Beruflich Dokumente
Kultur Dokumente
a r t i c l e i n f o a b s t r a c t
Article history: In particle comminution processes, the size of the original particles (mother particles) will continuously
Received 12 June 2011 decrease to finer sizes (daughter particles). This size reduction process can be mathematically
Received in revised form represented (in simplified form) by two major functions: selection function (breakage probability)
23 November 2011
and breakage function. In the present study, a new horizontal impact experimental system is
Accepted 12 December 2011
developed, constructed and tested. The experiments from the impact system are used for development
Available online 20 December 2011
and validation of the selection and breakage models. Ability of a number of statistical formulations to
Keywords: accurately describe the probability of the particles to break under an impact load and the probability of
Selection function the particle fragments size are examined and compared. Selected models are analyzed and general
Breakage function
expressions including the effect of the material and particle size are developed. The findings will
Particles comminution
significantly improve the accuracy of comminution processes simulation, and optimization of industrial
Breakage probability
Impact breakage processes.
Size reduction models & 2011 Elsevier Ltd. All rights reserved.
0009-2509/$ - see front matter & 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.ces.2011.12.012
Y. Rozenblat et al. / Chemical Engineering Science 71 (2012) 56–66 57
all of which are based on the same principle, namely the particle collisions and influencing the flow field. The particles are accel-
is accelerated by a gas flow in closed geometry (usually a tube) or erated by compressed air inside a 590 mm long stainless steel SS-
by means of gravity and collides with a rigid target at the end of 303 tube with an inside diameter of 15.8 mm. The acceleration
the acceleration. While other type of impact experimental system tube is designed such that it is possible to modify the distance
uses target to hit the particle (Petukhov and Kalman, 2004; between the outlet and the target. The impact cage in which the
Peukert and Vogel, 2001). By repeating the impact experiment particles collide with the rigid target provides a close environ-
for a sufficient number of particles, one can determine both the ment for the particles to ricochet. The target plate is able to spin
amount of particles that breaks and the fragment size distribution about its axis so that it is possible to investigate the effect of the
by sieving the impacted particles. impact angle (not shown in this paper). The unit where the target
As a result of such impact experiments different researchers is mounted has transparent covers (fiberglass), which makes it
have developed a variety of correlations to define the BP and the possible to measure the particle velocity using a high speed video
BF, such as will be presented later in Tables 2 and 6. It should be camera. Finally, the granular material that has been blown to the
noted that some researchers correlate BP with the impact velocity target is collected in the hopper. At the bottom of the hopper
(Duo et al., 1996; Kalman et al., 2004; Petukhov and Kalman, there is a demountable box that allows us to examine the results
2003; Boerefijn et al., 1998; Lecoq et al., 1999; Salman et al., of the impact. Additionally, there is a dust filter, which is
2002), while others with the kinetic impact energy (Austin, 2002; connected to the top of the collecting tank. This filter is actually
Cho and Austin, 2003; Peukert, 2004; Guigon et al., 1994). a fiber bag, which collects the fines.
To develop accurate correlations between the BP and the BF a This experimental system has the following advantages:
new improved horizontal impact air gun system was designed,
constructed and operated. The experiments carried out with this (1) it allows the feed rate of the particles to be changed;
system are used to develop the new BP. The BF was chosen on the (2) a wide range of particle sizes can be measured;
basis of the experimental data with appropriate considerations. (3) the impact particle velocity can be controlled;
The BP and BF models were compared to existing models from the (4) it offers the ability to work with hazardous materials;
literature and their dependence on different parameters was (5) experiments with different impact angles can be performed;
investigated. (6) large amounts of particles within a short time can be handled,
while keeping the experimental error within allowable
boundaries;
(7) different materials can be used for the impact target;
2. Experimental
(8) the apparatus can be adjusted such that erosion and wear
tests can be carried out and
Our developed horizontal impact system (see Fig. 2) consists of
(9) it is easy to use and maintain.
four major parts: a feed tank, an acceleration tube, an impact cage
and a collection hopper. The tested particles are inserted into the
feed tank and moved towards the acceleration tube by means of a Despite all these advantages, there are still some drawbacks.
vibrating feeder, which enables to control the flow rate of the For example, the angular velocity of the particles cannot be
particles. The particles feed rate was set in such a way that there obtained, and secondary impacts with the walls of the collecting
were only a few particles (2–3) every time in acceleration tube. cage can occur.
