Professional Documents
Culture Documents
170 180MMP 06 058PothinaPROOF
170 180MMP 06 058PothinaPROOF
170 180MMP 06 058PothinaPROOF
net/publication/261059813
CITATIONS READS
9 7,416
4 authors, including:
Some of the authors of this publication are also working on these related projects:
All content following this page was uploaded by Vladislav Kecojevic on 31 March 2014.
MINERALS & METALLURGICAL PROCESSING 171 Vol. 24, No. 3 • August 2007
PROOF COPY
1 the 80% passing size of the feed and product size distributions, particles on power consumption is not considered. This rep- 1
2 limiting its accuracy in representing the size reduction process. resents a handicap because irregularly shaped feed particles 2
3 Another drawback of this formula is that the machine-specific can break more easily than others due to stress concentrations 3
4 properties that affect energy consumption are not considered. at their irregular projections (Hogg, 2003). 4
5 After conducting experiments on a wide range of particles and The power drawn by a crusher depends on material specific 5
6 industrial comminution devices, Hukki (1961) concluded that parameters like feed size, shape, hardness and distribution, 6
7 no single energy–particle-size relationship was adequate for feed rate and machine specific parameters like crusher settings 7
8 all ranges of material sizes. and liner profile. The product size distributions of gyratory 8
9 According to Lynch (1977), energy losses due to transmission crushers/cone crushers mainly depend on crusher settings 9
10 and internal friction of the crushing device may be quantifiable, (gap) and to a lesser degree on feed rate and size, liner profile 10
11 but assessing the energy lost due to internal friction between the and condition (Anderson 1988; Anderson and Napier-Munn, 11
12 particles and their elastic and plastic deformations is not pos- 1988). Therefore, all these parameters need to be considered 12
13 sible. Therefore, understanding comminution/crushing cannot to arrive at the representative values for power consumption 13
14 be based on energy–size-reduction theories alone. Instead, this and product size distribution. Knowing the extent of influence 14
15 study asserts that a comprehensive understanding the effects of each parameter on power consumption is important. 15
16 of feed and product size distributions and operating variables Scientists at the Julius Kruttschnitt Mineral Research Cen- 16
17 of the crushing device that causes size reduction is necessary tre (JKMRC) conducted extensive research over the last 25 17
18 to accurately represent the process. years on size reduction, including the analysis of power draw 18
19 The number of parameters that affect energy consumption in comminution circuits (Morrell et al., 1992; Napier-Munn 19
20 in crushing is very large. However, only a few parameters af- et al., 1996). Using pendulum tests, Narayanan and Whiten 20
21 fect energy significantly. Therefore, a mathematical model that (1988) found a specific comminution energy (Ecs) required 21
22 considers only the significant parameters influencing energy for breaking a particular ore particle. Calculation of Ecs 22
23 consumption and reduces complexity without affecting the final values for a wide range of industrial ores/rocks resulted. The 23
24 result considerably represents an additional challenge. size distribution, represented using the t10 value, defined the 24
25 During size reduction, because particles undergo recurrent proportion of material passing 1/10th of original ore particle 25
26 breakage, industrial comminution operations can be explained size. A database of t10 values and their corresponding Ecs 26
27 as multiple breakage events (Epstein, 1948). The breakage rate values for a wide range of ores was the resulting development. 27
28 and product size distributions are also dependent on the type Narayanan and Whiten correlated the Ecs and t10 values with 28
29 of machine and method of applying force. Therefore, detailed the actual power drawn by industrial crushers. Using this 29
30 modeling of the comminution process would need to account provided a large database of values necessary to predict any 30
31 for these factors. crusher’s power draw. 31
32 It has been suggested that comminution models can be JKMRC also developed a simulation software package 32
