Indian Journal of Engineering & Materials Sciences Vol. 10, August 2003, pp. 269-276

Breakage parameters of some minerals and coals ground in a laboratory size ceramic mill

A Ozkan", M Yekelerb* & S Aydogana

"Selcuk University, Department of Mining Engineering, 42031 , Konya, Turkey

bCumhuriyet University, Department of Mining Engineering, 58140, Sivas, Turkey

Received 26 December 2002; accepted I May 2003

The kinetics of batch dry and wet grindings of calcite, barite, quartz, lignite and anthracite from feeds of sieve size -425+300 /Lm has been determined using a laboratory scale ceramic ball mill. The 51 values obtained were the highest (0.294 min· l

) for anthracite and the lowest (0.071 min· l) for quartz when ground dry . However, the wet grinding of these material s

gave higher 51 values by a factor of 1.28 for calcite, 1.15 for barite, 1.25 for quartz, 1.11 for lignite and 1.07 for anthracite comparing to the dry 51 values. The B i,j values also changed for all the materials ground as dry and wet. The simulations of the product size distributions for both first-order and non-first order breakage of the materials ground were in good agreement with the experimental size di stributions. There is a relationship between the 51 values and the y value of B ij

parameter: i.e., 51 values increase when y values decrease, indicating that faster breakage rates of top sizes produce more fines in the finer size distribution region (-ISO /Lm).

The grinding of minerals and other solids is widely used as a processing step in the production of ceramics, glass, cement, pigments and refining of ores. Thus, size reduction by grinding is an important industrial operation involved in many aspects of the mineral, metallurgical , power and chemical industries. Increased rates and efficiencies of milling have been sought through the optimization of milling by providing favourable physical and operational conditions for the mills used I, The design and scale­up of ball mills are important issues in size reduction processes, Therefore, various models are used for predicting the behaviour of large industrial-scale mills using the data obtained in small laboratory-scale mills2

. It is known that the rate of size reduction is enhanced by wet rather than dry grinding, but the degree of enhancement depends on the material being ground3

.4, BondS states that the capacity of wet grinding in the industrial scale is 1.3 times higher than dry grinding under the same operating conditions, Austin et at. I showed that the ratio of the specific rate of breakage values between dry and wet grinding varied from 1.1-2.0 for different materials , The degree of this acceleration depends on the slurry concen­tration and the fineness of grinding6

, The water eliminates or reduces the effect of slowing-down due

* For correspondence (e-mail: yekeler@cumhuriyet.edu.tr)

to coating of the balls and reagglomeration of fines in the mill. In addition, the water allows better transfer of the mechanical action of the tumbling balls to the stressing of the particles, leading to higher rates of breakage, On the other hand, as fines accumulate in the mill, the slurry density becomes more viscous and decrease in the breakage rate is occurred after a particular viscosity value7

,

This paper reports on the kinetics of dry and wet grinding of calcite, barite, quartz, lignite and anthracite, all from Turkey, and compares the results of dry grinding to those of wet grinding for these minerals and coals ground in a laboratory size ceramic ball mill ,

Theory The grinding rate constant, which is widely

accepted as first-order expression of grinding rate, is one of the important parameters required to evaluate a grinding process8

. The normal breakage rate is defined by a first -order breakage rate as gi ven below I:

rate of breakage of size i = SjWj(t)W . ., (1)

where Sj is the constant specific rate of breakage of particles of size i , and Wj(t) is the mass fraction of the total charge (W) that is of size i at time t of grinding. It is common to use sieving both to prepare test feed sizes and measure particle size distributions and size i

270 INDIAN J. ENG. MATER. SCI, AUGUST 2003

refers to a screen interval (usually a .fi interval) indexed by i=] for the largest size, 2 for the next smaller size, down to 11 for the final size interval. On the other hand, at long grinding times, Si may decrease as finely ground material accumulates in the mill causing non-first order breakage, i.e., a slowing­down of breakage rates.

