13.1
13.
Overview of comminution tests for ore characterisation T. Kojovic JK Mineral Research Centre, Universit Universityy of Queensland Email:
[email protected]
Introduction A significant problem during initial evaluation of drill core and definition of potential ore-types is the need to composite and blend in order to satisfy the high-cost/ large sample volume requirements of many ‘quantitative’ physical tests. Tese tests are essential for defining practical design parameters but can disguise variability and discourage an iterative approach to sample selection and ore-type definition based on proven similarity. Tis commonly leads to poor comparative statistics because the sample sets are small and, even though they are characterized by high quality analytical data, there is a high degree of ‘noise’ between and within the overlycomposited and poorly-validated ore types. A key focus of the AMIRA P843 GeMIII project is provision of small-scale ‘comparative’ tests that can define relative similarity and difference at the drill-corescale. Tere is a wide array of physical tests available for comparative testing and the GeMIII project has undertaken a review of the most promising comminution tests. Te review canvassed the two major types: • Tumbling tests (Bond ball mill work index (BMWI) and rod mill work index (RMWI); Semi-autogenous grinding (SAG) power index (SPI) and SAGDesign; Bond Abrasion; JK Ore Abrasion) • Impact/Comp Impact/Compression ression tests (Bond crushing work index (CWI); unconned compressive strength (UCS); Point Load index (PL index (PLT), T), Drop Weight (DWT), SAG mill comminution (SMC); EquoTip and the JK rotary breakage test (RBT)) Details on most of these tests can be found in NapierMunn et al. (1996) and ISRM (1985). SPI and SAGDesign are covered in Starkey et al. (1994, 2006)
and Starkey and Dobby (1996). Other tests designed for coarse rock particles include the Amdel-Orway “Advanced Media Competency Test”, the “MacPherson Autogenous Grindability Test” and the “Kilborn Test” (MacPherson et al., 1999).
Tumbling tests Tere are several tumbling tests which claim to be suitable for tumbling mill characterisation. e Bond test is the best known for rod and ball mills, whilst in recent years the SPI and SAGDesign test has become popular for SAG mills. Tese are reviewed below. Bond ball mill and rod mill work index tests (BMWI, (BMWI, RMWI)
In 1952, Fred Bond published his theory of comminution, which, together with the laboratory-scale tumbling tests he developed, have become the industry standard for estimating the specic energy (kWh/t) of rod and ball mills (Bond, 1946, 1952, 1961, 1963). Tis procedure involves conducting locked cycle grinding tests in a 12" (0.305 m) x 12" (0.305 m) diameter mill for ball milling and a 12" x 24" (0.61 m) mill for rod milling. Figure 1 shows the Bond ball mill test apparatus. Each mill is charged with a standard load of balls or rods. Te rock sample is crushed down to a nominal size distribution, which differs depending on whether a rod mill or a ball mill test is being conducted, then ground for a specified period. At the end of this period the ground material is taken out, screened at the target size (referred to as the closing size of the test, which is defined by a P80), and the oversize returned to the mill with additional fresh feed equal in mass to the undersize GeMIII (Amira P843) echnical echnical Report 1 – February, 2008
13.2
Figure 1. Bond Ball Mill Grindability Test apparatus
removed. Tis process is repeated, the grind time being iteratively adjusted so that for rod milling a recycle load of 100% is obtained and for ball milling 250%. A full Bond test typically has 7 to 8 cycles. e closing size is a controlling parameter parameter for the test, and is typically selected on the basis of the expected optimum liberation size for the target mineral mineral to be extracted. extracted. e Bond ball ball mill test requires 5kg of -3.35 mm material (or approximatel approximatelyy 2.2 m of NQ ½-core). e rod mill test requires 10kg of –12.7 mm material (or 4.4 m of NQ ½-core.) Bond (1952) provided equations to obtain the, socalled, rod and ball work indices from the results of the test. Te indices are used in a further equation relating specific energy to feed and product sizes. According to Bond’s third theory of comminution, the work input is proportional to the new crack tip length produced in particle breakage, and equals the work represented by the product product minus that represented represented by the feed. Te relationship is expressed as follows: W
=
10 Wi P
−
10 Wi F
(1)
where W = Specic energy = Power/r Power/roughput oughput (kWh/t) (kWh/t) W i = Work index (kWh/t) P = 80% passing size for the product (microns) F = 80% passing size for the feed (microns) e work index was dened by Bond (1952) as the comminution parameter which expresses the resistance of the material to crushing and grinding; numerically it is the kilowatt hours per tonne required to reduce the material from theoretically infinite feed size to 80% passing 100 µm. In practice Wi has to be determined from plant data
or by conducting laboratory grinding tests in which W, P and F are measured. For ball mills, Equation (1) is then used to calculate the specific power required to reduce a given F80 to the required P80 in an 8 ft diameter wet overflow ball mill. For a given throughput (t/h) the specic power (kWh/t) is converted to power draw (kW). Mill dimensions are then chosen to draw the required power, using an appropriate mill size-power relationship. Bond assumed that the net energy consumption per revolution of the test mills he used remained constant. Levin (1989) estimates that on average this value is 198.4x10–7 kWh/rev for a Bond ball mill, but states it is far from constant. Tis value was implicitly incorporated by Bond (1961) in his equation for determining the laboratory ball and rod mill work index, by calibrating his laboratory procedure with full-scale mill data. Despite reservations by some researchers (Morrell, 2004a) as to the form of the Bond equations, the Bond test has become the industry standard for estimating the comminution energy required to reduce rock from one size to another and has been applied to all comminution steps ranging from blasting to fine grinding. Various factors have been added, depending on the application, with the intention of improvin improvingg its accuracy accuracy.. Howev However, er, the basic equation has remained unchanged. SAG Power Index test (SPI)