33 classified as “black-box” models or fundamental models (Na- called JKSimMet to optimize industrial mineral processing 33
34 pier-Munn et al., 1996). The “black-box” models try to predict circuits (Awachie, 1983; Anderson, 1988). This package con- 34
35 product size distributions and energy expended from the feed tains a model that is capable of predicting the crusher power 35
36 size distribution, physical parameters of the feed material, and draw and product size distribution. JKMRC models include 36
37 particle breakage characteristics. Fundamental models, on the ore-specific parameters such as feed size distribution, hardness 37
38 other hand, consider each element of the comminution process, and feed rate and machine-specific parameters such as liner 38
39 such as interaction of ore particles within a machine and with profile and condition. However, these models depend on the 39
40 the machine largely on the basis of Newtonian mechanics. large database of ore-specific and machine-specific values 40
41 Several researchers (King, 1979; Herbst and Fuerstenau, 1980; generated over many years of research. For a typical industrial 41
42 Radziszewski and Tarasiewicz, 1989; Mishra and Rajamani, user, the overall procedure of using these various values and 42
43 1992; Inoue and Okaya, 1994) contributed to the development understanding the concepts behind them might prove very 43
44 of fundamental models. A main constraint for the slow progress complicated. Moreover, JKSimMet aims at overall optimiza- 44
45 in the development of these models has been the very high tion of the mineral-processing circuit rather than optimizing 45
46 computational requirements. individual units such as a gyratory crusher. 46
47 “Black-box” models can predict product size distribution as Apart from the development of simulation models for the 47
48 a function of mill/crusher feed size distribution and material whole comminution system, significant research has focused 48
49 characteristics and operating conditions using mass balance on developing optimum structural features of crushers. Wear on 49
50 equations to model the crushing process. The population balance critical parts of a crushing plant during its life affects production. 50
51 model is one of the “black-box” models proposed by Epstein Wear changes the profiles of crusher liners and even the volume 51
52 (1948), which has been applied in various size-reduction ap- of the crushing chambers. Svedensten and Evertsson (2001) 52
53 plications around the world (Herbst and Fuerstenau, 1968; developed a software package using visual C++ for optimizing 53
54 Kelsall et al., 1969; Austin and Brame, 1983; and Guillaneau plant performance, taking wear into consideration. Lindquist 54
55 et al., 1995). and Sotkovszki (2003) found that the nonlinear dependency 55
56 Lynch (1977) used Whiten’s model with certain modifica- between pressure and wear of cone crusher liners could be 56
57 tions to incorporate machine specific parameters for the crusher explained through variations in particle size and inaccuracies 57
58 at Mount Isa Mines Limited. However, this model was only in feed flow rates. Rosario et al. (2004a; 2004b) and Rosario 58
59 applicable to cone crushers and did not incorporate parameters and Hall (2006) used a laser-based tool for assessing wear in 59
60 such as the wear rate of liners and their profiles, which can gyratory crushers. As a result, a novel methodology to evaluate 60
61 influence energy consumption. Lynch predicted the power crusher liner profiles and its link to crusher performance was 61
62 consumption of the cone crusher from the size distributions of developed and new mantle profiles were designed. 62
63 feed and breakage matrices. In his model, the crusher power Comminution models developed and based on the compu- 63
64 draw is mainly due to the breakage process expressed by the tational power of computers are in general very complex and 64
65 breakage function/matrix. The effect of the shape of the feed require extensive effort to adapt them for a given crushing 65
MINERALS & METALLURGICAL PROCESSING 173 Vol. 24, No. 3 • August 2007
PROOF COPY
1 a particular crushing operation. This, in turn, allows the user Because Eqs. (2) and (3) are used for the same purpose, 1
2 to find the range of operating parameters that cause the least they are equated as 2
