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

net/publication/44206672

Effect of operating variables in Knelson Concentrators: a pilot-scale study

Article
Source: OAI

CITATIONS READS
15 1,764

1 author:

Sunil Kumar Koppalkar

McGill University
3 PUBLICATIONS   24 CITATIONS   

SEE PROFILE

All content following this page was uploaded by Sunil Kumar Koppalkar on 16 April 2015.

The user has requested enhancement of the downloaded file.

 PAPER 16

Effect of Operating Variables in Knelson Concentrators:

A Pilot-Scale Study

Sunil Koppalkar,
Graduate Student

and

Late Prof. A.R. Laplante

Department of Mining, Metals and Materials Engineering

McGill University, 3610 University Street
Montreal, Quebec, H3A 2B2
PH: (514) 398-4755, Ext: 09501
E-mail: [email protected]

Key Words: Gravity, Gold Recovery, Gravity Recoverable Gold, Knelson Concentrator

39th Annual Meeting of the January 23 to 25, 2007

Canadian Mineral Processors Ottawa, Ontario, Canada

247
 Proceedings of the 39th Annual Canadian Mineral Processors Conference – 2007

ABSTRACT

A Pilot Plant was installed in one of the grinding circuits of the Dome Mill, Ontario, Canada. Tests
were conducted on a KCCD12 Knelson Concentrator to systematically analyze the effect of
operating variables on the performance. The main objective of this test work was to explore the
effect of very coarse (6 mm) feed size on GRG recovery and performance of the Knelson
concentrator. Preliminary results have shown that coarse feed size is deleterious to maximizing
recovery. Size-by-size GRG recoveries have shown an unusual trend in the sense that recoveries do
not decrease monotonically with decreasing gold particle size. Similar trend observed in one other
full-scale unit has been compared. The observed trend is probably linked to the feed size distribution
and top size.

INTRODUCTION

Knelson Concentrators are the most widely used semi-continues centrifuge gravity separation units
for the recovery of gold and platinum group minerals (Xiao and Laplante 2004). The main
advantages over the competing centrifugal gravity units are their design, operating simplicity and
good separation efficiency that make them the most common choice for gravity circuit. Despite
some characterization at bench scale and a limited characterization at full scale, there is a lack of a
comprehensive study of impact of operating variables on efficiency that reflects realistic full scale
operating practice. GRG characterization studies, at bench scale, on these units provide information
about the occurrence of gold in ore samples but fail to provide adequate information about the effect
of operating variables for full-scale units. It is well known fact that such work is difficult to
undertake on full-scale units owing to the difficulties involved in sampling and systematic variation
of operating parameters.
To tackle these problems, a pilot facility consisting of a 12-in CD Knelson Concentrator, a feed
screen and tailing pump was installed in one of the grinding circuits of Dome Mill. This paper
describes the work carried out and the results.

Porcupine Joint Venture (PJV)

Porcupine Joint Venture was formed in 2002 between Placer Dome Inc. and Kinross Gold
Corporation. The JV utilizes the milling capacity of the Dome Mill to process ore from Placer Dome
and Kinross mines. The Dome Mill circuit has gravity concentration, cyanidation and CIP circuits to
recover gold. Gravity recovery has been used at Dome Mill since inception to recover gold: first by
using conventional jigs, which were eventually replaced by two Knelson Concentrators in 1993, one
in each of the grinding circuits. Three more Knelsons joined the circuit in 2002 to improve gravity
recovery. An Acacia Intensive Leach Reactor was also added in 2002 to process gold concentrates
replacing shaking tables (Chong et al 2004).

Dome Mill Circuit

Dome mill receives ore from two main sources: Dome’s open pit and underground mines, which
were lately combined to form a super pit, supplies relatively low grade ore at 2.35 g/t and a high

248
 Proceedings of the 39th Annual Canadian Mineral Processors Conference – 2007

grade of 6 to 12 g/t from Hoyle Pond Mine. Ore is crushed in three stages to produce a feed of 80%
passing ½-inch to the mills. A simplified Flowsheet of Dome Mill is shown in Figure 1.There are
two parallel grinding circuits, each consisting of a rod mill and a ball mill in closed circuit with
cyclones. A rod mill was added in grinding circuit-B in 2005. Gravity recovery is achieved by five
KCCD30 Knelson concentrators, two in grinding-A and three in grinding-B circuits, fed from
cyclone underflow. Knelson concentrates are further intensively leached in a Consep CS6000 Acacia
Intensive Leach Reactor. The Acacia loaded solution has a dedicated electrowinning circuit. Gravity
recovery accounts for up to 45% of the recovered gold depending on the ore type. Remaining gold is
recovered from CIP.