This condition is required in order to avoid particle–particle The impact velocities of the accelerated particles were mea-
sured by tracking them with a high speed digital camera (Cam-
Breakage Probability Breakage Function Record 1000, 2000 fps). The following sequence of the frames
(Fig. 3) demonstrates the particle collision and breakage. The
pictures were cropped from the original clip that had been used
for the particle velocity measurements.
Numerical Simulation In order to get statistically representative results, every impact
Strength Function
CFD DEM
Validation test (specific material, size and velocity) was carried out with a
Experiments
sample of no less than 50 g of particles. This minimum quantity
Fatigue Function
was determined by comparing the particle size distribution and
mass loss during the handling of the collided particles for each
Equivalence Function impact experiment. Each sample is weighed and sieved before
each experiment in order to define initial size distribution of
Fig. 1. Scheme of the general approach to the particle breakage phenomenon. the particles, while the size analysis of the particles was
Target Plate
Particle
t=4 msec t=5.5 msec t=6 msec
Fig. 3. Sequence of the particle impact (salt particle, 2.36–3.35 mm, V ¼ 10.3 m/s).
Table 3
BP models.
1
P ¼ 1 ð1Þ Petukhov and Kalman (2004) Logistic
1 þ ðV =aÞb
V b Cheong et al. (2004), Salman et al. (2002),
P ¼ 1exp ð2Þ Weibull
a Kapur et al. (1997), Djamarani and Clark (1997)
1 lnðV Þa
P¼ 1 þ erf pffiffiffi ð3Þ Pocock et al. (1998) Lognormal
2 b 2
P ¼ aV b ð4Þ
Duo et al. (1996) Range limitation
3 !
V
P ¼ 1arctg tgð1Þa ð5Þ Boerefijn et al. (1998) Range limitation
30
rV 2 lH
P¼a 2 ð6Þ Cleaver and Ghadiri (1993) Multitude of variables
K 2c f
100 100
90 90
80 80
70 70
Breakage Probability, S
Breakage Probability, S
60 Salt 60
50 3.35-4 mm 50 Potash
2.36-3.35 mm
40 2-2.36 mm 40 3.35-4 mm
1.4-2 mm 2.36-3.35 mm
30 1-1.4 mm 30 2-2.36 mm
0.71-1 mm 1.4-2 mm
20 20
Logistic model Logistic model
10 Weibull model 10 Weibull model
Lognormal model Lognormal model
0 0
0 5 10 15 20 25 30 35 40 45 50 55 5 10 15 20 25 30 35 40 45 50
Impact Velocity, V, m/sec Impact Velocity, V, m/sec
100 100
90 90
80 80
70 70
Breakage Probability, S
Breakage Probability, S
60 60
50 50
GNP
40 4-4.75 mm 40
3.35-4 mm Zeolite
30 30
2.36-3.35 mm
1.4-2.36 mm
20 2-2.36 mm 20 Logistic model
Logistic model
Weibull model
10 Weibull model 10
Lognormal model
Lognormal model
0 0
4 8 12 16 20 24 28 32 21 24 27 30 33 36 39 42 45 48 51 54
Impact Velocity, V, m/sec Impact Velocity, V, m/sec
where d0 is the mean size of the particles in a feed size interval, comparison of the median impact velocity of different materials
and AV and BV are the correlation parameters. Fig. 7 demonstrates per same particle size revealed the sensitivity of the materials to
that as the particle size becomes larger, the median impact impact loads. For example, salt is the weakest material and potash
velocity (impact velocity at which 50% of the feed particles break) is the strongest one (among the tested materials) in terms of their
becomes smaller for all the tested materials. Moreover, a ability to withstand impact loads.
Y. Rozenblat et al. / Chemical Engineering Science 71 (2012) 56–66 61
100 100
90 90
80 80
70 70
Breakage Probability, S
Breakage Probability, S
60 60
GNP
50 50
4-4.75 mm GNP
40 3.35-4 mm 40
4-4.75 mm
2.36-3.35 mm
30 30 3.35-4 mm
2-2.36 mm
2.36-3.35 mm
Logistic model
20 20 2-2.36 mm
in terms of impact velocity
Logistic model
10 10
in terms of specific energy
0 0
6 8 10 12 14 16 18 20 22 24 26 28 30 32 0 50 100 150 200 250 300 350 400 450 500
Impact Velocity, V, m/sec Specific Energy, E , J/kg
Fig. 6. Impact breakage probability in terms of (a) impact velocity and (b) specific energy for GNP particles in various sizes.