The cumulative primary breakage distribution (Bi ,j) is defined as the mass fraction of material broken from the larger size j that appears less than the upper size of a smaller size interval i on a primary breakage action . An empirical equation for this function I is as follow:

B· . = I.J

{ ly { l~ Xi- l (l <p Xi-l

~ + - ~ , n~i > .i ~ l ... (2)

i=j

where Xi is the top size of the size interval indexed by i. The parameters y, <!> and ~ are characteristic of the material being ground. On plotting experimentally determined Bi.1 values versus Xi on log-log scales, the slope of the straight line lower portion of the curve gives the value of y, <!> is the intercept of this part of the line extrapolated to X2, and ~ is determined to make the function fit the upper part of the curve9

-1I

.

Experimental Procedure The feed materials used in the grinding test were -

425+300 flm (40x50 mesh) sieve fractions of calcite, barite, quartz, lignite and anthracite from different regions of Turkey. These materials were prepared by crushing first in a jaw crusher then in a double roll crusher.

The grinding experi ments were performed in a cylindrical ceramic ball mill of 128 mm internal diameter operated at 70% of critical speed, using ceramic ball s of 25.3 mm diameter (Table 1). The Standard Sand B test conditions were used to give first order grinding kinetics when steel media and mills employed: 20% mill volume filled by the ball bed (assuming a bed porosity of 0.4) and 60% of the interstices of the ball bed filled with powder. Sieving schedules were established from a study of sieving kinetics , and a sieving time of 10 min on a Rotap sieving machine with a sample of 50 g taken from the ground product (by cone and quartering) was found to

be satisfactory. The grinding tests were calTied out dry and wet at I, 2, 4, 8, 16 and 32 min of grinding time.

Size analysis by sieving was performed by washing with tap water for at least 10 min starting from the top size, then each successive screen was washed in this manner, removed and put into an oven to dry . The dried material remaining on each screen was further dry sieved in a Rotap shaker for 10 min. Finally, each remaining fraction on the screen was weighed.

Results and Discussion Fig. 1 shows the firs t-order plots of dry and wet

grindi ng of calcite, barite, quartz, lignite and anthrac ite. It had been previously shown that good first order plots were only obtained for the feed sizes less than about 650 flm for the ceramic media at the given experimental conditions l2

. For the non-first order breakage of larger sizes called abnormal breakage rate, an approximate S value was experimentally obtained to be lower than the S va lues obtained for the feed sizes less than 650 fl m. This is due to larger feed sizes ground in the mill that are not nipped properly by the fixed ceramic ball sizes chosen . In the literature, it is generally noted that the abnormal breakage region for steel media (25.4 mm balls) is started when larger sizes than approximately 1000 flm . In addition , a mill with lower density grinding media will have both a lower capacity (tph) and a lower power draw and specific rate of breakage

Table I- Test conditions

Mill

Balls

Materia l

Inner diameter, mm

Length , mm

Volume, cm'

Critical speed, rpm

Operational speed, rpm

Matcrial

Diameter. mm

Specific gravity

Charge, kg

Fractional ball fill ing

Calcite

Barite

Quartz

Lignite

Anthracite

Powder-ba ll loading ratio

Pulp density, % (by volume)

128 212

2500

132

92

Ceramic

25.3

3.75

1.125 0.2

Specific Powder

gravity weight, g

2.69

4.38 2 .65

1.46

1.42

0.60

40

194

315 191

105

102

OZKAN el a/.: BREAKAGE PARAMETERS OF SOME MINERALS AND COALS

100,0 ..----_____ -,

-+- Dry I rlndiftl ($,-O,19Imin-')

.. -I---+-~w __ • -;:... . .:..,· ... ;.:;IS.:...".;,:;.O.2U:...,: • .:;. •• .;.." .....,...~-l

o 1 • 6 I 10 12 '4 GRINOING TIME, minul~~

(a)