Ore Sources
Dome U/G
Dome Pit
Hoyle Pond
11,500 TPD
3.5 g/tonne
solids
Crushing process water

Grinding Gravity Circuit

47% Au Production

Cyanidation/CIP Refinery Gold Dore

53% Au Production 440,000 oz/year

Tailings Effluent Treatment Environment

Figure 1: Simplified Flowsheet of Dome Mill (Chong et al 2004)

The cyclone overflow reports to a thickener and the thickened slurry at 60% solids feeds eight
leach tanks in series for about 15 hours. The leached slurry is then pumped to the CIP circuit to
recover gold. Gold bullion, 80 % Au and 15% Ag cast as bars, is shipped to Johnson Matthey
Ltd. for refining (Folinsbee et al 2005).

PILOT TEST WORK

A pilot plant (Figure 2) integrating a vibrating screen, 12-inch CD (centre discharge) Knelson
concentrator, dedicated control panel and a sump-pump was supplied by Knelson Concentrators Inc.,

249
 Proceedings of the 39th Annual Canadian Mineral Processors Conference – 2007

Langley, BC, Canada to the Dome Mill. The pilot plant installed in grinding circuit-B (Figure 3)
received same feed as that received by the full-scale Knelson concentrators.

Figure 2: KC CD12 Pilot Plant (Courtesy: Knelson Concentrators)

Figure 3: Schematic of Dome Mill Grinding Circuit with Pilot Plant

250
 Proceedings of the 39th Annual Canadian Mineral Processors Conference – 2007

Coarse feed size, minus 6 mm to the full-scale Knelson concentrators at the time of sampling proved
bottleneck for the test work. Modifications such as putting a 8-inch pipeline to handle tails, water
jets for smooth flow of tailings and bigger screen discharge chute reduced the plug ups to some
extent and hence 12 tests were conducted varying the operating variables systematically. The main
variables analyzed were: feed rate, feed size distribution and top size, fluidization water flow rate,
rotating velocity. Before starting a test, required screen panel was installed, flow rate and rotating
velocity to the pilot unit were set using the fully controlled ICS system. Feed to the pilot plant was
introduced by opening 4-in bleed valve provided on the feed pipe to the full scale Knelson
concentrator. Tail sample was collected throughout the test manually whereas concentrate recovered
fully at the end of the test. Flow rate of tailings and density were measured frequently during the test.
Samples collected during the pilot testing were weighed wet and filtered at the mill site. The wet
cakes of feed, concentrate and tails were then shipped to McGill University for further processing.
The samples received from the pilot test were oven dried. Tramp iron was separated from the
concentrates using hand magnets. The concentrate samples were then riffled twice using John’s
Riffler and only a quarter of the total mass was retained for size analysis. The remaining
concentrates were immediately shipped back to the Dome mill. Concentrate samples were dry
screened from 1700 to 25 µm. The dried tail samples were screened at 850 µm. The minus 850 µm
fraction ranging from 11 to 27 kg was processed on a 3-inch laboratory Knelson concentrator to
recover GRG lost to the pilot test tails. The tail samples of the lab tests were wet screened at 25 µm
using 600 g sample and then the plus 25 µm fraction was dry screened from 600 to 25 µm. All size
fractions were shipped to PJV for gold analysis.