Breakage Probability
Salt 100
30 Salt: 3.35-4mm
Potash Salt: 2.36-3.35mm
GNP 90 Salt: 2-2.36mm +20%
Salt: 1.4-2mm
Median Impact Velocity, V , m/sec
25 V =A *exp(-d /B )
Calculated Breakage Probability, S
Salt: 1-1.4mm
80 Salt: 0.71-1mm
Potash: 3.35-4mm
20
70 Potash: 2.36-3.35mm
Potash: 2-2.36mm
Potash: 1.4-2mm
60 GNP: 4-4.75mm
GNP: 3.35-4mm
15
50 GNP: 2.36-3.35mm
GNP: 2-2.36mm
Zeolite: 1.4-2.36mm
40
10
30
5 20
10
0
0.0 1.0x10 2.0x10 3.0x10 4.0x10 5.0x10 0
Original Particle Diameter, d , m 0 10 20 30 40 50 60 70 80 90 100
Experimental Breakage Probability, S
Fig. 7. Median parameter correlation.
Fig. 8. Comparison of the predicted probabilities to the experimented ones.
Table 4
Logistic model empirical parameters.
100
Material V 50 ¼ AV eðd0 =BV Þ
90
AV BV DI
80
m/s m – Salt: 3.35-4mm
Breakage Probability, S
70 Salt: 2.36-3.35mm
3 Salt: 2-2.36mm
Salt 36.5 2.1 10 2.3
60 Salt: 1.4-2mm
Potash 45.1 3.3 10 3 3.4 Salt: 1-1.4mm
GNP 34.5 3.8 10 3 4.2 50 Salt: 0.71-1mm
Zeolite 40.7 4.4 10 3 5.9 Potash: 3.35-4mm
40 Potash: 2.36-3.35mm
Potash: 2-2.36mm
Potash: 1.4-2mm
30
3.1.3. Explicit selection function GNP: 4-4.75mm
GNP: 3.35-4mm
When we combine Eqs. (7) and (8), the selection function takes 20 GNP: 2.36-3.35mm
the following form: GNP: 2-2.36mm
10 Zeolite: 1.4-2.36mm
1 Logistic model (S =1-1/(1+V )
SI ¼ 1 ð9Þ
1 þ ðV=AV eðd0 =BV Þ ÞDI 0
0 5 10 15 20 25 30 35
In order to obtain the best correlation of the proposed model Normalized Impact Velocity, V =(V/V )
(Eq. (9)), the deviation between the model and the experimental
data was minimized by the Matlab program. The most suitable Fig. 9. Validation of the Logistic Function model.
62 Y. Rozenblat et al. / Chemical Engineering Science 71 (2012) 56–66
Table 5 1/10 of the parent particle size (class j), dj is the representative
List of the BF models. size in class j and Di is the sieve size of fragments in class i. The
parent particle size (dj) is defined by the geometric mean value of
Model Reference
the parent sieve size interval, i.e. dj ¼ (Dj Dj þ 1)0.5. a stands for the
a b modulus of the upper-truncated Rosin–Rammler distribution.
d
Bi,j ¼ 1 1 ð10Þ Harris (1968)
d0 Since the size of the parent particles in the current study is
" #
d
b indicated by d0 and defined as an arithmetic mean value of the
Bi,j ¼ 1exp Cheong et al. (2004)
a ð11Þ initial sieve size of the particle, i.e., d0 ¼(Dj þDj þ 1)/2, and the fact
b
d
that BP depends on the impact velocity, instead of the breakage
Bi,j ¼ ð12Þ Schumann (1940) energy, it was decided to redefine the parameters of the original
a
ð9=ðd0 =Di Þ1Þa
Bi,j ¼ BðDi ; dj Þ ¼ 1ð1t 10j Þ ð13Þ Tavares equation. The form of the equation has remained the
Tavares (2004)
a b same as the original Tavares model (Eq. (13)), but t10j statistic
d d
Bi,j ¼ f þ ð1fÞ ð14Þ Austin and Luckie (1972) parameter was renamed to t10 and it was redefined as an
d0 d0
a 0
empirical parameter in addition to a. Additionally, Di sieve size
d 1 dd
Bi,j ¼
d0 2
1 þ tan h 0 ð15Þ Vogel and Peukert (2002) in class i was redefined as representative size of the fragments in
d
size class i and a new indication of it is di.
