and FuturePublishers

Reprinted from Mineral Processing, Recent AdvancesTrends, S.P. Mehrotra and R. Shekhar, eds. AlliedLimited, 1995, p. 159-170.

Recent Advances in Power Requirement andPowder (Product) Characteristics of Stirred Media Milling

Jie Zheng. Colin C. Harris and P. Somasundaran

Henry Krumb School of MinesColumbia University, New York, NY 10027

USA

ABSTR..f CT

Wltile applications of stirred media mills tor fine panicle production have continued to gro\v, there is alack of understanding of power requirements, optimum operating conditions and powder (product)characteristics underlying stirred media milling processes. Recent results of tests in laboratory stilTedmedia mills with media. limestone and )"ttria stabilized zirconia are presented and mill d:-namics.perfonnance and physico-chemical aspects are discussed. Four operational regions marked by sharptransitions are described: transition from static to dynamic friction; channelling; dispersion; andcentrifugation. Equations, including po\\'er and modified Reynolds number, have been established forrelating relevant operating and geometrical variables. Scale-up guidelines with respect to powerconsumption are also proposed. The best operating conditions for grinding limestone have been identified.EtTect of additives including that on the ground product is discussed using the example of ultrarinegrinding of zirconia. in which complexation by polymer was found to cause extraction of }"ttrium intosolution with resultant changes in surface ch~mical composition or" the product.

I. VTRODc'"CTION

It has been estimated that 25 billion kWh of U.S. electrical power production is consumed annually in orebeneficiation, with half of that spent in the grinding stage.! Tumbling, planetary, centrifugal. fluid energy.vibratory and stirred mills are primarily used in grinding operations. In comparison with other mills, stirredmedia mills (including Tower. sand, Coball and Perl mills) have especially anracted attention during recentyears because of their reported high energy efficiency, ability for grinding into the micron and sub-micronrange, and lower pro~uct contamination. Stirred media mills are used for fine particle production in manyindustries such as mineral. metallurgical. ceramic, electronic, pigments, paint and lacquer, chemical, bio-technology, rubber, agricultural, pharmaceutical, photographic, coal and energy.2-4 However, there is verylittle design information available, and scale-up of stirred media milling processes rarely reaches thedesigned capacity.S Detailed operating and performance characteristics of stirred media mills are alsoproprietary. This situation is in contrast to that in tumbling mills, for which information including type,dimensions, loading, speed and power consumption is freely available and reliable estimates of mill powerdraw can be made by means of dynamic and semi-empirical models.6 In this context, there is a need forfundamental understanding of power requirement, operating conditions, powder (product) characteristics,and the requirements for optimum performance in stirred media mills.

160 Min~raf Proc~.r.ring: R~c~nt Advanc~.r and Flltur~ Tr~,..d.r

In this paper, recent results of tests in vertical stirred media mills with media, limestone and ynriastabilized zirconia in our laboratory are given. Mill dynamics, operating performance and physico-chemical aspects in stirred media mills are also discussed.

MILL DY.VA,\1ICS

Power characteristics of stirred media mill have been studied by an approach developed from worlc onstilTed reactors operating under different conditions of impeller speed. design and dimensions, and mediaconcenU'ation, size and density. 7.8 Fig. I provides a schematic summary of distinct regions -displayed bytorque versus (4-pin) impeller speed curve for monosize glass spheres media with water as supernatant.This shows that the process passes through four regions: transition from static to dynamic friction;channelling; dispersion; and cenbifugation.7 Channelling is caused by cavities on the U'ailing edges of theimpeller pins. The progress from the region of channelling to that of dispersion results in the drop in torque(Fig. I CD), ~'hich is due to media dispersion into supernatant causing a drop in concenU'ation andconsequently a sharp lowering of viscosit)'. If there is no supernatant, the concenU'ation does not decreaseand the discontinuity does not occur. Because the dispersion region is relatively uniform and of knownconcenU'ation, it is especially amenable to analysis. To further understand channelling and dispersion f1o~.behavior, a model has been developed to predict power draw of multiple pin impellers from resultsobtained for milling with one pin situated at depths corresponding to each pin in the multiple pin case.s

TORQt.."E

A

I Bl'1PELLER SPEED

.A.B static to d~-namic friction regionBC chaMelling region; media rotation commences near CCD media begins to suspend into the supernatantDE media dispersing into available supernatantEF media full~. dispersed and uniform in concentrationFG ,,'o~exingGH centnfuging

Figure I: Schematic summar:-- of regions displayed by torque versus impcller speed cur\'e in media with supemauntliquid.