RESULTS AND DISCUSSION

Pilot Knelson performance was evaluated assuming gold recovered by pilot unit as 100% GRG and
GRG in the tailings proportional to gold recovered by the laboratory Knelson concentrator. Hence
GRG recovered, Ri for size fraction i, is:

M c × mic × Gic
Ri = × 100%
c c 30 % − 850μm t t t GRGit
M c × mi × Gi + 1000 × × × Qt × mi × Gi ×
60 100% 100%

where Mc : mass of dry concentrate (kg)

mi c : % mass of concentrate in size fraction i
Gic : gold grade of concentrate size fraction i (g/t)
%-850 µmt : % of the Knelson CD12 tailing finer than 850 µm
Qt : dry flow rate of Knelson CD12 tailings (t/h)
mi t : % mass of tailing in size fraction i (based on –850 µm)
Git : gold grade of tailing size fraction i (g/t)
GRGit : % of gold that is GRG in size fraction i of CD 12 tailing

Table 1 shows the experimental design conditions and metallurgical performance for the 12 short
recovery cycle (30 minutes) tests. It can be seen from the data that GRG recoveries fluctuated
between 37 and 80 % mainly because of the fluctuating feed rates from 6.6 to 14.1 t/h. It can also be

251
 Proceedings of the 39th Annual Canadian Mineral Processors Conference – 2007

seen that feed grades to the pilot unit were very high for the first five tests may be because of a
significant portion of ore from Hoyle Pond Mine, which supplies high grade material and thereafter
remained quite consistently at around 5 g/t.

Table 1: Experimental Design Conditions and Overall Metallurgical Results

Test # Design Screen Fluid.flow Rotating Dry Tons Concentrate KC CD12 Tails Feed
feed rate Size USGPM Velocity t/hr % GRG Au, g/t GRG non-GRG GRG
Recovery g/t g/t g/t
1 mid mid mid 796 8.8 74 5653 3.22 2.45 12.41
2 mid mid low 650 9.7 65 10275 3.77 3.34 10.7
3 mid mid low 650 9.6 70 11941 3.14 2.4 10.33
4 mid mid mid 531 11.3 65 11538 3.86 2.14 11.03
5 mid mid low 796 10.8 61 7768 3.94 3.78 10.02
7 mid mid mid 650 13.6 45 5379 3.24 2.52 5.88
10 mid mid low 531 9.3 59 4694 2.39 2.27 5.79
14 mid mid mid 650 14.1 60 3044 1.89 2.07 4.77
16 mid low low 650 6.6 80 3983 1.07 2.35 5.37
17 mid high high 650 7.8 54 1334 2.35 2.89 5.09
19 mid high mid 650 13.3 37 1463 2.84 2.42 4.52
26 high mid mid 650 13.1 50 8140 3.92 3.02 7.81

Legend: 1. Screen size: Low-1.0 mm, mid-2.0 mm and high-3.5 mm

2. Rotating velocity: Low-531rpm (40 G), mid-650 rpm (60 G) and high-796 (90 G)
3. Fluidization flow: Low- 2-4 m3 /hr, mid-4.5- 5.7 m3/hr, high- 6.4 m3/hr

Overall GRG recoveries for the 12 tests have been plotted in Figure 4 as function of feed rate for the
three top sizes chosen. It can be seen that much of the variations in GRG recovery are due to feed
rate with top particle size also being a significant contributor.

100

80
% GRG Recovery

60

40
Mid size(2.0mm)
20 Fine(1.0mm)
Coarse(3.5mm)
0
5 7 9 11 13 15
Feed rate, T/h

Figure 4: GRG Recoveries as a Function of Feed Rate for the Three Top Sizes

Coarse feed size has produced lowest recoveries among the tests. Figure 4 also shows the results at
mid top sizes, which do not exhibit any significant trend. Increase in feed rate had a negative impact
on recovery. The effect of finer top size (1.0 mm) could not be ascertained as only one test was

252
 Proceedings of the 39th Annual Canadian Mineral Processors Conference – 2007

conducted at low feed rate. It can be seen that feed rate and top size affect GRG recovery, which can
be fitted to the equation:

RGRG = 121.2 − 3.31 × Q − 11.65 × X

where Q is the dry feed rate in t/h and X , the top feed size in mm. The effect of fine top size is not
included in the equation as only a single data point available at low feed rate. The regression has
lack of fit of 6% and a significance of 99.7%. The two coefficients for Q and X are significant at the
95% level. Given the scatter of data, the correlation coefficient is low, 0.87.

100

80
% GRG Recovery

60

40
Mid size
Fine size
20
Coarse
Repeats(mid size)
0
5 7 9 11 13 15
Feed rate, t/h

Figure 5: Effect Feed Rate with Repeats (mid size) Highlighted

Figure 5 shows results of Figure 4 with the repeats at mid size (highlighted), fluidization flow and
rotating velocity. The scatter is in part due to normal experimental variation and probably due to the
change in GRG size distribution of the feed.