Introduction of the renamed parameters into original Tavares
model is shown by
a
parameters were evaluated based on least squares method. Bi,j ¼ Bðdi ; d0 Þ ¼ 1ð1t 10 Þð9=ðd0 =Di Þ1Þ ð16Þ
A detailed explanation of that process can be found in Breakage function is the fraction of the particles that trans-
Rozenblat et al. (2011). The most suitable parameters for all the ferred from j size class into a smaller size class i as a result of the
materials experimented upon are presented in Table 4. impact event and Bij is the conventional sign for it. Since in the
Figs. 8 and 9 confirm the validity of the proposed model (Eq. current work the symbols were changed and there are no longer
(9)). Fig. 8 compares the experimental results with the predicted indications of size class j, the breakage function sign has also to be
ones by means of the breakage probability correlation, and Fig. 9 changed into Bi,0, but for the sake of conventionality the sign of
presents the deviation of the experimental data from the model. the breakage function was unchanged, i.e. Bi,j.
The figures indicate that most of the results are within the The new form of BF is now dependent on the impact velocity
boundary of 720% error, which is reasonable. instead of the breakage energy, and finally, the representative size
of the parent particle size is the arithmetical mean value of the
3.2. Breakage function parent size interval.
Among all the materials that were tested in the present
The breakage function (BF) indicates the mass of the broken research, salt offers the widest range of sizes. Consequently,
particles that reaches a size interval less than the feed size further analysis is based on the results of salt and then the type
interval as a consequence of breakage. In other words, the BF of correlations will be applied to the rest of the materials
determines the particle fragment size distribution in terms of experimental data.
mass. Table 5 lists some common definitions of the BF.
Since the last part of Eq. (15) refers to fine dust (Vogel and 3.2.2. Parameter t10
Peukert, 2002) and the first part is identical to Eq. (12), these BF Parameter t10 was obtained from fits of the BF (Eq. (16)) to the
definitions can be neglected. Therefore, only five of BF expres- experimental data. In order to find the t10 correlation, this
sions (Eqs. (10)–(14)) will be analyzed. In order to compare the parameter was plotted versus two major variables, i.e. impact
models and select one of them to represent fragments size velocity and original particle size. Fig. 11a and b presents the
distribution, all of these models were fitted to all our experi- influence of the impact velocity and original particle size,
mental results. As an example, fitting of each model to the respectively.
experimental data of salt at an initial size interval of 3.35–4 mm Fig. 11a demonstrates that as the impact velocity becomes
is presented in Fig. 10. greater, the quantity of the small-sized fragments (i.e., dust)
Fig. 10 shows that from a physical point of view the greater the becomes larger. Fig. 11a and b also demonstrates that t10 depends
impact velocity, the smaller the size of the fragments and the on both the impact velocity and the original particle size. Since
larger the amount of the fragments; this observation is also parameter t10 has an upper bound of unity according to the
supported by Antonyuk et al. (2006). Fig. 10 also demonstrates Tavares definition (Eq. (13)), the t10 dependence on initial particle
that there are some expressions that describe experimental data size and impact velocity was expressed in the following form:
insufficiently. The choice of the suitable model that will represent
daughter size probability was made on the basis of the coeffi- t 10 ¼ 1expðaV b d0 Þ ð17Þ
cients of determination (R2). Table 6 presents the sum of the
where a and b are the model parameters.
fittings in terms of average R2. Table 6 demonstrates that the
The correlation of t10 was evaluated for experimental data of
highest average R2 for all of the materials is obtained using the
salt, since it has the greatest size variety.
Tavares model (Eq. (13)). Therefore, this model was selected to
In order to examine this relationship, the resulting t10 was
represent the particle fragment size distribution.
plotted versus the calculated one and presented in a comparison
chart (see Fig. 12).
3.2.1. Tavares BF definition The agreement of this correlation (Eq. (17)) with the experi-
The BF model described by Tavares (2004) (Eq. (13)) is based mental results for the rest of the materials was examined and
on the upper-truncated Rosin–Ramler distribution (Tavares, found to be satisfactory. Fig. 12 demonstrates t10 correlation
2004). This model depends on several variables including impact quality, while it was normalized by dividing to its maximal value.
energy, particle characteristics and stiffness of both the particles As one can see from Fig. 12, most of the results are within the
and the impact target. While t10j indicates the fraction that passes error boundaries of 720%.