3 energy consumption. 3
W = Emt (4)
4 The Bond (1952) equation is used to scale the energy 4
5 consumption in crushing. But the energy required to crush a This can be rewritten as 5
6 ton of material differs from machine to machine. As an alter- 6
Wi × WSR = (V × A × PF) × t (5)
7 native, Bond’s work index can be related to the values that 7
8 represent energy consumption in industrial crushers by some If A × t = At = Wi × Asrmt, then 8
9 scaling factor. Relating the energy required to crush a ton of 9
Wi × WSR = (V × At × PF) (6)
10 any material found by Bond’s equation to the corresponding 10
11 amperage constant would help find a new energy-scaling for- 11
Wi × WSR = Wi × (V × Asrmt × PF) (7)
12 mula. Simply, this method finds a way to effectively relate the 12
13 constant that represents the degree of breakage at a particular 13
WSR = V × Asrmt × PF (8)
14 size distribution to the part that represents size reduction in 14
15 Bond’s equation. The method developed to achieve the task WSR 15
16 is the platform for full-scale model development. Asrmt = 16
17 Bond’s equation is V × PF (9)
17
18 where 18
19 1 1 Asrmt is an amperage constant that corresponds to size 19
W = [Wi ] × 10 × − (1)
20 P80 F80 reduction. 20
21 21
22 The right side of the equation has two parts: a portion that Because V and PF are constants for a particular operation, 22
23 represents the energy required to crush a ton of specific mate- Eq. (9) shows dependence of Asrmt on WSR. From the above 23
24 rial (Wi) and a portion that represents size reduction (WSR), equations, evidently, for every crushing process a unique WSR 24
25 where exists, and in turn, an Asrmt value exists. The Asrmt value gives 25
26 the fraction of amperage constant (At) responsible for size 26
1 1
27 WSR = 10 × − reduction in the total amperage drawn by the motor. Another 27
28 P80 F80 distinctive value for each size reduction process that is very 28
29 sensitive to feed size distributions and selection and breakage 29
30 Rewriting the equation yields processes that occur in the crusher is the amperage factor, a, 30
31 described in Lynch’s equation (Lynch, 1977). Relating these 31
W = Wi × WSR (2)
32 two variables (a and Asrmt) is crucial in finding the energy 32
33 In general, the F80 and P80 values for a particular crush- consumption for a particular crushing process. 33
34 ing operation are the average values for a wide range of size With the Minitab statistical package and using the calcu- 34
35 distributions resulting from blasting and crushing operations, lated values, a relationship between the amperage factor a 35
36 respectively, which makes WSR constant for a specific op- and the amperage constant Asrmt was established as will be 36
37 eration. This is one of the main drawbacks in using Bond’s noted below. 37
38 equation. In fact, different blasting configurations will result One of the important objectives of the amperage constant 38
39 in different fragment size distributions and different feed size model is the necessity of finding burden and spacing values 39
40 distribution values for the crusher. As mentioned above, dif- that can produce the fragment size distributions suitable for 40
41 ferent size distributions may have the same F80 values but the gyratory crusher. The assumption is that the feed particles 41
42 the corresponding P80 values differ at a particular CSS value resulting from the blasts are in a range between γlower and 42
43 for the crusher. γupper. The product size requirements of the gyratory crusher 43
44 Each size reduction process at a particular feed size dis- are assumed to be in the range of Ωlower to Ωupper. Burden and 44
45 tribution, resulting from the particular burden and spacing spacing values vary between Blower and Bupper and Slower and 45
46 configurations, has its own unique set of F80, P80 and CSS Supper, respectively. For each feed size distribution, the CSS 46
47 values. Values for WSR and W are unique for different burden, varies between CSSlower and CSSupper. For each set of B, S and 47
48 spacing and CSS configurations. Based on these facts, finding CSS, a unique set of a and Asrmt values arise. 48
49 the equation that predicts the energy utilized in the crushing After establishing the relationship between the amperage 49
