Professional Documents
Culture Documents
net/publication/235997816
CITATIONS READS
0 523
3 authors, including:
Some of the authors of this publication are also working on these related projects:
All content following this page was uploaded by Rahul K Soni on 15 May 2014.
a
Indian School of Mines, Dhanbad-826004 (India)
b
Institute of Minerals and Materials Technology (IMMT), Bhubaneswar-751013 (India)
Abstract
Single particle breakage tests have been used by a number of researchers to investigate
some salient features of the complex comminution process. A twin pendulum setup is designed
to indirectly measure the losses in energy for single particle breakage which is assumed to be
equivalent to comminution energy. Calculated comminution energies are analyzed for their
relations against breakage characteristics therefore impact energies, specimen size and operating
work index. Breakage pattern of material also analyzed with the determination of single
distribution parameter (t10). Comminution energies and breakage behavior both analyzed against
the transfer patterns of energy; transfer of part of impact energy to useful energy in comminution
and its dependency on breakage characteristics such as material hardness and size of material are
presented and discussed in this article. Above mentioned terms are modeled to bring conclusions
in a mathematical platform.
Keywords: comminution energy, twin pendulum, size distribution parameter, size distribution.
*
Correspoding Author: Eswaraiah, C., Institute of Minerals and Materials Technology (IMMT)
Bhubaneswar-751 013 (India)
Email: eswar@immt.res.in
1. Introduction
models using first order kinetics which require the determination of breakage rate functions and
residence time distribution. Crushing and grinding are known energy intensive processes
(Schonert, 1979 [1]). In a typical mineral processing plant approximately 50 percent of total
energy consumption takes place by comminution only (Cohen, 1983 [2]). Single particle
breakage tests can provide near to accurate information on realistic breakage characteristics of
materials to develop comprehensive mathematical models and scale-up designs for comminution
process.
processes (Schuhmann, 1940 [3]; Harris, 1966 [4]; Schonert, 1972 [5]; Rumpf, 1973 [6]; Buss et
al., 1978 [7]; Beke, 1981 [8]; Tarjan, 1981 [9]). The elementary comminution can be regarded as
a single particle subjected to stress. During such a breakage micro-event, two main modes of
comminution have been identified (Gaudin, 1926 [10]; Committee on Comminution and Energy
Consumption, 1981 [11]). In the major mode, the particle is subjected to compressive stresses. In
the minor mode, the rubbed or jostled so that stress concentration exist at some surface sites on
the corners or protrusions, leading to abrasion. Relatively little research has been reported on the
abrasive mode of comminution (Crabtree et al. 1964 [12]; Kinasevisch, 1964 [13]; Rose, 1983
[14]) probably because it appears to be slower and less efficient means of grinding. In addition, a
vast amount of research has been directed towards the study of breakage phenomena to
understand most difficult elementary comminution events (Bickle, 1958 [15]; Calcott, 1969 [16];
Rumpf, 1973 [6]; Lowrison, 1974 [17]; Lynch, 1977 [18]; Committee on Comminution and
and utilization of impact forces lead to the development of pendulum devices. Bond, 1946 [20]
developed a twin pendulum apparatus to determine the impact crushing strengths of 76×50mm
rock particles under controlled crushing conditions. An attempt was made by Gaudin et al. (1946
[21]) to approximately quantify the rebound losses by using a special twin pendulum device.
Research on various designs of twin pendulum setup has been performed by several researchers
(Narayanan, 1986 [22]). Investigations reported can be broadly classified into (Narayanan, 1986
[22]):
(b) detailed energy balance to accurately determine the net energy utilized in comminution
2. Experimental
twin pendulum device is designed and fabricated similar to Narayanan et al., (1988 [23]). In
addition to referred setup a computer monitoring system added to setup for accurate
laser detection arrangement and a signal processor having compatibility with LabView-Data
presented in Fig.1. A ball pendulum weighing 4.685kg and a cylindrical pendulum weighing
6.853kg used as the input pendulum and rebound pendulum respectively. Diameters of both
pendulums are 10.4cm and are made of manganese-steel alloy. Length of the rebound pendulum
is 10.2cm. Each pendulum suspended from respective steel plates attached to main frame.