Y. Rozenblat et al. / Chemical Engineering Science 71 (2012) 56–66 63
100 100
4.4 m/s 4.4 m/s
90 Salt 90 11 m/s Salt
11 m/s
d =3.35-4 mm d =3.35-4 mm
14 m/s 14 m/s
80 80 16 m/s
16 m/s
18.1 m/s 18.1 m/s
70 70
23.9 m/s 23.9 m/s
Undersize, B
Undersize, B
60 Harris model [31] 60 Cheong et al. model [10]
50 50
40 40
30 30
20 20
10 10
0 0
0 500 1000 1500 2000 2500 3000 3500 0 500 1000 1500 2000 2500 3000 3500
Fragment Particle Diameter, d, μm Fragment Particle Diameter, d, μm
100 100
4.4 m/s 4.4 m/s
90 Salt 90 11 m/s Salt
11 m/s
d =3.35-4 mm 14 m/s d =3.35-4 mm
14 m/s
80 80 16 m/s
16 m/s
18.1 m/s 18.1 m/s
70 70
23.9 m/s 23.9 m/s
60 Schumann model [32] 60 Tavares model [33]
Undersize, B
Undersize, B
50 50
40 40
30 30
20 20
10 10
0 0
0 500 1000 1500 2000 2500 3000 3500 0 500 1000 1500 2000 2500 3000 3500
Fragment Particle Diameter, d, μm Fragment Particle Diameter, d, μm
100
4.4 m/s
90 11 m/s Salt
14 m/s d =3.35-4 mm
80 16 m/s
18.1 m/s
70
23.9 m/s
60 Austin & Luckie model [34]
Undersize, B
50
40
30
20
10
0
0 500 1000 1500 2000 2500 3000 3500
Fragment Particle Diameter, d, μm
calculated one with normalization by the maximal value and analysis examination of the proposed BF model (Eq. (19)) the
presented in Fig. 14 for all materials tested. experimental results were compared to the predicted ones. The
Fig. 14 shows that a correlation model (Eq. (18)) is satisfying. results are presented in Fig. 15.
Allowed boundaries for this model are 720%. Fig. 15 demonstrate that most of the results are within the
boundary of 720%, which is reasonable considering the compre-
3.2.4. Estimation of the most suitable BF model parameters hensiveness of the model.
Since primary parameters (t10 and a) of the BF model were
separately evolved, it was necessary to examine the overall BF
model parameters, which includes the combination of t10 and a 4. Conclusions
expressions (Eq. (19)). The estimation process of the secondary
parameters t10 (Eq. (17)) and a (Eq. (18)) was conducted such that The main idea of the current study was to select suitable
the calculated fragment size distribution will represent the models to describe particle population behavior under impact
experimental measured in the most accurate way. loads. For this purpose, a newly developed experimental impact
All of the BF parameters were estimated in the same manner system was constructed and tested. The result functions are
as the BP parameters. Evolved parameters significantly improved suitable for the impact event-driven or steady-state processes.
the ability of the BF model to reflect the experimental data of the These functions are part of the model array, which will represent
fragment size distribution. Final values of the BF model para- the complete mathematical definition of the comminution phe-
meters for all the materials experimented upon are presented in nomenon. Using this overall model, one can optimize any indus-
Table 7. trial process requiring particle size reduction.
The key findings are presented below:
3.2.5. Explicit breakage function
In order to obtain the explicit BF model, the primary para-
A new horizontal air gun experimental system was developed.
This system allows easy and convenient way of testing
meters t10 (Eq. (17)) and a (Eq. (18)) were substituted into the
new breakage function expression (Eq. (16)). The result of these
1.0
mathematical rearrangements is presented as follows: Salt
a 0.9 Potash
Bi,j ¼ 1ð1t 10 Þð9=ðd0 =Di Þ1Þ
Calculated Normalized Model Parameter, t
GNP +20%
c 0.8
where t 10 ¼ 1expðaV b d0 Þ and a ¼ aa V ba d0a ð19Þ Zeolite
0.7
Eq. (19) basically has the same form as the original Taveras
equation, only a dependence of t10 and a on the impact velocity 0.6
and mother particle size is suggested in this work. For the final
0.5
Table 6 0.4
Average R2 for all models.
