Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
net/publication/248475701
CITATIONS READS
9 271
2 authors:
Some of the authors of this publication are also working on these related projects:
All content following this page was uploaded by Luis Marcelo Tavares on 09 December 2017.
http://www.elsevier.com/copyright
Author's personal copy
Abstract
The traditional Bond work index proposed by F.C. Bond over 60 years ago is still today a useful tool for characterizing material
crushability and grindability in the minerals industry. Although not as popular as the Bond ball mill or the Bond rod mill work indices,
the impact work index, also called Bond crushability index, continues to be much used for crusher selection. The present work shows the
application of a model that combines measurements of particle fracture energies and a parameter that characterizes material amenability
to breakage by repeated impacts in the prediction of the Bond crushability index. Results show that a reasonably good agreement exists
between measured and predicted results, as long as losses in the pendulum apparatus, used to determine the Bond crushability index, are
taken into account in the simulations.
2007 Elsevier Ltd. All rights reserved.
0892-6875/$ - see front matter 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.mineng.2007.08.021
Author's personal copy
Nomenclature
A cross-sectional area of the impact load cell rod g acceleration due to gravity (m s2)
(m2) k stiffness of the particle prior to impact (GPa)
C wave propagation velocity in the ILC rod ~k stiffness of the damaged particle (GPa)
(m s1) mb weight of the striker used in the ILC (kg)
CB impact energy per unit particle thickness, also mp particle weight (kg)
called crushing resistance (J mm1) n number of impacts (–)
D damage variable, which corresponds to the frac- P 80% passing size in the crusher product (lm)
tion of any cross-section of the particle that is P(E) cumulative fracture energy distribution of parti-
not able to withstand load (–) cles in a sample or proportion of particle broken
d particle size (m) (–)
Dmax maximum value of the damage variable reached t time (s)
during loading the particle (–) tc time at primary fracture (s)
Dn amount of damage sustained by the particle in vo impact velocity (m s1)
the nth impact (–) W specific energy consumption (kWh t1)
E specific particle fracture energy (J kg1) w particle thickness – distance between loading
En specific particle fracture energy of a particle points (mm)
after the nth impact (J kg1) Wi Bond work index (kWh t1)
Ek specific input energy – kinetic energy of the drop Y modulus of elasticity of the particle (N m2)
weight per unit of particle weight (J kg1)
Ek,n specific input energy used in the nth impact on Greek letters
the particle on the Bond pendulum (J kg1) a local deformation of the system comprised by
E0k;n specific input energy used in the nth impact on particle and platens (m)
the particle for simulation of the Bond work in- ac local deformation of the system at primary frac-
dex, which takes into account losses in the pen- ture (m)
dulum (J kg1) c damage accumulation constant (–)
E50 median of the log-normal distribution of specific / release angle in the Bond pendulum (–)
particle fracture energies (J kg1) l Poisson’s ratio of the particle (–)
F 80% passing size in the crusher feed (lm) q rod specific gravity (kg m3)
F(t) load on the particle as a function of time (N) qs particle specific gravity (g cm3)
Counterweight
Hammers
ho
Collection
box
crusher power consumption from Bond’s law of comminu- impacting individual particles with a free-falling drop
tion, given by (Bond, 1952) weight and recording the force-time profile. From the
force-time profile different measures are calculated. The
1 1
W ¼ 10Wi pffiffiffi pffiffiffiffi ð3Þ specific particle fracture energy E is calculated by
P F " Z
tc Z tc
1
where F and P are the 80% passing sizes in the feed and E¼ vo F ðtÞdt þ g F ðtÞtdt
product of the crusher (in lm) and W is the specific energy mp 0 0
Z tc 2 Z tc #
consumption (in kWh/t). 1 1 2
A detailed description of the experimental apparatus F ðtÞdt F ðtÞdt ð4Þ
2mb 0 qAC 0
and the analysis of the data can be found elsewhere (Berg-
strom, 1985; Tavares, 2007). where tc is the time of primary fracture of the particle, mp is
the particle weight, vo is the impact velocity, g is the accel-
2.3. Impact load cell eration due to gravity, q is the density, A is the cross-sec-
tional area and C is the wave propagation velocity in the
The impact load cell (Fig. 2) consists of a long rod onto rod.