While Bond is the best known grindabilit grindabilityy test for rod and ball mills, in recent years the SPI (SAG Power Power Index) test has become popular for SAG mills. e SPI test is a batch test developed by Minnov Minnovex. ex. e test employs a 30.5 cm diameter by 10.2 cm long grinding mill charged with 5 kg of steel balls. e SPI test requires 2 kg of coarse rejects from drill core or
13.3
RVC cuttings (–19 mm material) – prepared by crushing to 100% –19 mm (80% –13 mm; Fig. 2a). e mill is run with several screening iterations until the sample is reduced in size to 80% at –1.7 mm (or 10-mesh Tyler sieve opening). Te length of time required to achieve a size of 80% minus 10-mesh, in minutes, is called the SAG Power Index, or SPI (Starkey et al., 1994; Starkey and Dobby, 1996). Typically the SPI is determined from a plot such as that shown in Figure 2b. e SPI is used to predict the specic energy and transfer size of an existing or new mill via a series of proprietary equations that have been calibrated using a full-scale plant database (Starkey and Dobby, 1996; Dobby et al., 2001). In its early development in 1991, a very simple equation was put forward for the SAG specic energy or power index W sag : W sag (kWh/t) = (0.1SPI + 2.2) x T80-0.33
(a)
(2) (b)
Te technique was further developed when it was realized that the SAG mill specic energy was inuenced by a number of factors and could not be described in such a simple fashion. Hence by 2001 the single equation given above was changed slightly in form (Equation 3) and linked to a series of supplementary equations describing various dependencies, including factors such as pebble crusher recycle load, ball load and feed size distribution. W sag (kWh/t) = K(SPI / T800.5)nf sag
(3)
Figure 2. (a) SPI mill being discharged. (b) Typical Typical plot from a SPI test.
Te final set of equations contain 18 empirical factors, that clearly enabled the Minnovex SAG mill model to t experimental data well (Fig. 3). However, it is not clear from the literature if the predictive capability of the test is as good. e SPI test and its subsequent co-development as a geometallurgical geometallurgic al mapping tool with the help of Minnovex echnologies Inc. has been well documented in the literature by Chris Bennett of Minnovex (e.g., Bennett et al., 2001). SAGDesign test
One of the most recent laboratory SAG tests is the Standard Autogenous Grinding Design or SAGDesign test developed by Starkey et al. (2006) to overcome the limitations of the SPI test for SAG mill design. e SAGDesign Consulting Group consists of Outokumpu Technology Inc., Dawson Metallurgical Laboratories Inc.
Figure 3. Minnov Minnovex ex SAG model calibration calibration to plant data.
GeMIII (Amira P843) echnical echnical Report 1 – February, 2008
13.4
and Starkey & Associates. e test was primarily designed to estimate the full size SAG mill pinion specic energy (kWh/t) needed to grind a given ore from F80 150 mm to P80 1.7 mm. e SAG mill pinion energy can then be used to size the mill and motor to treat a target throughput or estimate the expected throughput given an existing mill. e test requires approximately 10 kg of half-core crushed to 80% passing 19 mm. Grinding is then done in a 0.5 m diameter SAG mill to 80% passing 12 mesh (1.7 mm). e apparatus is shown in Figure 4. e SAG ground product is then used for a Bond BMWI test. e SAGDesign test was designed to duplicate industrial mill operating parameters, namely 26% load, 11% steel, 15% ore (constant volume), and 76% critical speed. e mill was then sized so that 4.5 litres (~7 kg of siliceous ore) would be sucient for one test. Eight 1.5 inch (38mm) square lifter bars were added to match the size of the ore and balls. Te ball charge is a half and
Figure 4. SAGDesign test mill.
half mixture of 51 mm- and 38 mm-diameter mm-diameter grinding balls. e SAG stage feed size was selected to be the same as for a MacPherson Autogenous Grindability est or 80% passing ¾ inches (19mm). e SAG test produces a product size that is 80% passing 1.7 mm, using repeated grinding cycles with removal of the minus 1.7 mm nes from the batch charge after each cycle. Te number of revolutions of the mill to achieve this end point is the SAGDesign SAG grinding result. It is expressed as revolutions, not minutes so as not to confuse the test with an SPI test where the result is measured in minutes. Soft ores typically require less than 300 revolutions, whereas hard siliceous ores may need over 2000 revolutions to achieve the target grind size. e SAG Mill Pinion Energy is estimated using a linear calibration equation expressed as: SAG Mill Pinion Energy, kWh/t = Revs × (16000 + g)/(447.3 g)
(4)
where g is the weight weight of the ore ore tested, i.e. 4.5 litres of ore. ore. Te term g accounts for the effect of ore specific gravity on the specic power requirement in SAG milling. A higher power draw results from a heavier charge resulting from a higher specific gravity given the same ore volume. However, to maintain the constant rock volume, the tonnage ground is greater for increasing specific gravity or weight, which is reflected in the divisor of the equation and hence reduced kWh/t. 16,000 g is the weight of the steel ball charge used in the test. Reproducibility for SAG grinding is claimed to be ±3% for duplicate tests on the same ore. Te test has been used for predicting throughput as well as new plant design. An example of the test results and data reduction for a new mill design is shown in able 1.