'''''r---------,

-+- DflJri!od inl\S, - O. I02IW." , _ wdpindin,IS,-O,IIJmin· l ,

a 11 16 20 H 11 Jl J6 GRINDING TIME. minulC'1

(d)

_ OoyP;ndi"'(.SJ - O.lHmlft· I ,

-- W_IfindinI I.S. - O,JI6 min·.) ,., +--.,.........r-.'--',.~.....,...~~~ o 1 • , • 10 12 14

GRINDING TIME, minllin

(b)

10(1.0..--------_-,

_ u.,a,Mdinl tS, - O.194m .. -I,

,. '-I--~W...:."';....·""...;;"~ISc.,."'':;''''.:.,:'._in' I.;..' ~~-I o 1 4 ~ • 10 U 14 16 'I

GRINDING TIME. minlile.

(e)

100.0-.:;---------,

-e- Ory ,nrodin,IS ,- 0.011 m;,.· I )

"' -I-_~W-"'."""";:",· ;,:,;· ':,::;ISc.,."'::;,:'''':'':''':::'-' '.;..'1 ~~~ • 11 16 10 H 11 Jl J6

GRI NDING TIME. mMlllk,

(c)

271

Fig. I- First-order plots for dry and wet grinding of -425+300 JLm (a) calcite, (b) bari te, (c) quartz, (d) lignite and (e) anthracite.

Table 2-Specific rates of breakage and other characteristics breakage values of the -425+300 JLm feed size materials

Mineral SI-dry SI .wel SI .wel / SI-dry aT.dry aT-weI (Xdry aWe!

Calc ite 0.198 0.254

Barite 0.275 0.3 16

Quartz 0.071 0.089

Lignite 0.102 0.113

Anthracite 0.294 0.3 15

values, giving the same comparable as one with high density balls l.l3.

The Sj, aT and a values for all materials studied are outlined in Table 2. The values of S can be fitted to be Eq. (3) gave an aT value by inserting a values and xo=650 Ilm for all materials in this work.

.. . (3)

The primary breakage distribution functions were determined by using the BII calculation method for grinding of calcite, barite, quartz, lignite and anthracite as shown in Fig. 2. The good B jj results are obtained when the time of grinding is chosen to give an amount of material broken out of the top size interval of about 20-30% (ref. 1) . Thus, the method is applied to one minute of grinding time data for calcite, barite, lignite and anthracite, and 2 min of grinding time data for quartz in this study. The equation for BlI calculation is as following:

1.28

1.15

1.25

1.11

1.07

0.26 0.32 0.60 0.54

0.36 0.38 0.62 0.45

0.10 0.13 0.80 0.80

0.12 0.13 0.43 0.35

0.40 0.43 0.75 0.75

Table 3--Primary breakage distribution function parameters

Mineral BiJ parameters

Dry grinding Wet grinding

y <P ~ y <P ~

Calcite 1.01 0.56 4.4 0.90 0.60 4.3

Barite 0.82 0.57 4. 1 0.66 0.5 1 4.1

Quartz 1.52 0.65 4.5 1.34 0.53 4.5

Lignite 1.62 0.57 6.4 1.52 0.63 6.2

Anthracite 1.63 0.63 4.7 1.30 0.59 5.6

log [0- P; (0)/(1- P; (t)] . H I = ] , 1>1

I . log [0- P2 (0)/(1- ~ (t)

... (4)

where Pj(O)= cumulative weight fraction at time 0 for size interval i , P/ t) = cumulative weight fraction at time t for interval i. The results of the primary breakage distribution calculations of all materials are given in Table 3.