50 operation for the given feed size distribution is essential. factor and the amperage constant, determining the energy 50
51 An alternative way to calculate the energy required for scaling formula for any feed size distribution at any given 51
52 crushing the material is given by Ohm’s law as CSS is possible. For example, for a particular set of B, S and 52
53 CSS values, a corresponding Asrmt value exists. Knowing the 53
Emt = (p × t) = (V × A × PF) × t (3)
54 value of Asrmt, the energy required to crush material (Emt) is 54
55 where calculated by 55
56 p is power required for crushing the material, 56
Emt = Wi × V × Asrmt × PF (10)
57 t is the time taken for crushing, 57
58 V is line voltage of the motor drive of the crusher, The required power (Psrhr) is calculated by 58
59 A is amperage drawn by the machine to crush material 59
60 and E × LF (11) 60
Psrhr = mt × RC
61 PF is the power factor. DE 61
62 where 62
63 The PF or cosf is the ratio between the real power supplied LF is the lump factor, 63
64 and apparent power received by the motor. Loss in transmission DE is the drive efficiency and 64
65 is due to phase angle (f) differences. RC is required production per hour. 65
MINERALS & METALLURGICAL PROCESSING 175 Vol. 24, No. 3 • August 2007
PROOF COPY
1 crusher has a closed-side setting adjustable 1
2 from 110 to 140 mm (4.3 to 5.5 in.) and is 2
3 capable of producing the required product 3
4 size in the range of 13 to 140 mm (0.5 to 4
5 5.5 in.). The power required to reduce the 5
6 maximum feed size to the required product 6
7 size is 352.4 kW, according to Bond’s equa- 7
8 tion (Pothina, 2006). A gyratory crusher’s 8
9 idle power is generally one-third of its rated 9
10 power (Vergne, 2005). Thus, this system 10
11 requires an additional usage of 100 kW. 11
12 This will increases hourly power usage of 12
13 the crusher to approximately 450 kW. The 13
14 compressive strength and abrasive index 14
15 of dolomite are 207 MPa (30,000 psi) and 15
16 620, respectively. 16
17 17
18 Results and analysis 18
19 Figure 3 shows the data related to the feed 19
20 size distributions. It can be observed that 20
21 increasing burden and spacing values will 21
22 increase the coarseness of the material. For 22
23 example, at B and S values of 1.5 and 1.85 23
24 m (5 and 6 ft), respectively, 100% of the 24
25 feed material is finer than 0.2 m (8 in.), 25
26 Figure 3 — Feed size distributions at different burden (B) and spacing (S) val- whereas only 30% of the feed material is 26
27 ues. finer than 0.2 m for B and S values of 6.5 27
28 and 8.00 m (21 and 26 ft), respectively. 28
29 Varying B and S in combination with 29
30 the CSS produces a range of product size 30
31 distributions, as shown in the Fig. 4. A 31
32 diagonal line (ideal slope) represents the 32
33 ideal product size distribution. The ideal 33
34 slope of the product size distribution is 34
35 the ratio between the x‑intercept (size of 35
36 particle) and the y‑intercept (cumulative 36
37 percentage). A value of one indicates 37
38 higher uniformity of the required product 38
39 size and minimum fines in the product. 39
40 Product size distributions close to the ideal 40
41 slope are preferred. Size distributions to 41
42 the right side of the ideal slope are coarser, 42
43 while those to the left side of the ideal 43
44 slope are finer. 44
45 Based on Levene’s test, the probability 45
46 plot of residuals, an F-test, and fitted plots; 46
47 the choice of linear regression to find the 47
48 relation between a and Asrmt was justified. 48
49 The p-value from Levene’s test (0.558) 49
50 conducted on the residual values resulting 50
51 from the linear regression supports the 51
52 initial assumption that the variance of the 52
53 residuals is constant. The F-test indicated 53
54 that cubic and quadratic regressions were 54
55 not statistically significant in representing 55
56 the relation between a and Asrmt. Due to 56
57 this, linear regression was used to estab- 57
58 lish a relation between a and Asrmt. The 58
59 obtained regression equation is 59
60 60
61 Asrmt = -22.24 + 0.3348 a (24) 61
62 From the Minitab output, the intercept 62
63 (-22.24) and the coefficient of a (0.3348) 63
64 Figure 4 — Product size distributions for different B, S and CSS. lies in the intervals (-30.87, -13.95) and 64
65 (0.3018, 0.3678), respectively, at a 95% 65