Table 1. Example of SAGDesign SAGDesign test mill results – New design
13.5
Te claim by the developers that the test can accurately estimate the full size SAG mill energy needed to grind from F80 150 mm to P80 1.7 mm is questionable since the test uses a standardized feed size, whereas in practice the SAG feed size has a signicant impact on the power requirements and hence throughput rates. Hence the test cannot be expected to yield accurate results unless the test and industrial operating conditions are similar.
• 0.2 to 0.3 hard ore (AG, ABC) • 0.3 to 0.6 medium medium hard hard ore (SAG, SABC) • 0.8 to 1.5 friable ore (SAG) where ABC refers to an AG mill + Ball mill circuit with pebble Crusher, and SABC refers to a SAG mill + Ball mill circuit with pebble Crusher.
Compression/impact tests
Ore abrasion tests
e conventional Bond abrasion test measures how much a given rock-type will abrade steel (i.e., how much steel wears away). It is used to determine steel media and liner wear in crushers, rod mills and ball mills (Bond, 1963). Tere is no correlation for wear in Autogenous grinding. e test, developed by Allis-Chalmers, consists of a rotating drum, into which dry ore samples are placed, with an impact impact paddle mounted on a centre shaft rotating rotating at a higher speed than the drum. Te paddle is made of standard alloy steel hardened to 500 Brinell. e abrasion index Ai is determined from the weight loss of the paddle under standard operating conditions. Te test requires 1.6 kg of -19.1+12.7 mm ore. Bond developed a set of correlations using Ai to predict the wear rate in lb of metal wear / kWh of energy used in in each comminution process. process. For example, the wear rate for ball mills and crushers is estimated using equations expressed as: Wet W et Ball Mills Mills
Balls Liners
lb/kWh = 0.35(Ao-0.015)0.33 lb/ kWh = 0.026(Ai-0.015)0.30
Crushers
Liners
lb/ kWh = (Ai+0.22)/11
In complete contrast to tumbling tests, where a distribution of rocks is simultaneously tested in a device which is used to infer rock properties, there are a number of tests in which single specimens are “squ “squeezed” eezed” until they break; direct measurements of the material strength are obtained from the tumbling tests. Examples are, the Brazilian, UCS and Point Load tests. A subset of such tests are so-called impact tests which can also be considered as squeezing the specimen, only in this case the squeezing is done very rapidly via an impactor. Such tests include the Hopkinson bar, Bond crushing work index, JK Drop-weight, SMC and JKRBT tests. A number of these are reviewed in the following sections. Bond Crushing Work Index test (CWI)
e Bond Crushing Work Index is used to estimate crushing power requirements requirements (Bond, 1946). It is used for representativee rock specimens in the size range -76+50 mm representativ which are broken under the impact of twin pendulums, each weighing 13.6kg; it is recommended that at least 20 rocks are broken, ~10kg, during the derivation of each CWI. e input energy of the twin pendulums is increased by progressively raising their release height. Eventually a height is reached where rock breakage occurs. Te energy to achieve breakage is converted to CWI (kWh/t) as follows: Eb = K (1 - cos α) (5)
e JKMRC abrasion test measures how much steel or rocks will abrade a rock-type (i.e., how much rock wears away). It is used to assess ore amenability for SAG or AG milling. However this is clearly not the same test as the Bond abrasion test. e ore abrasion resistance is measured using a procedure based on a tumbling test as part of the standard JKMRC AG/SAG mill ore characterisation test CWI = 53.49 (E/t)/SG (6) work. is test requires 3 kg of -55+38 mm rocks which are tumbled for 10 minutes in a 30 cm diameter mill at 70% where Eb = Bond crushing crushing energy for an individual individual rock critical speed. e amount of nes generated is expressed (J, average of 10 breaks) as the ta parameter, where ta= t10/10. Smaller values of ta K = apparatus constant (82) indicate more resistance. Te test result is typically used α = the angle through which the pendulums fall to select the appropriate SAG milling conguration, for (degrees) example: SG = the specic gravity of the individual rock t = the thickness of the rock specimen (mm)
GeMIII (Amira P843) echnical echnical Report 1 – February, 2008
13.6
Equation (6) is only valid for rocks in the specied size range. Tere is limited published data available to determine the accuracy of this test for predicting crushing power requirement. An initial assessment of published literature suggests there is no relationship between UCS and CWI, which is not surprising if the fracture frequency in the test samples is high. Also since the CWI is an estimate of the energy required to achieve breakage, the test has poor precision due to operator-dependency (Doll et al., 1999; Angove and Dunne, 1997). The unconned compressive strength test (UCS)