272

1.00

€ ~ >= u

2 '" 0.>6 '-' "' '" 0. 10

~ iJi '" > >= ~ :> :>i :> V

0 .0 1

1.00

iff ~ 1= u

2 On ~ 1.112 1.)2

w 0" 06) '-' " :.i 01 0

~ '" > 1= ~ :> :>i :> U

0.01

10

INDI AN 1. ENG. MATER. SCI, AUGUST 2003

1.00 1.00

3- . ~

is S 5 5 E E 2 2 ~ 4.1 4.1 '" '-' :.i 0. 10 :.i 0 .10 ;:; <

'" ::i ., ., '" '" > > >= >= ~ ~ :> :> :>i :>i

-4-- O r)' ,rindln, :> _ Dry" iftdlll, :> - Wtt arinditl l V _ WCl lriltdifll U

O.Ot 0.0 1

100 1000 10 100 1000 SIEVE SIZE. pm SIEVE SIZE.. pm

(a) (b)

1.00

.~

s 5

I G Ila 1>'1

2 1 6) I lD 06) O.

'" " '-' < '" 0 10 <

~ W > 1= ~ :> :>i

-+- D". , rindin. :> -+- Dry ,rind,,,. _ Wtll, indi"l U _ Wttlrindilla

0.01

100 1000 10 100 1000 SIEVE SIZE. pm SIEVE SIZE, pm

(d) (e)

10

_ D'l' lrilldinl -+- Wn ,' indi",

100 SIEVE SIZE.)!m

(c)

1000

Fig. 2-Primary cumulati ve breakage distri buti on functions for dry and wet grind ing o f -425+300 /-! m (a) calcite, (b) barite, (c) quartz, (d) lignite and (e) anthracite .

100.0

'" N Vi

~ r I-

ffi ~

" ~ 10.0 ' .l '-' ~ '" > 1= ~ :> :>i :> 0 s ... ·"" u - Si ..... IMion

1.0

10 100 1000

SIEVING SIZE, pm

(a)

100.0

;:j Vi z < ~ ffi 10.0 Z

'" " u.s

~ '-' ~ w 1.0 > 1= ~ :> :>i 0 S.ui ... :> u _ S ..... t.lion

0.1

10 100 1000

SIEVING SIZE, pm

(d)

100.0

'" N Vi z g '" ~ " I- 10.0 r '-' ~ '" > 1= < :5 :>i :> U

1.0

10

100.0

;:j Vi z g '" '" ~ " I- . 0.0 r '-' ~ '" > 1= ~ :> :>i :> U

1.0

10

100

SIEVING SIZE. 101m

(b)

1000

100 1000 SIEV ING SIZE. I'm

(e)

IOO.O,---------"'V-r--,

0.1 +--____ ~~-___ ~~ 10 100

SIEVING SIZE, pm

(c)

1000

Fig. 3--Simulated and experimental product s ize di stributions o f dry gro und (a) calc ite, (b) barite, (c) q uartz , (d) ligni te and (e) anth rac ite.

OZKAN et al.: BREAKAGE PARAMETERS OF SOME MINERALS AND COA LS

100.0 100.0 1000

"' "' ~ N 9! ;;; 11 1

~ Z 11. Z

iO 11.0 g g

'" '" ' .l ::i "' "' 10.0 )O.~

~ 60 Z Z c: c: "if- "/. "I-.... 10.0 l .l .... 10.0 .... G " § Q

~ "' "' ~ ~ 1.0

l .a

"' "' "' > ~i""lin ... ,,""""CI > > 1= 1= 1= -< j j Grird"'I "irnc..mi .... lC . ..J ::J ::J ::J :>i . Sievin, :>i . Sicv;", :>i . Sicv;", ::J ::J ::J

_ Simul . .... " " U _ Sin .. lal lcoro U _ SimuIMIM.on V 1.0 1.0 0. 1

10 100 1000 10 100 1000 10 100 1000

SIEVING SIZE. ~m SIEVING SIZE. ~m SIEVI NG SIZE, ~m

(a) (b) (c)

