very close to those determined by 310

20 direct power logging, while the specific energy inputs calculated 311

10 by assuming 60 J/rev are reasonably close to, albeit a little lower 312

0
than, those determined using Bond’s third law. As noted earlier, 313
10 100 1000 10000 direct power logging measures the power input of the labora- 314
Particle size (µm) tory-scale Bond ball mill, while Bond’s third law actually up-scales 315
-3.35 mm feed -2.36 mm feed Option B feed the result to that of a 2.44 m industrial wet ball mill. Thus, the 316
-3.35 mm product -2.36 mm product Option B product operating work indices reported in Table 4 are larger than the 317
actual Bond work indices, as they represent the laboratory-scale 318
Fig. 6. Bond test comparative feed and product particle size distributions.
Bond mill. 319
The results in Table 4 also show that the 2.36 mm Option B 320
284 Fig. 5 shows that the air classified fine product from Option A HPGR product had the lowest Bond work index (14.0 kW h/t) and 321
285 exhibits a very wide size distribution, that is, it contains an unex- subsequently required less power input than the jaw crushed 322
286 pectedly large portion of coarse particles (2.5% of 0.3–3.0 mm). 2.36 mm feed, which had a Bond work index of 15.3 kW h/t. Note 323
287 Considering the P80 of the fine product is approximately 50 lm, in Fig. 6 that the size distributions of the Option B Bond test and 324
288 the presence of coarse particles is of concern and an indication of 2.36 mm jaw crushed feeds were almost identical, so the differ- 325
289 coarse ‘‘by-pass’’ to fines, which may create potential downstream ence in Bond work index can be attributed to micro-cracking rather 326
290 processing problems. Further investigation is required to confirm if than to differences in the feed size distributions. This result agrees 327
291 this result is common for all similar air classifiers or only for the with observations by Daniel (2007) that HPGR renders a sample 328
292 particular unit and operating conditions tested. more amenable to comminution by introducing micro-cracks. 329
The work index for the standard 3.35 mm material (15.0 kW 330
h/t) was similar to that of the jaw-crushed 2.36 mm material, 331
293 3.3. Ball mill grindability tests but the Bond test required a higher power input due to experienc- 332
ing a larger size reduction ratio in the test. 333
294 In addition to the Bond test described for Option B above, which
295 used 2.36 mm HPGR-crushed feed, Bond tests were also com-
296 pleted using 3.35 mm standard Bond test feed and 2.36 mm 4. Conclusions 334
297 jaw crushed feed, all closed at 75 lm to produce a similar product
298 to that of Option A (P80 = 50 lm) for comparison. The feed and The aim of this work was to investigate the comparative energy 335
299 product size distributions for each Bond test are shown in Fig. 6. use of HPGR with air classification and HPGR with screening and 336
300 It should be noted that the product P80 of these tests (57 lm) is lar- subsequent ball milling to produce a relatively fine product from 337
301 ger than, but similar to, that of Option A (50 lm). Comparative a 10 mm top size feed aggregate material. The overall circuit 338
302 Bond work index, operating work index and specific energy input energy consumptions are summarised in Table 5. From this table, 339

Table 4
Bond test energy input comparison results.

Parameter Sample
Feed 3.35 mm Feed 2.36 mm Option B 2.36 mm
Feed P80 (lm) 2134 1492 1444
Product P80 (lm) 58 58 57
Grindability (g/rev) 1.12 1.16 1.27
Bond work index (kW h/t) 15.0 15.3 14.0
Operating work index (kW h/t) 17.5 17.2 16.1
Specific energy input: power logging (kW h/t) 19.3 ± 4.3 18.2 ± 5.7 17.1 ± 1.8
Specific energy input: @93 J/rev (kW h/t) 19.3 18.1 16.5
Specific energy input: Bond’s third law (kW h/t) 16.5 16.1 14.9
Specific energy input: @60 J/rev (kW h/t) 14.9 14.4 13.1

Table 5
Overall circuit option specific energy input comparison results.

Stage Option A Option B Option B Units Method


HPGR from 10 mm to 50 lm 7.9 kW h/t Direct power logging
Air classifier 6.1 kW h/t Direct power logging
HPGR from 10 mm to 2.36 mm 2.8 2.8 kW h/t Direct power logging
Ball mill from 2.36 mm to 57 lm 17.1 kW h/t Direct power logging
Ball mill from 2.36 mm to 57 lm 14.9 kW h/t Bond’s third law
OVERALL 14.0 19.9 17.7 kW h/t

Please cite this article in press as: Jankovic, A., et al. Evaluation of dry grinding using HPGR in closed circuit with an air classifier. Miner. Eng. (2014), http://
dx.doi.org/10.1016/j.mineng.2014.10.023
MINE 4510 No. of Pages 7, Model 5G
10 November 2014