Connection of pendulums with steel plate made through adjustable sleeve and cotton rope of
1.5mm diameter. The pendulum length can be adjusted with a turn buckle on each of the two
ropes. Also, the pendulums alignment can be made through provided arrangement with the main
frame.
This arrangement enabled the pendulum to be suspended in its most stable position so
that friction during its motion is minimized. Furthermore, the potentially uneven motions of the
rebound pendulum after a collision are rapidly corrected by the two ropes that tend to hold the
pendulum in the equilibrium position avoiding any twisting motion. Input pendulum can be
raised to a known height by the use of an attached nylon release thread. A 3-fin (with identical
spacing) arrangement is attached to the rebound pendulum as shown in figure 1 and figure 2.
The motion of the rebound pendulum via 3-fins is monitored through interruption of laser beam
and processed through data acquisition electronic card. Data acquired from setup is further
processed with LabView which resulted in oscillation time periods. Various setup dimensions
Ore sample of BHQ, coal and magnetite with Bond’s operating work index as 17.5, 13 and
12.7 respectively were collected from different locations followed by their primary
comminution to obtain the closed size range samples of (-18+15)mm, (-9.5+8.0)mm and (-
al., 1988 [23]) was performed to obtain the parameter of linear relationship between rebound
pendulum time period and square of its angular displacement. For each sample, particles in the
desired number were taken according to their respective density such that the approximate
weight of sample becomes 50-60 gm. Then one by one particle from each sample was loosely
attached to the active face of rebound pendulum and input pendulum with different heights was
impacted on the particle. In the presented work 10 oscillation after each impact was monitored,
see Narayanan, et al., (1988 [23]) for complete experimental procedure. Each material and size
of it was experimented for three heights of input pendulum i.e. 5cm, 10cm and 15 cm, therefore,
three levels of impact energy. So, three materials, three sizes and three levels of impact energy
Mathematics for twin pendulum experiments can be derived from fundamental physics.
Detailed derivations (Narayanan, et al., 1988 [23]) shows conservation of linear momentum in
experimental setup as
where M is the mass of pendulum, U is the initial velocity before collision, V is the final
velocity after collision and subscripts 1 and 2 indicate input and rebound pendulums
M 1 + M 2 V2
e=( ) −1 (2)
M1 U1
Coefficient of restitution easily leads to the calculation of final velocity of input pendulum
after collision. Energy of each pendulum before and after collision can be calculated with
measured velocities. Difference of energy for pendulums before and after collision can be
EC = Ei − E r 1 − E r 2 (3)
where subscripts c, i, r1 and r2 are for comminution energy, input energy, residual energy of
position is computed from oscillation time period, T, and calibration constants, a and b, use of
the relationship
T −a
A = 16 * (4)
b
Narayanan, 1988 model (Single parameter, t10 model) and Klimpel and Austin (1984) model are
Narayanan and Whiten (1988) have widely used a novel procedure for estimation of
breakage distribution functions of ores from the so called t-family of curves. In this method, the
product size distribution can be represented by a family of curves using marker points on the size
distribution defined as the percentage passing at a fraction of the parent particle size. They have
proposed empirical equations for relating the reference curve data t10with the impact energy.
Narayanan et al., (1983) [24], Pauw et al., 1988 [25], Bourgeois, 1993 [26] and King, 2002
[27] proposed Eq. (5) to represent the breakage distribution
9 α
[{ } ]
n −1
t n = 1 − (1 − t10 ) (5)
In the Eq. (5) given above, n in the equation is the ratio of geometric mean specimen (feed)
and product sizes and tn is the cumulative fraction passing to the corresponding n values. So t10
is the cumulative fraction passing by the geometric mean product size 1/10th to the geometric
mean specimen size. For example in case of specimen size of (-9.5+8.0) mm, t10 is the
(9.5 * 8.0)
cumulative fraction passing by the size of = 0.8718 mm.
10
2.3.2. Klimpel and Austin (1984)
xi γ x
Bi, j = ϕ ( ) + (1 − ϕ )( i ) β (6)
xj xj
Model consists of more number of parameters. In the equation ܤ, is the cumulative amount
breaking from the jth size to below of ith size. ݔ and ݔ are the sieve sizes at ith and jth level
respectively, where j value is greater than i value. ߮, ߛ and ߚ in the equation are model
parameters.