which solid-state strain gauges are attached (Tavares and Experiments were conducted using the impact load cell
King, 1998; Tavares, 2007). The experiment consists of measuring 63 mm of diameter on particles of size fraction
Drop weight
release system
Data acquisition
board
Collection box
Laser
source
Photodiode
Strain gauges
Computer
Signal
conditioner
63.0–53.0 mm. Impact velocities used in testing the differ- particle fracture energy) after each impact at a known
ent materials ranged from 3.3 to 5.5 m/s, which are, impact energy Ek is given by (Tavares and King, 2002)
although higher, in the order of magnitude of those used
En ¼ En1 ð1 Dn Þ ð9Þ
in the Bond pendulum (which range from 0.5 to about
3.5 m/s). Details of the equipment and apparatus can be and
found elsewhere (Tavares, 2007). 2c5
The distribution of particle fracture energies can be 2c Ek;n
Dn ¼ ð10Þ
described well with the log-normal distribution, given by ð2c 5Dn þ 5Þ En1
1 ln E ln E50 which must be solved iteratively, given that Dn appears
P ðEÞ ¼ 1 þ erf pffiffiffi ð5Þ
2 2r E implicitly in Eq. (10).
A minimum of two experiments is required to determine
where E50 and rE are the median and the standard devia-
c, the only parameter in the model:
tion of the distribution, respectively.
Additional experiments were conducted to determine
• a measurement of the distribution of particle fracture
the amenability of the various materials to fracture by
energies from a test at a loading energy that is suffi-
repeated impacts (Tavares and King, 2002). The response
ciently high for primary breakage of all particles in the
of materials to repeated impacts is likely to play a role in
sample (parameters in Eq. (5));
the sequential loading events that particles suffer while they
• impacts at a constant low-energy level to determine the
move through the crushing chamber of jaw, gyratory and
cumulative fraction of broken particles. In these experi-
cone crushers. This, indeed, becomes evident from the pro-
ments the impact energy is given by Ek;m ¼ mb v2o =ð2mp Þ,
cedure used to determine Bond’s crushability index
where mb is the ball weight, mp is the particle weight and
described previously. Thus, particles sieved in the range
vo is the velocity of the ball at the instant of collision.
from 22.4 to 19.6 mm have been subjected to repeated
low-energy impacts in order to estimate the parameter c
Results of experiments used to determine the parameter
of the damage model (Tavares and King, 2002). In order
c are shown in Fig. 3, where the fraction of broken particles
to do that, particle fracture energy distributions were deter-
is presented as a function of number of impacts, along with
mined for this size range as well.
model predictions.
In the case of Bond’s standard pendulum test, the spe-
3. Modeling and simulation procedure
cific impact energy varies according to
The load–deformation response that results from impact 117
E0k;n ¼ ½1 cosð10 þ 5 ðn 1ÞÞ ð11Þ
of a spherical particle can be described by a combination of mp
continuum damage mechanics (Kachanov, 1958) and Hertz
contact theory (Tavares and King, 2005), where n = 1 for the first impact (/ = 10), and so on, and
E0k;n is given in J/kg when mp is measured in kg. Eq. (11) as-
d 1=2 ~ 3=2 sumes free-fall conditions in the pendulum and that no
F ¼ ka ð6Þ
3
where F is the load, a is the deformation and 100
~k ¼ kð1 DÞ ð7Þ
/kg
J
2
is the stiffness of the particle, with k = Y/(1 l ), where Y
.5
80
46
g
J/k
The damage variable D may be described by the power .5
81
law relationship 60
c
a
D¼ ð8Þ
ac 40
95
90
4. Results and discussion
70
50
Typical experimental results from Bond crushability
30 tests are presented in Fig. 4, which shows the significant
10
variability of the measurements, which is primarily asso-
5 ciated with the variability found in the properties of indi-
Copper ore 1
Bauxite 1
vidual particles contained in each sample. A summary of
Limestone 4 the experimental results for all materials tested is pre-
1 sented in Table 1, which shows the different characteris-
1 10 100
tics of the materials tested, with values of (median)
Crushing Work Index - Wi (kWh/t)
impact Wi that varied from about 4–20 kWh/t. Fig. 5
Fig. 4. Distribution of impact (crushing) work index for selected materials shows typical experimental results from particle fracture
measured using the Bond pendulum (symbols: experimental data; lines: energy measurements which contain significant variability
log-normal fit).
as well.