6 A. Jankovic et al. / Minerals Engineering xxx (2014) xxx–xxx

340 it can be seen that the specific energy input of the Option A circuit with air classification. The circulating load in the HPGR would be 377
341 (HPGR with air classification) was 14.0 kW h/t using direct power expected to be lower if the cut size of the air classifier was coarser, 378
342 logging, while it was 19.9 kW h/t for the Option B circuit (HPGR but ball milling might be more energy efficient at coarser grind 379
343 with screening and ball milling) using direct power logging. That sizes. It would therefore be of interest in future work to evaluate 380
344 is, the Option A circuit consumed 29.5% less energy per tonne of HPGR with air classification against ball milling and screening for 381
345 ore processed than the Option B circuit when compared at labora- coarser grind sizes in the range of about 150–300 lm. 382
346 tory-scale. Alternatively, the specific energy input of the Option B
347 circuit was 17.7 kW h/t when the ball mill specific grinding energy Acknowledgements 383
348 was calculated using Bond’s third law. That is, the Option A circuit
349 consumes 20.8% less energy per tonne of ore processed than the This work was supported by Metso Process Technology and 384
350 Option B circuit when the ball mill in the latter circuit is scaled Innovation in collaboration with the CSIRO Minerals Down Under 385
351 up to a 2.44 m industrial wet mill. This second comparison, how- flagship. The authors thank Venkata Nunna (CSIRO) for his valuable 386
352 ever, does not take into account any potential efficiency increase assistance, Tirsha Raynlyn (CSIRO), Tugs Tsedenbaljie (Metso) and 387
353 with up-scale of HPGR and air classifier equipment. The calculated Jeff Douglas (CSIRO) for assistance with the experimental work, 388
354 energy savings of Option A are considered conservative because and Ralph Holmes and Warren Bruckard (CSIRO) for proof reading 389
355 the P80 of the product was 50 lm while that of Option B was of the draft manuscript. 390
356 57 lm. On the other hand, this study has not taken into account
357 the power consumption of ancillary material handling equipment References 391
358 that would be required in industrial scale circuits, or the capital
Aydoğan, N.A., Ergün, L., Benzer, H., 2006. High pressure grinding rolls (HPGR) 392
359 cost of equipment and grinding media.
applications in the cement industry. Miner. Eng. 19, 130–139. 393
360 These findings are in line with other published results (Van der Bond, F.C., 1961. Crushing and grinding calculations part 1. Br. Chem. Eng. 6, 394
361 Meer et al., 2012; Aydoğan et al., 2006). Consequently, the use of 378–385. 395
Daniel, M.J., 2007. Energy efficient mineral liberation using HPGR technology. PhD 396
362 HPGR and air classification for energy efficient grinding in the min-
Thesis, University of Queensland, Brisbane, Australia. 397
363 ing industry is promising. One concern is that the air classifier Gupta, A., Yan, D.S., 2006. Mineral Processing Design and Operation. Elsevier, 398
364 product contained an unexpectedly large proportion of coarse par- Oxford. 399
365 ticles, which may be due to deficiencies of the actual laboratory Hilden, M., Suthers, S., 2010. Comparing energy efficiency of multi-pass high 400
pressure grinding roll (HPGR) circuits. In: XXV International Mineral Processing 401
366 unit or operating conditions; however, verification of this would Congress (IMPC) 2010 Proceedings. The Australasian Institute of Mining and 402
367 require further investigation. Another consideration is that the Metallurgy, Brisbane, pp. 801–811. 403
368 50 lm grind size used in this study is very fine (similar to that used International Mining, 2012. Better processing. Int. Min., 128, September. 404
Ito, M., Furukawa, T., Fujii, S., Tanaka, H., 1993. Particle comminution in a 405
369 in the cement industry), while the circulating load in the HPGR was 406
centrifugal air classifier. Adv. Powder Technol. 4 (1), 9–23.
370 700%, which is quite high. The circulating load in an industrial scale Rosario, P., Hall, R., 2008. Analyses of the total required energy for comminution of 407
371 HPGR could potentially be higher due to roll edge effects and a lar- hard ores in SAG mill and HPGR circuits. In: 5th International Mineral 408
Processing Seminar (Procemin 2008), University of Chile, Santiago, pp. 129–138. 409
372 ger operating gap. Thus, it may be necessary to reduce the top size 410
Van der Meer, F.P., önol, S., Strasser, S., 2012. Case study of dry HPGR grinding and
373 of the feed and coarsen the grind size in order to decrease the cir- classification in ore processing. In: 9th International Mineral Processing 411
374 culating load. For example, Van der Meer et al. (2012) reported a Conference (Procemin 2012), Gecamin, Santiago, pp. 32–34. 412
Weir Minerals, 2011. Introduction to High Pressure Grinding Roll Technology in 413
375 circulating load of 380% when grinding a magnetite feed from a
Mining. Weir Minerals, Madison, USA. 414
376 top size of 7 mm to 80% passing 90 lm using HPGR in closed circuit
415

Please cite this article in press as: Jankovic, A., et al. Evaluation of dry grinding using HPGR in closed circuit with an air classifier. Miner. Eng. (2014), http://
dx.doi.org/10.1016/j.mineng.2014.10.023
View publication stats

You might also like