100.'3 100.0

"' ~ N ;;; '" z z -< n.. -< " iO .... ::i '" 10.0 "' Z ~ "' ;f. ;f. !;: tr 10,0

is Q

~ to ~

"' 1.0

"' > > 1= 1= -< j 5 ::J Grif,dio,.titrc . ...... UIC' :>i . Sicvint :>i . Sic ~ int ::J ::J U _ Simulaiool U _ SimuL.1ion

0 .1 1.0

10 100 1000 10 100 1000

SIEVING SIZE.).lm SIEVING slzr:.. ~m

(d) (e)

Fig. 4--Simulated and experimental product size distributions of wet ground (a) calcite, (b) barite, (c) qual1z, (d) lign ite and (e) anthrac ite.

)0 30 30

;; 25

~ 20 'e

25

~ 20 'e

25

~ 20 'e uS ui uJ ::;: 15 ::;: 15 ::;: 15 ;:: ;:: ;:: :il 10 -' « "-

:il 10 -' « "-

OJ 10 V> ....

~

10 15 20 25 )0 )5 40 10 15 20 25 30 35 40 10 15 20 25 30 35 40

REAL TIME, minutes REAL TIME. minutes REAL TIME. minutes

(a) (b) (c)

30 30

25

~ .~ 20

25

" g 20 e

uJ ui ::;: 15 ::;: 15 ;:: ;::

"' 10 V> -'

OJ 10 V>

-' < "-~

10 15 20 25 )0 JS 40 10 15 20 25 )0 35 40 REAL TIME, minutes REAL TIME, minutes

(d) (e)

Fig. 5--False time (apparent fi rst order grindi ng time) versus real grinding time for dry and wet grinding of (a) ca lcite, (b) barite, (c) quartz,(d) lig nite and (e) anthracite .

273

274 INDIAN J. ENG. MATER. SCI, AUGUST 2003

I.0 -r+t--- ----------,

" - 0 .9

"' t < ~ 0.' ~ 8 ~ 0.7

;< ~

' .6

-+-""' ........ ~WctpMd~

0.' +--...--r_-.___,-....... -...:.,.~=-1 10 IS 20 2j 10 JJ

GRINDING T IME (I), minutes

(a)

1.0 ........ r-""t------ -----,

:E 0,9

'" t < ~ .. 8 ~ 0.7

i>

~ ' .6

.......... Dry"...,..

~Wapind" O~ +_-..._-r_~.___,-....... -~-~

10 " N ~ ~ ) 5 GRINDING TIME ( I), rm.u.n

(d)

1.0 .... ..----________ ,

~ 0.9

§ -< to.. 0.'

~ ~ 0'

~ 0.'

--- Dry .. iMine -*,""" ·Wct srin6i-.

'~ +---r--'---r--~~r-~~~ 10 IJ 20 15 )0 "

ORJNDlNG TIME ( I ), mirMcs

(b)

I.o .... rt-------___ --,

:g 0.9

§ -< ... • .1

~ 8 ~ ' .7

~ • .6

~ DIy"""

o~ +--..._-.--_,r_ ....... --+-~-w--~~.__·~ • 10 IS 2t l' )0 )S

GltIND/NG TIME ( I~_

(e)

I ~ ~p-------------,

:B: 0 .9

g -< 1.1. 0.' ~ 8 ~ 0.7

~ 0.'

-e- Ory p indirlt

-4- Wctpin6irla 0.' +--..._-.--_,r_ ....... -~-_=._-''_l

10 IS 10 23 )0 JS QRINDING TIME (I ). mtnukl

(c)

Fig. 6--Slowing-down fac tor (I() versus real grinding time for (a) calci te, (b) barite. (c) quartz, (d) ligni te and (e) anthraci te gri nding.

1.0

. •.. ~ 0.' z ~

8 0.' .; z i ~ •..

0.' 10

1.0

. ... g ! 0.' ~ . -8 0.' .; z i ~ 0.'

o~

10

100

I , I I

- ..,-- -... .,.,.. ~ PASSINO SIZE. ta,.