A wide research on equation (6) concluded its failure for breakage representation in
several ways. Equation is having more number of parameters of parameters as compared to
equation (5), and after performing regression and therefore determining the parameters it found
difficult to correlate those with the breakage characteristics i.e. impact energy, particle size and
material hardness. Parameters dependency on these characteristics found highly unpredictable.
Only parameter φ showed weakly that it is specimen size dependent and independent of ore and
specific input energy.
Calibration of the experimental setup is a critical step in for research. Equation (4) can be re-
written as
A
T = a + b( ) 2
4
Above linear equation needs calibration of rebound pendulum to obtain the values of a and b.
For this rebound pendulum was raised to heights of 5, 10, 15, 20, 25 cm and then allowed for
oscillations. 10 swings of oscillation in each case monitored and resulted in time period for each
case. A linear regression as per equation (4) resulted in values of a and b as 1.466 and 0.0007.
not attempted to examine breakage distribution against Klimpel model. However, physical
experiments described in subsection 2.2 were followed by number of steps for analysis.
Breakage product were first segregated in different size ranges and then examined for their
differential weights, which were fitted into equation (5) in the MATLAB curve fitting
toolbox. It is important to note that tn values are on the fractional basis and a percent passing
value can be obtained by multiplying it with a factor of 100.Table 1 gives the limiting
and 15cm. These energies were converted into corresponding kg-cm (Ei) values followed by
energy values in Kwh/ton. Time periods supplied by LabView, through equation (4) computed
the angular displacement, followed by the computation of the vertical displacement of rebound
pendulum resulted in the energy and velocity of rebound pendulum. Let, U1 and U2 are the
velocities of input and rebound pendulum before collision and after collision they obtain the
velocities V1 and V2 respectively. So, the coefficient of restitution was calculated from equation
(2). Which led to the tabulated values of U1, V1 and V2, and therefore the energies of input and
rebound pendulums before and after collision. Thus equation (3) can be used to calculate the
values of comminution energy, Ec, divided by mean particle weights to obtain specific
comminution energies, E cs. It is important to note that losses in energy are considered in the
research as comminution energy, E c. Table 2 below summarizes the mean particle weights of
2
E r1 = M 1 *V1 (9)
2g
3.3.1. Breakage distribution of the materials was analyzed first for their
Single parameter t10 model, is concluded to be a better option for representing breakage of
material. Model consist of two parameter that are α and t10. Dependency of α and t10 to breakage
characteristics such as impact energy (EI) or comminution energy (ECS), specimen particle size
and material hardness as Bond’s work index (WIo) are shown in article on individual basis.
Figure 4 and 5 below shows the variation in parameter ‘t10’ with comminution energy under
size (-9.5+8.0)mm
Fig. 5: Relationship between the size distribution parameter ‘t10’ and comminution energy for all
sizes of quartz
It is self explainable that how t10 is increasing with the increase in comminution energy. As
and reflected in terms of t10. While it is difficult to generalize the conclusions from figure 5, as
lower specimen sizes of a material shows more breakage, which can be taken as more breakage
for smaller particles that does not states true with the findings of earlier research. Reasons to
explain the phenomena are given in the article. (1) t10 is the measurement of size fraction passing
1/10th of the specimen size, and so it is somehow measurement of fine generation and here when
size is decreasing it leads to increase in fines generation and that is measured in terms of t10. It is
notable to see the comminution of coal marked by some of the authors, Sahoo, 2006 [28] have
suggested use of t50 instead of t10 for breakage analysis of coal in their handling systems. The
former breakage product percent which is 5 times smaller than of later is a suitable parameter for
coal as it is a brittle and easily breakable material, and so needs comparatively smaller size to be
considered for a analysis of fines generation. In addition, Weedon et al., 2000 [29] also
suggested consideration of t50 for the degradation of iron ores by twin pendulum device. (2)