MINERALS & METALLURGICAL PROCESSING 177 Vol. 24, No. 3 • August 2007
PROOF COPY
1 and CSS values, 4.00 to 4.25, 4.92 to 5.23 1
2 and 0.135 to 0.140 m (13.1 to 13.9, 16.1 2
3 to 17.2 and 0.44 to 0.46 ft), respectively, 3
4 result in minimum total costs in the range 4
5 of $1,934,933 to $2,001,867. Values of 5
6 B, S and CSS at 4.25, 5.23 and 0.140 m 6
7 (13.9, 17.2 and 0.46 ft), respectively, will 7
8 result in minimum annual total costs of 8
9 $1,934,933. 9
10 Currently, the subject mine operates 10
11 with B, S and CSS values of 3.75, 4.61 11
12 and 0.125 m (12.3, 15.1 and 0.41 ft), 12
13 respectively, with total costs per year of 13
14 $2,100,612. The optimized values en- 14
15 able annual savings in total costs from 15
16 $98,745 to $165,679, i.e., 4.7% to 7.9%, 16
17 respectively. 17
18 To test the developed model, the values 18
19 obtained from the crusher manufacturer 19
20 are compared with the values obtained 20
21 from this study. Figure 12 shows the 21
22 product size distribution at B, S and CSS 22
23 values that produces minimum crush- 23
24 ing and minimum total annual costs. It 24
25 Figure 7 — Feed and product in-range (%) at different B, S and CSS values. also can be seen that the manufacturer’s 25
26 curve representing suggested percentage 26
27 passing values is similar to the values 27
28 generated by the model developed for 28
29 this study. 29
30 30
31 Summary and conclusions 31
32 The output results indicate that increasing 32
33 burden and spacing values will increase 33
34 the coarseness and size of the feed mate- 34
35 rial. This in turn overloads the crusher 35
36 at some stage and produces over-size 36
37 material. An increase in the closed-side 37
38 setting will cause a decrease in power 38
39 consumption. The burden, spacing and 39
40 closed-side setting values that result 40
41 in minimum annual crushing and total 41
42 costs within the constraints, i.e., feed size 42
43 distribution required (90% to 98%) and 43
44 product size distribution (85% to 90%), 44
45 were determined. 45
46 The range of values that results in mini- 46
47 mum crushing energy cost are as follows: 47
48 burden at 3.25 to 3.50 m (10.7 to 11.5 ft), 48
49 spacing at 4.00 to 4.31 m (13.1 to 14.1 ft) 49
50 and closed-side setting at 0.110 to 0.120 50
51 m (0.36 to 0.39 ft). The range of values 51
52 that results in minimum total energy costs 52
53 are as follows: burden at 4.00 to 4.25 m 53
54 (13.1 to 13.9 ft), spacing at 4.92 to 5.23 54
55 m (16.1 to 17.2 ft) and closed-side setting 55
56 at 0.130 to 0.140 m (0.43 to 0.46 ft). The 56
57 suggested ranges can save annually from 57
58 $98,745 to $165,679, i.e., 4.7 % to 7.9 %, 58
59 respectively, in total costs. The results of 59
60 this study, specifically the mathematical 60
61 and software model developed in MS 61
62 Excel, can be used by industry profes- 62
63 sionals to evaluate different blasting and 63
64 Figure 8 — Energy consumption to crush a metric ton of material at different crushing scenarios and to predict energy 64
65 B and CSS values. consumption in this process. 65
MINERALS & METALLURGICAL PROCESSING 179 Vol. 24, No. 3 • August 2007
PROOF COPY
1 Hukki R.T., 1961, “Proposal for solomonic settlement 1
between the theories of Von Rittinger, Rick and
2 Bond,” Trans SME/AIME, pp. 403-408.
2
3 Hogg, R., 2003, “Particle characterization,” Principles 3
4 of Mineral Processing, M.C. Fuerstenau and K.N. 4
Han, eds., SME, Littleton, Colorado, pp. 9-20.
5 Inoue, T., and Okaya, K., 1994, “Grinding mechanism
5
6 in centrifugal mills,” 8th European Symposium on 6
7 Comminution, Stockholm, pp. 431-440. 7
8 Kawatra, S.K, Eisele, T.C., and Wlaqui, H.J., 2001, 8
“Optimization of comminution circuit throughput
9 and product size distribution by simulation and 9
10 control,” <http://www.osti.gov/bridge/servlets/ 10
11 purl/837186-Yhbgdr/native/837186.pdf> 11
Kelsall, D.F., Reid, K.J., and Stewart, P.S.B., 1969,
12 “The study of grinding processes by dynamic
12
13 modeling,” Electronic Engineering Trans. Inst., 13
14 EE5(1), pp. 84-95. 14
15 King, R.P., 1979, “A model for the quantitative estima- 15
tion of mineral liberation by grinding,” Institute
16 Journal for Mining, pp. 207-220. 16
17 Lindquist, M., and Sotkovszki, P., 2003, “Work 17
18 hardening in cone crusher liners,” <http://web. 18
sagmilling.nsf/Articles/691CC7491F79F1FA872
19 56DA8006214DC>.