Tis test is usually carried out as standard when characterising new ore bodies. It is not used for any predictive purposes in comminution but typically is an indicator of whether an ore is likely to be easy or difficult to treat. It is used by crusher manufacturers manufacturers to determine the expected machine stresses, and also in the drill-andblast, and geotechnical elds. e UCS test requires cylindrical core specimens with a length:width ratio of 2.5–3.0 and a diameter of about 50 mm; 20 pieces are normally recommended (~2.5 m of NQ core) e end of a specimen needs to be ground flat perpendicular to the long axis. e sample is placed in a test holder that is mounted in a press which stresses the rock until fracture occurs (Fig. 5). e UCS is then determined using the equation: σUCS
= Ppeak /A o
Figure 5. Specimen Holder used in Unconfined Compressive Strength (UCS) Test
(7)
where σUCS = Unconned Unconned Compressive Strength (MPa) Ppeak = Peak compressive load (N) Ao = Averag Averagee cross-sectional area of the specimen (mm2) Results for a given rock-type tend to be highly variable, which may explain why it has failed to show any signicant correlation with comminution machine performance (Doll et al., 2003). Much of this variability is related to the presence of structural features in the samples, which tend to induce premature failure when stressed. Figure 6 shows data from 11 mines around the world (Fluor Wright W right database) which suggests there is no universal correlation between UCS and Bond BMWI. However, in a small subset of rock types which have minimal rock fractures there may be a relationship as shown by Doll et al. (2003). Te level of variability in the results may explain why data obtained from UCS tests have failed to show any significant correlation with comminution machine performance. Point Load test (PLT)
e Point Load test (PLT) is a geomechanical test used to measure rock fragment strength (Broch and Franklin, 1972). Historically Historically the point load test was used as a quick and simple method to predict tensile and compressive strength e.g., UCS (Butenuth, 1997). e PLT measures the Point Load Strength (Is) of the rock sample. It uses the ISRM standard procedure procedure (ISRM, 1985). e test can be performed with portable equipment or using a laboratory
Figure 6. Plot of Bond ball mill Work Index (BMWI) versus UCS
13.7
testing machine, hence may be conducted in either the field or the laboratory. It consists of a two-column loading frame with two point-sha point-shaped ped “platens” between which the rock is placed (Fig. (Fig. 7). One of the platens is eectively stationary (though its initial starting position can be adjusted) whilst the other is free to move through the application of pressure, delivered via a hand pump and piston arrangement. arrangement. As the hand pump is activated activated the pressure and hence, force applied to the rock, is increased and eventually causes causes the rock to fail. Te peak pressure pressure applied is indicated on a pressure gauge. Reichmuth (1968) and Broch & Franklin (1972) carried out extensive testwork using the point load tester and developed the initial formulae for computing a strength index (Is) from the measured pressure. Brook (1985) subsequently modied the equation to account for dierent rock shapes (e.g., half drill core). Drill cores were found to provide the most consistent data and the test was therefore originally specied for 50 mm-diameter core, leading to the common standard designated as Is(50). Despite corrections for shape, size eects were apparent in the data and hence a correction was developed to convert data into Is(50) Is(50) equivalent. Rock samples samples may be in the form of either core (diametral and axial tests), cut blocks (block test), or irregular lumps (irregular lump test). Te irregular lump test offers the greatest convenience,
as sample preparation is not required. ypically, it is recommended that 25 pieces of rock (~0.25 m ½ NQ core) be used in each PLT. Te standard formula for point load strength calculation is as follows: Is(50) = FP/D2e
(8)
where F = size correction correction factor = (De/50)0.45 P = force at failure and is calculated from the pressure and geometry of the hydraulic system De = (4A/π)0.5 A = minimum cross-sectional cross-sectional area area of the specimen Te units of the point load strength Is(50) are MPa and whereas the test is considered to cause tensile failure it is converted to compressive strength (i.e., UCS) by: UCS = 24 × Is(50) e PLT is extensively used in drill-and-blast and geotechnical fields but to date has not been used with respect to comminution. However, recent data suggests this index may provide a useful guide to comminution behaviour (Fig. 8).
fixed platen
rigid frame
moveable platen
pressure gauge with max. reading pointer
hydraulic piston
hand pump release valve
Figure 7. Point Load Tester and schematic schemati c of test set-up
GeMIII (Amira P843) echnical echnical Report 1 – February, 2008
13.8 18
640 640
17
600 600
t 16 / h W k 15 G A S
560 560
14
480 480
13
440 440
h p t 520 G 520 A S
400 400
12 .0 44 .0
55..00
6..0 0 6
7..0 0 7
8.0
t10 = A × (1 – e–b
×
Ecs)
(9)
where A and b are are ore specic parameters, and Ecs Ecs is the specic breakage energy (kWh/t).