In order to establish the relationship betWeen po\\"er consumption and the process variables.dimensionless groups including power and Reynolds number are used. Several researchei'S9-10 use theviscosity and density of the liquid phase rather than those of slurTY and media combination to calculate

R~cenJ A.n-ancu in Power Reqmts. and Po\.'d,-r (Product) Characteristics o/Stirred .~le.iia .~li//ing 161

power and Reynolds numbers. An alternative method which is adopted in t:1e current \york considers themix(,.lr~ of pulp and grinding media as a non-Ne\\"tonian po\ver la\v liquid. Combining the pv\ver 13\"-equ:l.tion with the viscosity definitio.I and the assumption that the average liquid shear rate :s proportionalto impeller speed. I I the effective viscosit). can be obtained:

,u = Kan-I,Y"-' (()

A detailed procedure for determining the effective viscosity has been developed based on th~ t\...oassumptions of proportionality: shear rate to impeller speed; and shear stress to torque. '7 Relationshipsb~t\veen effective viscosit), (m). flo\v index (n). consistency (K), im~ller speed ~). and concentrationhave been evaluated for a number of design geometries, media particle size and solid concentration,7-S Theviscosity value can be incorporated into the Reynolds number:

.v2-.. D2 p

Kalf-I(2):VR~ =

Thtr~fore. th~ power number equation tor laminar conditions can be ~xpres:;ed as:

.\'~ = J,\":-. o2p l. I I

..i3)

Kan-1

wh~re C is a design factor dependent on impeller;tank geometry \vhich can be detemlined b~ employingthe pc\ver and Reynolds number correlation evaluated using Newtonia.'\ calibration liquids. For the (u!ldispersion condition [Fig. 1 EF]. the a','~rage dens it:-. can be calculated from the volumetric concentration.c. using the equation

0",'";.'t=Cp:; "'(l-c),q

Power and Reynolds number can be calculated using the viscosity and tile d~nsity ~.:tern1ir\.:d \\'ith.:quations (1) and (~), The relationship bet\v~en po\ver numb.:r and modi tied R':~"t1olds numb~r for ';.pln.I-pin and disc impell.:r at different conditions has be~n correlated,7.8 Figure 2 presen.s a summar:' of plotSof po\ver num'o~r VS, R~ynolds number for testS with 4.pin imp~llers for diff.:rer\t imp.:ll.:r sp.:.:d anddimensions. m.:dia size. concentration and d.:nsity. Th.: data is correlated by a straight line of slop.: -1 torth.: wide range of variables studied. Based on this result. certain scale-up guidelin.:s \vith r.:sp.:ct to po\verconsumption are propos~d: using a small mill to determine the values of C. K. n and a par~eters andassuming that they remain constant in scale-up. po\ver consumption for large units can be calculated usingthe equation:

P = CKan-',y"-'DJ (5

162.\linc,.a/ P,.oce.r.ring: Rec~nI Advance.r and FUlflre T,.~nds

100

. 6.':9.7:0.~':80

. 6.'; 9. 7;O.~'; sa

. 6.';9.7;O.~';55. 65;97:0~5:52

C 6.5;9.7;O.~5;50

0 6.5;9.7;O.~5;"

6 6.5;9.7;O,~';~5

0 6.'; 9.7;O.~5;40

. 10; 17;O.~';50

~ 1~;20.2;0.~';50

0 6.5;9.7;1;50

. 6.';9.7;1.05;50

~ 6.';9.7:~.~;50

X 6.5; 1.5; ~.~; 50

'to 6.5; 1.5;~. 7; 50 .

~ 10wc

~Z«w~0~ 1

10 100REYNOLDS NUMBER

1000

.~/JLL PERFOR,\-UVCE

R~c.nt Advances in Power R~qmlS. and Powder (Product) Characteristics ofSti,.,..d ."fedia ."tilling 163

0.8 0.3 - . .I Ef ~S 70

.9

0.03

0.7-M(8 Q.6(.)

0.5C\l(e 0.4

~0.3

.. \, \ 'T0

;~Ev~e%0.2-to.

.OJ~C"50. 0.10

e-

-M

0.02 (8

Co)

..c~'-'

0.01 >~

O'

.0

, ..-? .

., !

..' ,j

0' . ,.. ",'.

.0

: b: !c :~.oO,. . ~. :: . 0,0 ~

...::..-r/Q'

9c9

\

.

60 -.r.50~C\l(4oE

-~

300.2

0.1 0.0 - 120400 800 1200 1600

Impeller Speed. N (RPM)

0.000

Figure 3: Effect of impeller speed with the supernatant (conditions: D - 6.Scm; T = II.Scm; t - 15rnin; I: - 6000;

Rz3;d=3;da2.0Smm;V-1SOcm3).