Eq. (1) proposed by Fred Bond to calculate crushing
resistance in the crushability test assumes that no losses
momentum losses occur, an assumption that is later ana-
occur during impact. Simulation of the crushability test
lyzed in the paper.
using data measured using the impact load cell, where no
The simulation is carried out until all particles have bro-
losses during fall of the drop weight occur, requires that
ken. Ek,n (Eq. (11)) is then transformed in the crushing
any loss in momentum that might take place during the
resistance per unit thickness CB, given by
crushability test be taken into account.
117 Tests have been conducted in the Bond pendulum and
CB ¼ ½1 cosð10 þ 5 ðn50 1ÞÞ ð12Þ
w the impact load cell by impacting a 20-mm diameter
Table 1
Summary of results from Bond crushability and impact load cell tests
Rock types Sample Specific gravity Bond crushablity testa Impact load cell
qs (g/cm3)
Particle weight, mp Thickness d CB (J/mm) Wi E50 r2E b c
(kg) (mm) (kWh/t) (J/kg)b
Acid rocks 1 2.64 0.3784 51.0 0.605 12.25 59.0 1.605 3.06
2 2.79 0.3966 44.3 0.793 15.20 144.1 0.677 2.76
3 2.80 0.3430 43.0 0.361 6.90 54.8 0.614 4.34
4 2.69 0.1664 30.1 0.473 9.41 68.5 0.805 2.78
Copper 1 2.93 0.4371 48.3 1.081 19.73 194.3 0.635 5.25
ores 2 2.92 0.4839 45.5 0.736 13.48 149.7 0.509 5.46
3 3.39 0.5538 48.5 0.781 12.32 133.5 0.532 3.08
4 3.25 0.4812 50.5 0.770 12.67 177.3 0.491 2.04
5 3.25 0.2312 36.5 0.288 4.74 47.3 0.984 2.95
Iron ores 1 3.81 0.4759 34.2 0.512 7.19 25.1 0.759 3.06
2 3.84 0.3358 44.4 0.291 4.05 24.5 0.845 3.73
Limestones 1 2.71 0.2609 38.0 0.407 8.03 121.1 0.402 4.58
2 2.88 0.4047 49.0 0.434 8.07 80.3 1.314 4.83
3 2.63 0.3315 47.8 0.353 7.18 35.6 0.670 6.38
4 2.73 0.3884 45.0 0.253 4.95 15.7 0.506 5.41
5 2.73 0.4423 45.5 0.761 14.90 67.2 0.919 4.28
Bauxites 1 2.41 0.3135 50.5 0.470 10.44 67.6 0.591 5.76
2 2.42 0.2206 49.5 0.542 11.98 96.4 0.715 7.61
3 2.49 0.1751 47.0 0.227 4.88 68.7 0.462 2.98
4 2.41 0.1894 46.0 0.341 7.56 52.2 0.795 7.33
5 2.43 0.1885 46.7 0.445 9.80 66.1 0.766 6.40
6 2.45 0.1845 46.0 0.274 5.98 71.2 0.395 4.95
Coals 1 1.68 0.1688 26.0 0.221 7.04 25.7 0.585 1.49
2 1.55 0.2171 34.8 0.103 3.55 16.7 0.636 4.16
3 1.78 0.2043 38.6 0.191 5.72 18.7 0.707 4.68
a
Median values reported for the various measures.
b
Results for particle sizes in range 65–53 mm.
Author's personal copy
99
a 100
Copper ore 1
Bauxite 1
Cumulative distribution (%)
Limestone 4
95
P(E0)
90 50
n = 1 (φ = 10o)
70
50
30 0
10 100 600
E k, 1
10
5
b 100
P (E1)
1 10 100 1000
50
Specific particle fracture energy - E (J/kg)
n = 2 (φ = 15 o)
Fig. 5. Distribution of particle fracture energies of selected materials
tested in the impact load cell in size 63.0–53.0 mm (symbols: experimental
0
data; lines: log-normal fit). 10 100 600
Ek,2
c 100
15
P (E3)
50
Diameter of impression (mm)
n = 4 (φ = 25o)
10
0
10 100 600
Ek,4
5
d 100
P (E5)
0
1 10 100
Impact energy Ek (J)
0
10 100 600
Fig. 6. Comparison of impression size in lead from impact of a 20-mm Ek,6
diameter tungsten carbide ball using the impact load cell and the Specific particle fracture energy (J/kg)
pendulum device. Error bars show the standard deviations of the
measurements. Fig. 7. Simulation of Bond pendulum test using the continuum damage
model of particle fracture and data from acid rock 3. Solid lines show how
the distributions of particle fracture energies change with each impact and
tungsten carbide ball positioned between the striker(s) and the dashed lines the impact energies and proportion broken after each
a lead plate. In the case of the Bond pendulum, one ham- impact. Hatched areas represent the fraction broken in previous impacts.