(a)

100

/

I I

, I I

I

- ..,-- - ............

10% PASSING SIZE, I'M

(d)

. g ~ 8 ~ ~ 0 01

1000

. ! t ! ~ 8

" z i ~

1000

I»

0.'

• .1

0.'

u

0.' 10

I.'

0.'

...

.. , •.. .~

10

100

I

/

- ..,--- ......... 10% PASSING SIZE. " .

(b)

100

I

,,/

- ..,-- -........ ~ P" SSING SIZE, .. _

(e)

I..

. ... !! ~ • .1

~ 8 .. , .; z Ii ~ u

- ""---wcs ..... .~

1000 10 100 1000

.... PASSING SIU. JU'II

(c)

1000

Fig. 7-Slowing-down fac tor (I() versus the 80% passi ng size of the product for (a) calci te, (b) barite, (c) quartz, (d) lignite and (e) anthracite in the mill.

OZKAN et al.: BREAKAGE PARAMETERS OF SOME MINERALS AND COALS 275

Figs 3 and 4 show the particle size distributions of dry and wet grinding of the calcite, barite, quartz, lignite and anthracite at various grinding times, respectively. The results of dry and wet grinding up to 4 min of grinding time for calcite, barite and quartz were simulated using the characteristics parameters of a, aT, y, <1> and ~ in the simulation program l4

. At longer grinding times, the simul ated results were finer than the experimental results, indicating a slowing­down effect. Slowing-down effect was seen at longer grinding times for lignite and anthracite; it started at 8 min for dry lignite grinding, while it started at 16 min for wet lignite grinding. The dry grinding of anthracite reached the slowing-down effect at 4 min, the wet grinding reached the slowing-down effect at 8 min of grinding.

The slowing-down effect was treated us ing the false time concept9 by making the simulator produce a match to a specified point on the product size distribution and designating the grinding time necessary to achieve this match as the false time 8, where 8:S;t. Fig. 5 shows the relation between the false time and the real time of all materials. Fig. 6 gives the relationship between the slowing-down factor (K) and the real grinding time for these minerals and coals. Differentiation of the curve in Fig. 5 gives the val ue of the slowing-down factor K.

K= d8/dt ... (5)

Fig. 7 shows the relationship between the slowing­down factor K and the 80% passing size of the particle size distribution in the mi ll. The grinding clearly changes its character when the material in the mill is finer than about 300 /lm. However, some materials showed sharp drops in rates, as the fines accumulated in the region of 80% passing sizes of about 100-300 /lm ranges, depending on the materials given in Fig. 7.

Fig. 8 shows the values of SI for calcite, barite and quartz minerals as a function of the y values . As the SI values increase, the y values decrease showing that faster breakage at top sizes is associated with a larger proportion of fines in the size distributions (lower y value). This relationship was found based on these three minerals ground as dry and wet in the laboratory size ceramic mi ll with an inverse linear equation as given in Fig. 8. The coals did not give the same relationships as the minerals given in Fig. 8 due to havi ng different petrographic composition and grindabi lity characteristics from those minerals l5

.

0.40 -r------------------,

\: 'E .-:: ~ 0.30

Ul 0 < .. ~ < ~ 0.20 (). SI = -0.30 r + 0.52 co u... 0 Ul f- a Barite (dry) ;:i • Barite (wet) U 0.10 t:L: (). Calcite (dry) • U .. Calcite (wet) <> Ul 0.. <> Quartz (dry) (/)

• Quartz (wet)

0.00

0.60 0.80 1.00 1.20 1.40 1.60 1.80 Y VALUE OF Bij PARAMETER

Fig. 8--Variation of SI values with the y value of Bi .i for cal cite, barite and quartz minerals.

Conclusions (i) The SI values, from higher values to lower

values, were in the order of anthracite, barite, calcite, lignite and quartz. Moreover, the SI values for wet grinding in the first-order breakage region were higher than the dry values for those materials ground in the ceramic mill, between 1.07 to 1.28 times as high .