19
20 Lynch, A.J., 1977, Mineral Crushing and Grinding Cir- 20
21 cuits: Their Simulation, Optimization and Control, 21
Vol. 1, Elsevier, Amsterdam.
22 Mishra, B.K., and Rajamani, R.K., 1992, “Analysis of
22
23 media motion in a ball mill,” Comminution: Theory 23
24 and Practice,” K. Kawatra, ed., SME, Littleton, 24
25 Colorado, pp. 427-440. 25
Morrell, S., Napier-Munn, T.J., and Andersen J., 1992,
26 “The Prediction of power draw for comminution 26
27 machines,” Comminution: Theory and Practice, 27
28 K. Kawatra, ed., SME, Littleton, Colorado, pp. 28
405-426.
29 Mular, A.L., Halbe D.N., and Barrett D.J., 2002, Mineral
29
30 Processing Plant Design, Practice, and Control, 30
31 Proceedings, Vol.1, SME, Littleton, pp. 584-600. 31
32 Napier-Munn, T.J., Morrell, S., Morrison, R.D., and 32
Kojovic, T., 1996, Mineral Comminution Circuits:
33 Their Operation and Optimization, Vol. 45, JKMRC, 33
34 Brisbane. 34
35 Narayanan, S.S., and Whiten, W.J., 1988, “Determina- 35
tion of comminution characteristics from single
36 particle breakage tests and its application to ball
36
37 mill scale-up,” Trans of Inst of Min. and Metall., 37
38 245, C115-124. 38
Pothina, R., 2006, “A Gyratory Crusher Model for
39 Energy Savings,” M.S. Thesis, The Pennsylvania
39
40 Figure 11 — Interaction of important parameters for different B, S and CSS 40
State University, University Park, Pennsylvania.
41 values. Radziszewski, P., and Tarasiewicz, S., 1989, “Autog- 41
42 enous mill design using comminution energet- 42
ics,” Proc. Conference on SAG ’89, Vancouver,
43 pp. 773-782. 43
44 Rosario, P.P., Hall, R.A., and Maijer, D.M., 2004a, 44
45 “Improved gyratory crushing operation by the 45
assessment of liner wear and mantle profile
46 redesign,” Minerals Engineering, Vol. 17, No.
46
47 11-12, pp. 1083-1092. 47
48 Rosario, P.P., Hall, R.A., and Maijer, D.M., 2004b, “Liner 48
49 wear and performance investigation of primary 49
gyratory crushers,” Minerals Engineering, Vol.
50 17, No. 11-12, pp. 1241–1254. 50
51 Rosario, P.P., and Hall, R.A., 2006, “Integration of 51
52 gyratory crusher liner wear and operational per- 52
formance for better crusher management,” CIM
53 Bulletin, Vol. 99 (1095), pp. 1-8.
53
54 Svedensten P., and Evertsson V., 2001, Computer 54
55 assisted optimisation of crushing plants for both 55
machine parameter and wear tolerances, <http://
56 www.mvs.ch alme rs.se/~psvede>
56
57 ThyssenKrupp, 2005, Gyratory Crushers, <http://www. 57
58 eprocessingplants.com/eng/html/kreiselbrecher. 58
59 en.shtml>, pp.1-11. 59
Vergne, J.D.L, 2005, Hard rock miner’s handbook
60 rules of thumb, <http://www.mcintoshengineer- 60
61 ing.com/Hard%20Rock%20Handbook/hardrock. 61
62 htm> 62
U.S. Department of Energy, 2000, “Mineral Process-
63 ing Technology Road Map,” U.S. Department
63
64 Figure 12 — Product size distributions compared to manufacturer’s suggested of Energy, <http://www.eere.energy.gov/indus- 64
65 distribution. try/mining/>. 65