Earlier researches were mostly on drop weight tests and their breakage distribution
representation with t10, in which some of the researches might have found increase in t10 with
increase in specimen size. It can be notable to consider that losses in drop weight tests are nearly
constant with respect to input energy, when it is compared to the case of twin pendulum tests, as
in twin pendulum test a significant part of energy is going to the oscillations of rebound and
impact pendulum after collision. Figure 6 shows the transfer patterns of energy for comminution
for magnetite. Three levels of energy input therefore, 5cm, 10cm and 15cm height of input
pendulum, are marked in figure as EI in Kwh respectively. In figure comparison for three sizes of
specimen is shown, which concludes better energy transfer in lower size particles, and therefore
Fig. 6: Transfer pattern of energy and its comparison for three specimen sizes of
magnetite
Fig. 8: Variation of absolute specific energy with input energy for three sizes of quartz
Fig. 9: Variation in specific comminution energy with input energy for (-6.3+5.6) mm size
It important to note that in figure 6 and 7 the maximum values for ECS/EIS does not exceed
the value of 0.6, which means that at least 40% of the input energy is lost in oscillations and
unwanted vibrations of input and rebound pendulum after collision. Also, in figure 4 and 9 coal
has shown unpredictable behavior, as it is having a Bond’s work index value of 13.0 that falls
between the values, 13.5 for magnetite and 12.7 for quartz. However, in figures the
corresponding values for coal do not lie in between of magnetite and quartz, this is due non-
mineral characteristics of coal. Coal is highly brittle, easily breakage material and generates
more fine, and so in figure 4 it is having high values for t10. Also, coal is a lighter material in
comparison, therefore shows increased values of ECS in figure 9, as ECS is derived by dividing
As mentioned earlier, Narayanan, 1988 model (Single parameter, t10 model) requires to
determine two parameters, therefore t10 and α. Analysis for t10 for its dependency on breakage
characteristics is already mentioned in the article. One needs to determine α also to work with
this model.
Fig. 10: Dependency of parameter ‘α’ on energy input for three ores
Fig. 11: Dependency of parameter ‘α’ on specimen size of ores for three levels of input
energy
Though, Sand et al., 2004 [32] reported that α decreases with the increase in specimen size
while the present research concludes a contradictory result. Also, Sand et al., 2004 [32] drew an
COR is found with increase in input energy. However, the rate of increase in COR shows
significant decrease, when it moves from 46.85 to 70.27 Kwh input energy as compared to its
move from 23.42 to 46.85 Kwh input energy. In disagreement to above Mishra, 2003 [30]
suggested decrease in COR with increase in input energy, as a result of comparatively high
energy dissipation at higher impact velocity. Researchers have found vague conclusions in this
regard; it mostly depends upon the range of impact velocity. However, result found in
presented work for COR shows high agreement with Narayanan, 1985 [31].
Abbreviations
References
[1] Schonert, K., 1979. Energy aspects of size reduction of brittle materials, Zement-kalk-
Gips (Translation No. 1/79), No. 3: 40-44.
[2] Cohen, H.E., 1983. Energy usage in mineral processing, Trans. I.M.M., 92: C160-C164.
[3] Schuhmann, R., 1940. Principles of Comminution, A.I.M.I. Technical publication No.
1187.
[4] Harris, C.C., 1966. On the role of energy in comminution: A review of physical and
mathematical principles, Trans. I.M.M., 75: C37-C56.
[5] Schonert, K., 1972. Role of fracture physics in understanding Comminution phenomena,
Trans. S.M.E./A.I.M.E., 252: 21-26.
[6] Rumpf, H., 1973. Physical aspects of Comminution and new formulation of a low of
comminution, Powder Technology, 7: 145-159.
[7] Buss, B., Shubert, H., 1978. Results of investigation into comminution micro-processes
(German), Neve Bergbautechnik, (12): 706-710.
[8] Beke, B., 1981. The process of fine grinding, 15-65 (Dr. W. Junk, Publishers:
Boston/London).
[9] Tarjan, G., 1981. Mineral processing, vol. 1: 35-185 (Akademiai Kiado: Budapest).
[10] Gaudin, A.M., 1926. An investigation of crusher phenomena, Trans. S.M.E./A.I.M.E.,
73: 253-310.
[11] Committee on Comminution and Energy Consumption, 1981. Comminution and Energy
Consumption, National Materials Advisory Board Report No. NMAB-364, 50-160
(National Academy Press: Washington, D.C.).