0.3
Material Salt Potash GNP Zeolite Average
0.2
Reference 2
R R
2
0.1
Salt Salt
0.35 0.35
~15 m/sec
~20 m/sec
0.30 3.35-4mm 0.30 ~25 m/sec
2.36-3.35mm ~35 m/sec
2-2.36mm ~40 m/sec
0.25 0.25
1.4-2mm ~44 m/sec
Model Parameter, t
Model Parameter, t
0.15 0.15
0.10 0.10
0.05 0.05
0.00 0.00
0 5 10 15 20 25 30 35 40 45 50 55 5.0x10 1.0x10 1.5x10 2.0x10 2.5x10 3.0x10 3.5x10 4.0x10
Impact Velocity, V, m/sec Original Particle Diameter, d , m
Fig. 11. Dependence of t10 parameter on (a) impact velocity and (b) original particle size.
Y. Rozenblat et al. / Chemical Engineering Science 71 (2012) 56–66 65
Salt Salt
2.0
1.65 ~15 m/sec
3.35-4mm ~20 m/sec
1.8 2.36-3.35mm
1.50 ~25 m/sec
2-2.36mm ~35 m/sec
1.6 1.4-2mm ~40 m/sec
1-1.4mm 1.35 ~44 m/sec
Model Parameter, α
Model Parameter, α
0.71-1mm
1.4 Power model 1.20
1.2 1.05
1.0 0.90
0.8 0.75
0.6 0.60
0 5 10 15 20 25 30 35 40 45 50 55 5.0x10 1.0x10 1.5x10 2.0x10 2.5x10 3.0x10 3.5x10 4.0x10
Impact Velocity, V, m/sec Original Particle Diameter, d , m
Fig. 13. Dependence of a parameter on (a) impact velocity and (b) original particle size.
Potash 90
0.8 GNP
80
0.7
70
Calculated Undersize, B
0.6
60
0.5
50
0.4
40
0.3
30
0.2
20
0.1
10
0.0
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 0
0 10 20 30 40 50 60 70 80 90 100
Experimental Normalized Model Parameter , α
Experimental Undersize, B
Fig. 14. a parameter validation.
Fig. 15. Comparison of the predicted fragment size probabilities to the
experimented ones.
DI wideness of logistic distribution function (dimension- Cleaver, J.A.S., Ghadiri, M., Rolfe, N., 1993. Impact attrition of sodium monohydrate
less) crystals. Powder Technology 76, 15–22.
Cho, H., Austin, L.G., 2003. An equation for the breakage of particles under impact.
Di , Dj sieve size of fragments in size interval i or j (m) Powder Technology 132, 161–166.
d0 original particle size (m) Cleaver, J.A.S., Ghadiri, M., 1993. Impact attrition of sodium monohydrate crystals.
H hardness (Pa) Powder Technology 76, 15–22.
Duo, W., Boerefijn, R., Ghadiri, M., 1996. Impact attrition of fluid cracking catalyst.
i,j size intervals (dimensionless) In: Proceedings of the 5th International Conference on Multiphase Flow in
Kc fracture toughness (N/m2) Industrial Plants, Amalfi, Italy, pp. 170–179.
l cube dimensions (m) Djamarani, K.M., Clark, I.M., 1997. Charcterization of particle size based on fine
and coarse fractions. Powder Technology 93, 101–108.
P breakage probability (dimensionless) Fahernwald, A.W., Newton, J., Herkernhoff, E., 1938. Velocity of hit in rock
R2 coefficient of determination (dimensionless) crushing. Engineering and Mining Journal 139, 43–46.
SI particle breakage probability or selection function Guigon, P., Thomas, A., Dodds, J., 1994. Experimental study of a jet of particles on a
plate. IFPRI Progress Report.
(dimensionless)
Harris, C.C., 1968. The application of size distribution equations to multi-event
t sequential time (ms) comminution process. Transactions of the American Institute of Mining,
V impact velocity (m/s) Metallurgical and Petroleum Engineers 241, 343–358.
V50 median of impact velocity (m/s) Kalman, H., Rodnianski, V., Haim, M., 2009. A new method to implement
comminution functions into DEM simulation of a size reduction system due
r particle density (kg/m3) to particle-wall collisions. Granular Matter 11, 253–266.