100
Low fluidization flow
80
% GRG Recovery

60

40
High fluidization flow
20

0
10 100 1000
Particle size, µm

Figure 6: Average GRG Recovery Curves for 12 Tests as Function of Particle Size

253
 Proceedings of the 39th Annual Canadian Mineral Processors Conference – 2007

Figure 6 shows recovery curves for low fluidization flows and at high fluidization flow rates with the
average GRG curve in the middle. It can be seen that the recovery trend is consistent for both high
and low fluidization flow rates.

100 GRG Recovery

Feed size distribution
80
%GRG Recovery
60

40

20

0
10 100 1000
Particle size, µm

Figure 7: GRG Curve (Average of 12 Tests) with Feed Size Distribution

Figure 7 shows the average GRG curve for the 12 tests with feed size distribution. The shape of the
recovery curve that emerged from the averaged data is unusual, in the sense that the general trend of
GRG recovery - decreasing with decreasing particle size is not observed. It can be seen that there is a
recovery trough corresponding to the mode of the size distribution, which is at a size of 212 µm. In
order to compare the unusual trend, performances of two full-scale Knelson concentrators have been
chosen, a KCCD 20 Knelson concentrator operated at Golden Giant Mine and the other a KCCD30
Knelson concentrator operated at Marvel Loch Plant, Australia.

100

80
% GRG Recovery

35-60
60
45-60
60-60
40 45-120
60-120
80-120
20
S.D.

0
10 100 1000
Particle Size, µm

Figure 8: GRG Recoveries as a Function of Particle Size with Feed Size Distribution (S.D)
(Legend: The first number indicates fluidization flow rates (US gpm) and the second, G-force)
Golden Giant Plant (Laplante 1996)

254
 Proceedings of the 39th Annual Canadian Mineral Processors Conference – 2007

The performance at different fluidization flows and G forces for the KCCD20 Knelson concentrator
at Golden Giant plant is shown in Figure 8. GRG recoveries at different fluidization flow and G-
force have a decreasing trend with decreasing gold particle size. It may be seen that there is no
corresponding recovery trough at the mode of feed size distribution. But the curves indicate that
there is no significant impact of either fluidization flow rate or G force on the shape of the recovery
curves something similar suggested by the Dome Mill pilot Knelson test work as seen in Figure 7.

100
GRG Recovery
80
% GRG Recovery

Feed Size Distribution

60

40

20

0
10 100 1000
Particle size, µm

Figure 9: GRG Recovery for KC CD 30 Knelson Concentrator with Feed Size

Distribution, Marvel Loch Plant (Laplante, 2003)

In Figure 9, GRG recovery for Knelson Concentrator, Marvel Loch Plant operated at 30 t/h is shown
with feed size distribution. It can be seen that the shape of the recovery curve has a striking
similarity to the pilot tests curve and also has a similar feed size distribution.

100
CD30
80 CD12
% GRG Recovery

60

40

20

0
10 100 1000
Particle size, µm

Figure 10: Comparison of GRG recoveries of Pilot Unit with the Marvel Loch Plant

255
 Proceedings of the 39th Annual Canadian Mineral Processors Conference – 2007

The GRG recovery curves for Dome Mill Pilot Plant (KCCD12) and Marvel Loch Plant (KC CD30)
Knelson concentrators have been plotted together in Figure 10. The similarity between these curves
is striking. These two curves display an unusual shape of recovery curves. Both these curves had a
similar feed size distribution, which is definitely coarser than that of Golden Giant feed size
distribution curve. These curves also suggest that operating conditions, like, feed rate, rotating
velocity, fluidization water flow and screen aperture size, have little impact on the shape of the
recovery curve except the feed size distribution. A coarser feed size distribution similar to the pilot
test data will recover particles between 25 and 106 µm but would impact negatively on the recovery
between 106 and 425 µm. On the other hand, a finer feed like the Golden Giant would produce a
monotonically decreasing recovery with decreasing particle size. This would be linked to the natural
resistance to flow or percolation, which is significant at a particle size where the gangue is most
abundant.