9 9..0 0
IS(50) - MPa
Figure 8. Correlation between point point load strength Is(50) and SAG mill performance
Drop Weight Test (DWT)
e JKMRC drop weight apparatus and its associated data reduction technique were developed so that the relationship between specific energy input and resultant product size could be determined (Napier-Munn et al., 1996). Tis relationship is used in size-reductionmodelling for crushers and mills in the JKSimMet mineral processing (Wiseman and Richardson, 1991). e test apparatus comprises an impactor (Fig. 9); the mass of the impactor can be varied together with the height from which it can be dropped. dropped. A rock specimen specimen is placed on an anvil and is broken by the impactor. Te broken pieces are collected and sized. By varying the drop height and mass (input energy, J or kWh) as well as the rock size a range of specic impact energies (kWh/t) can be generated. ese
Figure 9. JKMRC Drop Weight Weight Tester Tester
are then related to the size distribution of broken products through the use of a so-called t10 parameter that is defined as 1/10th passing the original particle size. It is normally related to the specific energy using the following equation:
e typical test requires 65kg of -75+10mm rocks (~ 5 m of PQ core). e t 10-Ecs curve is inuenced by the size of the fragments tested. For this reason reason the full DWT examines ve size fractions ranging from 63mm down to 13.2 mm (able (able 2). In the standard test 3 energy levels are used for each size fraction, with 10–30 pieces of rock in each size fraction per test. Table 2. Standard DW test conditions. Size Range (mm) -63+53 -45+37.5 -31.5+26.5 -22.4+19 -16+13.2
Number of particles/ test 10 15 30 30 30
Expected Ecs (kWh/t) 0.10 0.25 0.10 0.25 0.25 0.25 0.25
0.40 1.0 1.0 1.0 1.0
2.5 2.5 2.5
Figure 10a illustrates the results from a test on one size fraction, -16+13.2 mm. e slope of this plot at the origin, A*b, is related to the strength of the rock; an A*b or slope with a larger gradient is indicative of weaker rock. Te parameter A is the t10 asymptote sill or maximum degree of breakage. Tis limit indicates that at higher energies little additional size reduction occurs as the Ecs is increased, i.e., the size reduction process becomes less efficient. Te parameter parameter b is related to the shape of the t10Ecs curve, with lower values of b indicating a harder ore. As there is some interaction between A and b in the impact breakage equation (9), JKMRC uses A*b for comparison as it is better defined. It is possible that both the A and b parameters could be related to rock texture (e.g., grain boundaries, mineralogical composition/associations and micro-cracks). Tis has yet to be established and is the subject of further work. In the example shown in Figure 10, 5.4% of the product from the sample broken at 0.25 kWh/t was ner than 1.45mm, based on the initial average size of 14.5mm. Similarly, the sample
13.9
(a)
T10 (%) 43.2
19.8
5.4
A*b =23 (hard ore) 0.25
1.0
2.5
Specific Energy (kWh/t)
(b)
Figure 10. (a) he relationship between between fines produced and specific breakage breakage energy for a single particle size Weight test. (hard ore). (b) Example t10 - Specific Energy relationship from a standard Drop Weight
broken at 2.5 kWh/t produced 43.2% of nes (smaller than 1.45 mm). e maximum degree of breakage, A, for this sample was 100%; b was 0.23, giving a relatively low value for A*b of 23 indicating a very high resistance to impact breakage. A complete set of results comprising 15 tests is used to determine the A and b parameters of Equation 9. is is carried out using proprietary JKMRC Drop Weight Test software that is routinely used to t the ore specific parameters to a set of given t 10-Ecs results. An example of a complete set of Drop Weight test results is shown in Figure 10b. Drop Weight test A and b parameters can be used to compare the impact hardness of different ore types (Fig. 11). Note that A and b values are not used directly to estimate mill energy requirements but are used in JKMRC
100 90
A=40 & b=0.3 b=0.3
80
A=55 & b=0.8 b=0.8 A=70 & b=1.3 b=1.3
70 ) 60 % ( 0 50 1 t 40
soft Decreasing resistance to impact
30 20
hard
10 0 0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
Ecs (kWh/t)
Figure 11. Example of comparative comparative results, reflecting different A and b values.
GeMIII (Amira P843) echnical echnical Report 1 – February, 2008
13.10
Figure 12. Relationship between between A*b and PLI, with GeM GeM EH512 sample results shown in pink.
100 t2 80
t4
Each vertical section represents a size distribution
60 t10 40 t25 20
t50 t75
0 0
10
20 Breakage Index, t
30
40
(%) 10
50
Figure 13. Standard JKMRC Breakage Map showing tn vs t10 for n=2, 4, 10, 25, 50 and 75.
2 1 3
Figure 14. Sample pieces cut from from 50 mm quartered core. core.
Figure 15. Comparison of results: SMC tests on pieces of ¼-core versus DWT on irregular lumps of rock ( t10-Ecs). Source JKTech website (2008).
13.11
AG/SAG models to simulate the mill performance. Extensive data from the Amira P483 Mine-to-Mill project has shown that A*b can be correlated quite well with the point load index, as illustrated in Figure 12. is is most likely due to the dependence of the b parameter on the energy to rst fracture (Tavares and King, 1997). Te A*b parameter has been shown to be a useful rank of ore hardness and almost all of the major mining companies use the DWT to obtain ore impact breakage characteristics. ere are now more than 20 JKMRC drop weight testers installed world-wid world-widee and the DWT is one of the standard ore breakage characterisation methods. Prediction of product size distribution
In JKMRC comminution models, the key output from a breakage event is expressed in terms of the t 10. Te parameter t10 is significant as it can be used to estimate the full product size distribution given the ore breakage map or family of t curves from t2 (1/2 of the mean initial size) to t75 (1/75th of the mean initial size; Fig. 13). Spline interpolation can then be used to reconstruct the full size distribution curve, given the six points on the curve and initial particle size (Napier-Munn et al., 1996).