Kalman, H., Hubert, M., Grant, E., Petukhov, Y., Haim, M., 2004. Fatigue behavior of
impact comminution and attrition units. Powder Technology 146, 1–9.
Subscripts Kapur, P.C., Pande, D., Fuerstenau, D.W., 1997. Analysis of single-particle breakage
by impact grinding. International Journal of Mineral Processing 49, 223–236.
calc/exp calculated or experimental value Lecoq, O., Guigon, P., Pons, M.N., 1999. A grindability test to study the influence of
material processing on impact behavior. Powder Technology 105, 21–29.
norm normalized value Piret, E.L., 1953. Fundamental aspects of grinding. Chemical Engineering Progress
49 (2), 56–63.
Petukhov, Y., Kalman, H., 2004. Empirical breakage ratio of particles due to impact.
Powder Technology 143–144, 160–169.
Peukert, W., Vogel, L., 2001. Comminution of polymers: an example of product
Acknowledgment
engineering. Chemical Engineering & Technology 24/9, 945–950.
Petukhov, Y., Kalman, H., 2003. A new apparatus for particle impcat tests. Particle
This research was supported by the German-Israeli Foundation & Particle Systems Characterization 20, 267–275.
Peukert, W., 2004. Material properties in fine grinding. International Journal of
for Scientific Research and Development (GIF). Mineral Processing 74, s3–s17.
Pocock, J., Veasey, T.J., Tavares, L.M., King, R.P., 1998. The effect and quenching on
References grinding characteristics of quartzite. Powder Technology 95, 137–142.
Rozenblat, Y., Portnikov, D., Levy, A., Kalman, H., Aman, S., Tomas, J., 2011. Strength
distribution of particles under compression. Powder Technology 208,
Andrews, E.W., Kim, K.S., 1999. Threshold conditions for dynamic fragmentation of 215–224.
glass particles. Mechanics of Materials 31, 689–703. Schubert, W., Khanal, M., Tomas, J., 2005. Impact crushing of particle particle
Austin, L.G., 2002. A treatment of impact breakage of particles. Powder Technology compounds—experiment and simulation. International Journal of Mineral
126, 85–90. Processing 75 (1–2), 41–52.
Antonyuk, S., Khanal, M., Tomas, J., Heinrich, S., Morl, L., 2006. Impact breakage of Salman, A.D., Biggs, C.A., Fu, J., Angyal, I., Szabo, M., Hounslow, M.J., 2002. An
spherical granules: experimental study and DEM simulation. Chemical Engi- experimental investigation of particle fragmentation using single particle
neering and Processing 45/10, 838–856. impact studies. Powder Technology 128, 36–46.
Austin, L.G., Luckie, P.T., 1972. Estimation of non-normilized breakage distribution Schumann Jr., R., 1940. Principles of comminution, I. Size distribution and surface
parameters from batch grinding. Powder Technology 5, 267–277. calculations. AIME Technical Paper 1189, 1–11.
Brosh, T., Kalman, H., Levy, A. Impementation of comminution functions in DEM Tavares, L.M., 2004. Optimum routes for particle breakage by impact. Powder
simulation of dilute-phase pneumatic conveying. Grannular Matter, in press. Technology 142, 81–91.
Brosh, T., Kalman, H., Levy, A. DEM simulation of particle breakage in jet-mill, in Vogel, L., Peukert, W., 2002. Characterization of grinding-relevant particle proper-
preparation. ties by inverting a population balance model. Particle & Particles systems
Boerefijn, R., Ning, Z., Ghadiri, M., 1998. Disintegration of weak lactose agglom- Characterization 19, 149–157.
erates for inhalation applications. International Journal of Pharmaceutics 172, Weedon, D.M., Wilson, F., 2000. Modelling iron ore degradation using a twin
199–209. pendulum breakage device. International Journal of Mineral Processing 5,
Cheong, Y.S., Reynolds, G.K., Salman, A.D., Hounslow, M.J., 2004. Modelling 195–213.
fragment size distribution using two-parameter Weibull equation. Interna- Yashim, S., Saito, F., Horita, H., 1981. Single particle testing by double pendulum
tional Journal of Mineral Processing 74 S, s227–s237. type impact testing apparatus. Kagaku Kogaku Ronbunshu 7/1, 83–89.