(ii) The Bi ,j values for wet grinding were essentially the same as the dry values, however, the y values for wet grinding were lower than the dry y values , which showed that wet grinding had produced more fine products in the fines distribution region for all materials.

(iii) The slowing-down effect in the mi ll started at 4 min of dry grinding for calcite, barite, quartz and anthracite, and 8 min for lignite. However, the wet grinding was subjected to slowing-down effect in the mi ll at 4 min for calcite, barite and quartz, at 16 min for lignite and at 8 min for anthracite.

(iv) The simulations of the product size distributions for both first-order breakage and slowing-down were in good agreement with the experimental data.

(v) The breakage data obtained from the laboratory size batch mi ll could be applied to industrial applications by choosing the scale-up mill option in the simulation program for each material studied.

276 INDIAN J. ENG. MATER. SCI, AUGUST 2003

(vi) Variations of SI values with the y value of B ij

were found with the following relationships for calc ite, barite and quartz minerals; however, the similar equation could not be obtained for

anthracite and lignite . SI = -0.30 Y + 0.52

N omencla tu re aT characteri stic constant, min·1

Bi.1 cumulati ve primary breakage function of size 1; fraction broken to less than size Xi in one breakage

integer denoting 12 size interval Pi(O) cumul ati ve weight fraction of time 0 for size

interval i Pi( t) cu mulative weight fraction of time t for interval i P2(0) cu mul ative weight fraction of time 0 for the second

interval P2(t) cumulat ive weight fraction of time t for the

second interval Si specific rate of breakage of material of size i, min-I t time of grinding, min W total powder mass in the mill \VI(t) fraction of mill charge in size interval i Xu standard size Xi size of particles, mm ex characteri stic constant y

<I>

~ e K

characterist ic constant characteri stic constant characteristic constan t fa lse time, min slowing-down fac tor: ratio of specific rate of breakage at time t to normal specific rate of breakage at time zero

References Austin L G, Klimpel R R & Luckie PT. Process engineering of size reduction: Ball milling (SME/AIME, New York), 1984.

2 Datta A & Rajamani R K, 1m J Min Proc, 64 (2002) 181-200.

3 Tangsathitkulchai C & Austin L G, Powder Teclmol, 42 (1985) 287 -296.

4 Ozkan A & Yekeler M, Indian J Eng Mater Sci, 9 (2002) 181-1 86.

5 Bond F C, Br Chem Eng, 6 (1960) 378-391 , 543-548.

6 Tangsathitkulchai C, Powder Teclmol , 124 (2002) 67-75. 7 Klimpel R R, Powder Technol, 32 (1982) 267-277.

8 Kotake N, Suzuki K, Asahi S & Kanda Y, Powder Technol, 122 (2002) 101 -108.

9 Shah I & Austin L G, Ultrafine grinding alld separaTion of industrial minerals (AIME, New York). 1983,9- 19.

10 Austin L G, Yekeler M, Dumm T F & Hogg R, Part Part Sys Charac, 7 (1990) 242-247.

II Yekeler M, Ozkan A & Austin L G, Powder Technol, 11 4 (200 I) 224-228.

12 Ozkan A, Detennination of the breakage alld welling characteristics of minerals alld their correlaTiolls, Ph D Thesis, Dept of Mining Eng, Cumhuri yet University, Sivas, Turkey, 200 1.

13 Yon Seebach H M, Effect of va pours of organic liquids ill the comminution of cement clinker in tube mills (Research Institute Cement Industry , Dusseldorf), 1969.

14 Austin L G, Yildirim K, Luckie P T & Cho H C, Two stage ball mill circuit simulator (PSU, Univ Park, USA), 1989.

15 Rogers R S C, Austin L G, Industrial practice of fine coal cleaning (AIME, Littleton, Colorado) 1988,279-296.