[12] Crabtree, D.D., Kinasevisch, R.S., Mular, A.L., Meloy, T.P., Fuerstenau, D.W., 1964.
Mechanics of size reduction in comminution systems. Part I. Impact, abrasion and
chipping grinding, Trans S.M.E./A.I.M.E, 229: 201-206.
[13] Kinasevisch, R.S., Crabtree, D.D., Mular, A.L., Meloy, T.P., Fuerstenau, D.W., 1964.
Mechanics of size reduction in comminution systems. Part II. Interpreting size distribution
curves and the comminution event hypothesis, Trans. S.M.E./A.I.M.E, 229: 207-210.
[14] Rose, H.E., 1983. Material removal by surface abrasion in masses of particles, Trans.
I.M.M., 92: C138-C146.
[15] Bickel, W.H., 1958. Crushing and Grinding-A bibliography, 20-75 (Dept. of Scientific
and Industrial Research, HMSO, London).
[16] Calcott, T.G., 1969. Recent developments in size reduction theory in Symposium on
Size-reduction, 33-38 (University of Sydney).
[17] Lowrison, G.C., 1974. Crushing and Grinding-the size reduction of solid materials, 35-
120 (Butterworths, London).
[18] Lynch, A.J., 1977. Mineral crushing and grinding circuits, their simulation, optimization,
design and control, 8-45 (Elsevier Scientific Publishing Co. Amsterdam).
[19] Austin, L.G., Luckie, P.T., Klimpel, R.R., 1984. Process Engineering of Size Reductions:
Ball Milling. 181-229 (S.M.E./A.I.M.E.: New York).
[20] Bond, F.C., 1946. Crushing tests by pressure and impact, Mining Technology, Technical
preprint No. 1895, 169: 58-65.
[21] Gaudin, A.M., Hukki, R.T., 1946. Principles of comminution–size and surface
distribution, Trans. S.M.E.A.I.M.E., 169: 67-94.
[22] Narayanan, S.S., 1986. Single particle breakage tests: a review of principles and
applications of comminution modeling, Bull. Proc. Australas. Inst. Min. Metall., 291(4):
49-58.
[23] Narayanan, S.S., Whiten, W.J., 1988. Determination of comminution characteristics from
single-particle breakage tests and its application to ball-mill scale-up, Trans. Inst. Min.
Metall., Sect. C: Mineral Process. Extr. Metall., 97: C115-C124.
[24] Narayanan, S.S., Whiten W.J., 1983. Breakage characteristics of ores for ball mill
modeling, Proceedings of Aus. IMM, 286, June, pp.31-39.
[25] Pauw, O.G., Mare, M.S., 1988. The determination of optimum impact-breakage routes
for an ore, Powder Technology, Vol 54, pp.3-13.
[26] Bourgeois, F., 1993. Micro-scale Modelling of Comminution Processes. Thesis,
University of Utah.
[27] King, R.P., 2002. Modeling and Simulation of Mineral Processing Systems, first ed.
Butterworth-Heinemann Publishers, NY.
[28] Sahoo, R., 2006. Review: An investigation of single particle breakage tests for coal
handling system of the gladstone port, central queensland university, rochampton,
queensland, 4702, Australia, powder technology, vol. 161, issue 2, pp. 158-167.
[29] Weedon, D.M., Wilson, F., 2000. Modeling iron ore degradation using a twin pendulum
breakage device, Int. J. of mineral processing, vol. 59, issue 3, pp. 195-213.
[30] Mishra, B.K., 2003. A review of computer simulation of tumbling mills by the discrete
element method: Part I—contact mechanics, Int. J. Miner. Process., 7, 73-93.
[31] Narayanan, S.S., 1985. Development of a laboratory single particle breakage technique
and its application to ball mill modeling and scale-up, PhD thesis, University of
Queensland, Australia.
[32] Sand, G.W., Subasinghe, G.K.N., 2004. A novel approach to evaluating breakage
parameters and modeling batch grinding, Minerals Engineering, Extractive Metallurgy
Program, Department of Mineral Science and Extractive Metallurgy, Murdoch University.