SAG mill comminution test (SMC) e SMC test (Morrell, 2004b) is similar to the Drop Weight W eight Test Test and uses the same apparatus. However, However, the SMC test was designed to make use of quartered drill core, i.e., core which has been cut into a number of identical pieces using a diamond saw (Fig. 14). Original
core diameters up to 85 mm (PQ) are suitable. Crushed core can also be used. For example, the minimum sample weight for NQ (47.6mm) diameter core core samples is 1.7 kg kg (~0.8 m ½-core), based on the typical SG of 2.8. e key elements of the SMC procedure are as follows: • ve specic specic energy energy levels are used – 0.25, 0.50, 1.0, 2.5 and 3.5 kWh/t • 20 quartered quartered core core pieces are broken broken at each energy energy • three orientations are used at each energy energy (7+7+6 (7+7+6 = 20; Fig. 14) Broken fragments from all three orientations are sized on a single sieve that defines the t 10 size. Te percentage of undersize from sieving the broken products is plotted against the input energy, in a similar way to the t 10 versus specic energy used in the DWT data reduction technique. In the SMC test proprietary algorithms are used to estimate the DWT equivalent A and b parameters and an impact strength index, called the Drop Weight index (DWi; kWh/m3), from the gradient of the percent of undersize versus input energy trend. As the ore impact strength increases so does the value DWi. One of the key assumptions in the SMC test is that the average of the results from three orientations is expected to reect the DWT result on irregular lump particles of similar size (Fig. 15). is is likely to be dependent on texture/bedding planes within the samples, however published documentation supporting this assumption is limited (JKTech, (JKTech, 2008) and a clear assessment has not been possible. Independent research at the JKMRC on Ernest Henry core samples shows a strong effect of orientation on the degree of breakage in Drop Weight tests as shown in Figure 16: orientation 1 produces the most breakage, 3
70
60
50
40
Decreasing resistance to impact 30
' orient -1
20 orient -2 orient -3
10
Figure 16. Effect of sample orientation orientation on degree of breakage in DWT. Orientation of samples as in Figure 14.
ave orient
0 0. 0
0. 5
1. 0
1.5
2. 0
2.5
3. 0
3.5
4.0
S p e c if i f ic ic I m p a c t E n e r g y , E c s (k (k W h /t /t)
GeMIII (Amira P843) echnical echnical Report 1 – February, 2008
13.12 Results for 1.5 kWh/t - DWT Cubes vs Fragments 30 brk EH1 DWT 1.5
25
cube EH1 1.5r DWT
d 20 e n i a t e R 15 % s s a M10
5
Figure 17. Comparison of Drop Weight Weight Test Test breakage results for cube-shaped and irregular lumps of rock.
0 0.01
0.1
1
10
10
Sieve Size (mm)
10
) m . u c / h W k ( I W D x e d n I C M S
8
6
4
2
0 0
2
4
6
8
Point Load Strength (MPa)
10
12
Figure 18. Relationship between DWi DWi derived from SMC tests and Point Load Strength
Inspection window
Figure 19. Photograph of the prototype JKRBT device, with rotor-sta rotor-stator tor showing through the inspection window. window.
13.13
the least. leas t. In addition, additio n, Drop Weight Weight Tests Tests on Ernest Er nest Henry Henr y 13 mm cubes versus 16 x 13.2 mm fragments suggest that cubes break more than irregular lumps of similar size/ volume (Fig. 17). Hence, the assumption that the average of the results for the three orientations used in the SMC test is equivalent to Drop Weight Test results for irregular lumps of rock is called into question. e DWi derived from the SMC test has been used to develop an empirical model for the AG/SAG mill specic energy, as per the Minnovex SPI approach noted previously:
rotor-stator system for rapid breakage characterisation. Te operating system consists of a vibrating feeder, a rotor-stator impacting device plus drive system, and an operation control unit. Like the Drop Weight tester, the JKRBT device also requires the ore particles to be pre-sized into narrow fractions. Particl Particles es of the selected size are fed into the rotor-stator impacting impacting system via a vibrating feeder. feeder. Te vibrating feeder controls the feed rate to ensure that breakage takes place in a single particle mode. After impact breakage, the product is collected from a container underneath the rotor-stator system. Te t10 values are determined using the same data reduction techniques noted above for the Drop Weight test. e JKRBT device can generate the standard AG/ SAG mill parameters A and b in 1/10th of the time it takes using the standard Drop Weight test. Validation and commercialization of the new device is in progress. ests carried out to date have confirmed the device offers a rapid method for determining the hardness of drill core samples within the context of the GeMIII project. Te new JKMRC breakage characterisation device was developed through partial nancial support from the AMIRA P9N project. Te device overcomes some of the limitations in the existing impact tests, including the precision of the energy input, time required to run the test, and the smallest particle size that that can be readily readily tested. tested. Comparativ Comparativee breakage tests using the new device and the traditional JKMRC Drop Drop Weight Weight tester suggest that the two devices generate the identical breakage–energy relationship for the same ore of the same sam e size (Fig. 20). Similarly, when the tests were compared across the full size and energy range,
Specic Energy = K.ƒ(F80,DWi,BL,SPEED,L/D) where F80 = 80% passing passing size for the the mill feed BL = ball load SPEED = mill speed L = mill length D = mill diameter e DWi, like the parameter A*b, can be correlated with point-load point-load strength, as shown shown in Figure Figure 18.