mer was kept at rest while the other was moved to different
angles. The mean diameter of the impression was then pre- 104:5
cisely measured and results are related to the energy Ek;n ¼ ½1 cosð10 þ 5 ðn 1ÞÞ ð13Þ
mp n0:13
applied in the test (Fig. 6). The figure shows that a loss
of momentum occurs in the pendulum test, which depends
directly on impact energy and, thus, on the release angle of This correction corresponds to a reduction in the calcu-
the pendulum /. An expression that can be used to correct lated energy that varies from about 15% at lower angles up
the impact energy in the Bond pendulum is 0.89/n0.13 to as high as 40% for the largest angles used in the test.
(where n is related to the release angle – in degrees – as A step-by-step description of the simulation procedure is
/ = 10 + 5(n 1)), so that in the simulations of weaken- presented as follows, along with a worked example that
ing by repeated impacts, the impact energy in each loading uses data from acid rock 3 (Table 1). It assumes that all
stage should be calculated from a modification of Eq. (11) material parameters (E50, rE and c) have been determined
given by previously.
Author's personal copy
(Refahi et al., 2007) that was based on material mechanical particle undergo as a result of repeated impacts plays a sig-
properties such as tensile strength, Young’s modulus and nificant role in the value of the Bond crushing work index.
Poisson’s ratio, and the simulation package FLAC3D.
In that investigation it was recognized that the Bond crush- Acknowledgements
ability index is more sensible to the presence of structural
defects, irregular shape of particles and plastic behavior The authors would like to acknowledge the financial
of the rock that was not appropriately accounted for in support from Brazilian Agencies CNPq and FAPERJ.
their simulations. Since these material characteristics cer-
tainly are found to affect particle breakage behavior in References
the impact load cell, then the validity of the approach used
in the present work is demonstrated. Austin, L.G., 2004. A preliminary simulation model for fine grinding in
Rather than aiming to demonstrate that the Bond high speed hammer mills. Powder Technol. 143–144 (25), 240–252.
impact work index should be determined using the present Bergstrom, B.H., 1985. Crushability and grindability. In: Weiss, N.L.
(Ed.), . In: SME Mineral Processing Handbook, vol. 2. Littleton, pp.
procedure instead of Bond’s pendulum, the present work
30-65–30-68.
shows that data measured in the impact load cell are con- Bond, F.C., 1947. Crushing tests by pressure and impact. Trans. AIME
sistent with those obtained using standard technical mea- 169, 58–66.
sures, such as the impact work index. As a result, it Bond, F.C., 1952. The third theory of comminution. Trans. AIME 193,
represents a sound basis for more advanced modeling 484–494.
Kachanov, L.M., 1958. Time of the rupture process under creep
and scale-up procedures of crusher selection for use in
conditions. Izv. Akad. Nauk AN SSSR 8, 26–31 (in Russian).
the future. Refahi, A., Rezai, B., Mohandesi, J.A., 2007. Use of rock mechanical
properties to predict the Bond crushing index. Miner. Eng. 20 (7), 662–
669.
5. Conclusions Tavares, L.M., 2007. Breakage of Single Particles: Quasi-static. In:
Salman, A.D., Ghadiri, M., Hounslow, M.J. (Eds.), . In: Handbook of
Particle Breakage, vol. 12. Elsevier, pp. 3–69.
A simulation procedure based on the combination of the Tavares, L.M., King, R.P., 1998. Single-particle fracture under impact
damage mechanics model of particle fracture and measure- loading. Int. J. Miner. Process. 54, 1–28.
ments of the distribution of particle fracture energies has Tavares, L.M., King, R.P., 2002. Modeling of particle breakage by
been successfully used to predict the Bond crushability repeated impacts using continuum damage mechanics. Powder Tech-
nol. 123, 138–146.
index and leads to results that agree reasonably well with
Tavares, L.M., King, R.P., 2005. Continuum damage modeling of particle
measurement results. The work demonstrates that energy fracture. ZKG Int. 58, 49–58.
losses in the pendulum device must be taken into account Unland, G., Szczelina, P., 2005. Coarse crushing of brittle rocks by
in the simulations and that the progressive weakening that compression. Int. J. Miner. Process. 74S, S209–S217.