Other researchers have also observed similar recovery troughs. Ling (1996) conducted tests on a 3-in
Knelson concentrator and observed a dip in recovery at intermediate sizes and linked it to the higher
ratio of diameter of intermediate size tungsten to gangue particles compared to the ratio of diameters
of finer tungsten to gangue particles and hence the intermediate tungsten has a lower ability to
percolate through the small voids of a relatively compact separation zone. Moreover, intermediate
size tungsten particles; unlike coarser tungsten particles do not posses mass large enough to intrude
forcefully into the separation zone. Laplante et al (1996) have postulated a particle shape effect to
explain the recovery trough. An increase in feed density decreases the porosity of the bed, and makes
the percolation of fine flakes (75-300 µm) more difficult than fine spheres (-37 µm), which do not
experience any percolation problems. Similar recovery troughs observed in magnetite and tungsten-
based test work, respectively (Buonvino, 1993; Ling, 1996; Ancia et al 1997), however do not
support the effect of particle shape and suggest particle size as the main cause of the recovery
trough. Increase in feed rate decreases the retention time in the bowl, thereby affecting recovery.
Decreasing fluidization flow rates can also affect recovery because of percolation problems. Tests
conducted on Aurbel’s KC CD20 Knelson concentrator (Vincent, 1997) have yielded lower
recoveries for the intermediate size fractions, which were attributed to sub optimal fluidization flow
rates.

CONCLUSIONS

The KC CD12 Pilot test work at Dome Mill has yielded gold recovery curves with unusual shape.
The shape of the recovery curve shows peaks between 25 and 106 µm and above 600 µm. This is
linked to the feed size distribution to the pilot unit. The Pilot KC unit has produced higher GRG
recoveries than expected for the feed rates deployed. The manufacturer specified feed rates are at 8-
10 t/hr. The test work has produced higher recoveries that have significantly been affected by feed
rate and top particle size. The results show that there is no significant impact of either fluidization
flow or rotating velocity on recovery.

256
 Proceedings of the 39th Annual Canadian Mineral Processors Conference – 2007

ACKNOWLEDGEMENTS

The author expresses his sincere gratitude to his supervisor, late Professor A.R. Laplante for
designing and supervising the pilot test work and also analyzing the results of this test work. The
author thanks the staff of PJV Mill for the permission to conduct pilot tests and Knelson
Concentrators for sponsoring the sampling program. The help and assistance during test work by
both PJV staff and Knelson Concentrators is gratefully acknowledged. Thanks are also due to the
PJV Assay department, for assaying all of the samples. The author would also thank Dr. Liming
Huang, Zhixian Xiao colleagues, for their help during the pilot test work, Ray Langlois for his help
in processing the samples at the laboratory.

REFERENCES

Ancia P., Frenay J., and Dandois P. “Comparison of Knelson and Falcon Centrifugal Separators,
Innovation in Physical Separation Technologies”, Richard Mozley Symposium, Falmouth,
United Kingdom, pp.53-62, 1997

Buonvino M, M. Eng, thesis, McGill University, 1993

Chong T. Y., Strickland D., Folinsbee J.A, Ron Millions, Fullam M., Grewal I. “Recent Gravity
improvements at the Porcupine Joint venture”, Proceedings-36th Annual Meeting of Canadian
Mineral Processors, Paper30, Ottawa, 2004

Folinsbee J.A and Chong T.A, Fullam M. “Continuous improvements of Porcupine joint venture
(PJV) Dome Mine’s gravity circuit. Treatment of Gold Ores”, First International Symposium,
44th Annual Conference of Metallurgists of CIM, Calgary, Alberta, Canada, pp65-79, 2005

Laplante A.R. “Report on Golden Giant Plant, Hemlo Gold Mines”, 1996 (personal
communication)

Laplante A.R., Shu, Y., and Marois, J. “Experimental Characterization of a Laboratory

Centrifugal Separator”, Canadian Metallurgical Quarterly, Vol.35, No.1, pp.23-29, 1996

Laplante A.R. “Marvel Loch site visit report”, 2003 (personal communication)

Ling J. “Variable Speed Knelson Concentrator”, PhD thesis, McGill University, 1998

Vincent F, M. Eng. thesis, McGill University, 1997

Xiao Z., and Laplante A.R. Characterizing and recovering Platinum group minerals- A review.
Minerals Engineering, Vol.17, pp.961-979, 2004

257
View publication stats