JK rotary breakage test (JKRBT) e new JK rotary breakage test (JKRBT) characterisation device employs a precise and accurate control of energy and can test particles across a wide range of sizes, from 1 to 30 mm. e prototype JKRBT device, shown in Figure 19, was designed and manufactured at the JKMRC pilot plant workshop in 2005 to test the concept of using a
40
120
(a)
35
D WT WT
R BT BT
(b)
100
30 80
25
) % ( 20 0 1 t
b * A
60
15 40
DWT
10
RBT
20
RBT repeat
5
Model Fit 0
0 0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
1
2
3
4
5
6
7
8
9
10 10
Sample
Ecs (kWh/t)
Figure 20. Comparison of (a) (a) t10 versus Ecs and (b) breakage parameters A*b as determined by JKRBT and DWT DWT tests
GeMIII (Amira P843) echnical echnical Report 1 – February, 2008
13.14
the statistical analysis indicated that the two machines can generate identical breakage parameter A*b values. Figure 20 compares the A*b values of 10 ore types determined by industrial JKRBT and Drop Weight tests. Over the life of the GeMIII project, the objective objective is to provide A*b measurements on 1500 to 2000 drill core samples represented in the Level 3 Geometallurgical Matrix. e JKRBT procedure therefore has been simplied to suit this task and the small sample volumes being tested as part of the physical testing protocol. Hence, a single size fraction (which could be drawn from a much smaller volume of rock) was selected with the understanding that the A and b parameters determined from the reduced test would satisfy the GeMIII requirements for comparative testing. Te procedure adopted within GeMIII is as follows: follows: 1. e size fraction fraction 11.2 x 9.5mm is being being used as it is available in sucient quantities post crushing of NQ and HQ half-core intervals. 2. Four specic specic energy levels are being used: 0.2, 0.2, 0.5, 1.0 and 2.0 kWh/t. 3. Te breakage products are sized to determine determine the t10 percentage (i.e., < 1mm). 4. Te A and b parameters are determined using standard JK data reduction techniques. Te GeMIII project has used the prototype JKRBT device to characterise the impact hardness of ore samples supplied by Level 1 sponsors. e approach has generated an excellent rst pass denition over a very rapid time scale, as illustrated by the chart in Figure 21 which shows
the frequency distribution of almost 150 test results for ve Cadia East drill cores. Te single particle size A*b parameters can be corrected to equivalent full Drop Weight Test parameters using established empirical rules governing the effect of particle size or supporting tests on other particle sizes. e corrected A*b estimates, combined with Bond BMWI data, can be readily converted to mill throughput predictions given the new or existing mill design and operating conditions and JKMRC proprietary models, as illustrated in Figure 22 for a single Cadia East drill hole. Clearly testing of the JKRBT device is showing signicant promise and should enable the GeM III project to provide vital geostatistical information on the throughput capacity of a new or existing mill circuit.
Ranking of comparative testing methods Te question of which test can best suit the GeM III requirements for comparative testing was addressed by ranking the above tests plus EQUOtip (Section 5) on the basis of the key needs for such a test. Ideally the comparative test needs to be: • Low cost • Fast • Relevant to comminution performance and rock texture • Applied on a small sample size • Reproducible.
Figure 21. Summary of Phase 1 Results for 5 Cadia East drill holes – as generated from the JKRBT device.
13.15 90
Mill Dimensions 36’x18’ 1740
80
Ball Load 12% 70
78% of Critical Spee Speed d
60
b * A
25% Mill Load
1520 1310 1090
50
870
40
650
30
T h r o u g h p u t ( t p h )
440
20 10 0 1100
1150
1200
1250
1300
1350
1400
1450
1500
Depth (m)
Figure 22. Down-h Down-hole ole A*b and mill throughout profile for a drill hole from Cadia East.
Each test was ranked according to the above criteria on a scale of 1 to 10, and a combined score for each test was calculated by adding the ranks together and normalizing the result to a percentage-scale (Table 3). Cost was not included in the ranking as the JKRBT device has yet to be commercialized. Te comparison suggests that the new JKRBT device appears to be the best choice, followed by SMC and PLT/SPI. is is not surprising as the standard Drop Weight Weight Test Test (DWT) is time consuming consum ing and requires requi res a large sample. e SMC test is much faster than the DWT,, but still DWT stil l too slow for GeMIII project requirements. e PLT PLT test is too tedious for use with large numbers of samples and is imprecise for the application; in addition, the relevance of the SPI is questionable. On the basis of this evaluation, the JKRBT device appears the ideal choice for the comparative testing tasks within the GeMIII project.
References Angove, J.E., and Dunne, R.C., 1997, A Review of Standard Physical Ore Property Determinations: World Gold Conference 1997, Singapore, 1-3 September September.. Bennett, C., Dobby, G.S., Kosick, G., 2001, Benchmarking and Ore Body Profiling – the keys to effective production forecasting and SAG circuit optimization: SAG 2001 Conference, Vancouver, September, Vol I, p. 289-300. Bond, F.C., 1946, Crushing Tests by Pressure and Impact: Trans SME/AIME, SME/AIME , v. 169, p. 58-66. Bond, F.C., 1952, The Third Theory of Comminution: Trans AIME, 1952, v. v. 193, p. 484-494. Bond, F.C., 1961, Crushing and Grinding Calculations: AllisChalmers publication, no. O7R9235B. (also in British Chemical Engineering, v. 6, nos. 6 and 8). Bond, F.C., F.C., 1963, Metal Wear Wear in Crushing and Grinding: Allis-Chalmers Publication Publication no. 07P1701. Broch, E., and Franklin, J.A., 1972, The Point Load Test: International Journal of Rock Mechanics, Minerals & Science, v. 9, p. 669-697.
able 3. Ranking of available comminution tests for their suitability in GeMIII project comparative testing. est Point Load UCS EquoTip DWT SMC Bond Abrasion SPI SAGDesign JKRBT
Spe peed ed 9 3 10 2 6 6 8 6 4 8
Samp Sa mple le Precision 8 5 3 5 10 4 3 6 8 8 6 7 6 5 8 8 6 8 8 9
Relevance 7 7 2 9 9 9 5 7 8 9
Rank 73% 45% 65% 50% 78% 70% 60% 72% 56% 85%
Position 3 8 5 7 2 4 6 3 6 1
GeMIII (Amira P843) echnical echnical Report 1 – February, 2008
13.16
Brook, N., 1985, The Equivalent Core Diameter Method Starkey, J.H., and Dobby, G., 1996, Application of the of Size and Shape Correction in Point Load Testing: Minnovex SAG Power Index at Five Canadian SAG International Journal of Rock Mechanics, Minerals, Plants: Proceedings of the Conference on International Science and Geomechanics, Abstract 22, p. 61-70. Autogenous and Semi-autogenous Semi-autogenous Grinding Technology Butenuth, C., 1997, Comparison of Tensile Tensile Strength Values of (SAG 1996), Vancouver, Vancouver, British Columbia, p. 345-360. Rocks Determined by Point Load and Direct Tension Starkey, J.H., Hindstrom, S., and Nadasdy, G., 2006, SAGDesign SAGDesig n Tests: Rock Mechanics and Rock Engineering, v. 30, no. testing – What is it and Why it Works: Proceedings 1, p. 65-72. of the Conference on International Autogenous and Doll. A., Barratt, D.J. and Wood, K., 2003, Comparison of Semi-autogenous Grinding Technology (SAG 2006), UCS to Bond Work Indices,
. Verwa erwaal, al, W. W. and Mulder, A., 1993, Estimating Est imating rock strength stre ngth with Dobby, G., Bennett, C., and Kosick, G., 2001, Advances in the EQUOtip hardness tester: International Journal of SAG Circuit Design and Simulation Applied to the Rock Mechanics and Mining Sciences & Geomechanics Mine Block Model: Proceedings of the Conference Abstracts, v. v. 30, p. 659-662. on International Autogenous and Semi-Autogenous Walters, Walters, S., and Kojovic, T., T., 2006, Geometallurgical Mapping Grinding Technology (SAG 2001), Vancouver, British and Mine Modelling (GEMIII) – the way of the Columbia, v. 4, p. 221-234. future: Proceedings of the Conference on International ISRM - International Society of Rock Mechanics, 1985, Autogenous and Semi-autogenous Semi-autogenous Grinding Technology Suggested Method for Determining Point Load (SAG 2006), Vancouver, v. IV, p. 411-425. Strength: International Journal of Rock Mechanics, Wiseman D.M., and Richardson J.M., 1991, JKSimMet - the Minerals, Science and Geomechanics, Abstract, v. 22, mineral processing simulator. Proceedings 2nd Can no. 2, p. 51-60. Conf on Comp Applications in the Min Ind , (Eds. Paulin, Levin, J., 1989, Observation on the t he Bond Standard Grindability Grindability Pakalniss and Mular), Pakalni Mular), University University of British Columbia Columbia Test, and a Proposal for a Standard Grindability Test for and CIM, v. II, p. 427-438. Fine Materials: SAIMM, v 89, no.1, p. 13-21. JKTech, JKT ech, 2008, . MacPherson, A.R., 1989, The Development of Autogenous Grinding and Semi-Autogenous Grinding: Proceedings of the Conference on International Autogenous and Semi-Autogenous Grinding Technology, (SAG 1989), Vancouver, British Columbia, p. 5-7. MacPherson, A.R. Consultants, 1999, Glossary of Comminution and Ore Hardness Terms, Lakefield, Ontario, Canada. 32p. Morrell, S., 2004a, An Alternative Energy-Size Relationship To To That Proposed By Bond for The Design and Optimisation of Grinding Circuits: International Journal of Mineral Processing, v. 17, no. 3, p. 437-445. Morrell, S., 2004b, Predicting the Specific Energy of Autogenous and Semi-Autogenous Mills from Small Diameter Drill Core Samples: Minerals Engineering, v. 17, no. 3, p. 447-451. Napier-Munn, T. T. J., S. Morrell, Morrison, Mor rison, R., R ., Kojovic, Kojovic , T. T. , 1996, Mineral Comminution Circuits: Their Operation and Optimisation: Brisbane Australia, JKMRC University of Queensland, JKMRC Monograph Series in Mining and Mineral Processing 2, 413 p. Reichmuth, D.R., 1968, Point-Load Testing Testing of Brittle Materials to Determine Tensile Tensile Strength and Relative Brittleness, Proceedings 9th US Symposium Rock Mechanics, University Universi ty of Colorado, p. 34-159. Tavares, L.M., and King, R.P., 1998, Single-Particle Fracture under Impact Loading: International Journal of Mineral Processing, v. 54, p. 1-28. Starkey, J.H., Dobby, G., and Kosick, G., 1994, A New Tool for SAG Hardness Testing: Proceedings Canadian Mineral Processor’s Conference, Ottawa, p. 12.