The results with and without supernatant indicate that solid concentration is a very impor::1.,t factor inthe grinding process. Effect of total solids (media and particles) concentration on grindi.'g limestone isshown in Fig. ~. The torque or power increases with increasing solid concenc-ation, and t.".~ best encri~~fficiency is obtained at the solid concentration corresponding to the value at which the:~ is minin"..11supernatant liquid in the system. At the very high concentration (80%), only those solids t:nedia ar.dparticles) around the center of the impeller are stirred causing grinding of the contained F~iclcs. \\.hilet~yond the lmpeller pins the solids remain almost stationary and particles in that region re:n:!l:1 unground.Also wet grinding is more energy efficicnt than dry grinding (c=IOOOo).

In the study of the effect of solids concentration, solids include both media and paniclts. a.~d thtretore.the effect of their relative proportions must be evaluated. Specific energ). increases with incr~;lSlng ratio or-media to particle volume. and the ratio for the bcst energy efficiency and product finen~ss is 2.8 forgrinding limestone using glass beads as media. This corresponds to the ...oids in thc media ?:l.:king ~ingjust occupied b~. the limestone particles. This critical ratio of filling is given by

1-8.8.(1- s, )

Rz t61

and conflnncd in this case as em - O.~ and ep 20.47 giving R = 2.8.

Just as the ratio of media to particle volume has an effect on grinding, the ratio of media to feed particlesize may also have an effect. In fact. the optimum ratio of media to feed particle size is found to ~ 12: 1 fo.-

164 .\finera/ Processing: Recent Advances and Future Trends

the case of grinding limestone using glass beads. Furthermore, glass media is found to be more energyefticient for grinding limestone than steel. The best ratio may depend on media density and minerals beingground. Th~ optimum range of media to fecd size suggested in the literature i S is 7: I to the ma.-<imum 20: I.Fl>r the case of grinding coal using steel balls, the optimum ratio is reported to be 20:1.16

Dimensional relationships bet\\'een impeller. \vhich displaces the media to cause panicle fracture. andtank. containing media and panicles also affect grinding. The torque depends principally on the impellerdiameter but hardly on the tank diameter (see Fig. 5). The lower the ratio of tank to impeller diameter, thefiner the product. Ho\vever. the maximum energy efficiency is obtained at the highest ratio of tank toimpeller diameter.

it is clear from above discussion that the grinding process is affected by many operating variables,\\hich in turn influence pulp viscosity. A procedure to determine the effective viscosity during the grindingprocess rather than at its termination \Vas described earlier in this paper. Based on this procedure. theerTective \"iscosit)' values \vere calculated and the relationship be~'een effective viscosity and solidconcentration for the cases of media only. and media and particles combined are compared and shown inFig. 6 together \vith the best fit equations.

The colTelation bet\'"een grinding at1d energy input is described by the equations relating energyefficiency and increase of specific surface area respectively as a function of volume-based energy for mostcases studied (see Figs. 7 and 8). The results using the half 4-pin impeller are included in the colTelations,suggesting that grinding may not be very dependent on the particular design of this type of impeller.Ho\\"ever. some conditions, namely ,lower solid concentration, smaller media size, higher density and drygrinding. are excluded from the present colTelations, probably due to different and less efficient grinding.

Recent .4dvances in Power Reqmts. alfd Po".der (Product) Characteristics o/Stirred .\fedia .\filling 163'

1.4

0° :0 60 0,0:I

:;.~~;)~1.2 :..

:Jto::'O-e 1.0-.

D(cm) T(cm)0 10 11.8. 8.5 11.8~ 8.5 8.5

0.8 :-.i

0.8 :-

j-':j~':j

- ':;' ";' v0.4 i . i . . 'i i i i i . . '. . . .

;

0.20 2 4, 5 8 10 12 14

Grinding Time (Min.)Fisure S: Effect ot- impeller and tank diameter on torque.

.,n~-IOO"tl,I-(, 'O.68»)-I.S~

03 04 0.5 0.6 0.7 08 0.9

Solid concentration by Volume (~'o)

Figur~ 6: Th~ r~l:1tionship ~t~etn etTc:cti\'~ viscosit)' and solid conc~ntrotion.

z-Q)~ct~0

~

166 ,\{ineral Processing: Recent Advances and FUtUre Trends

.

.~ ~o 0 0

.V o~c 0

? 100 I I I III I'I .. I II II'I " I "11'1

~C\l<

E--C:J

~()CQJ . c=60%'(j v d= lmm 0S ~ steel mediaC:J 10 a dry grinding>.. . half 4-pin y

~ 0 all other conditions~ ~. I I ". 1,1 I I I 1"1.1 I I I I I I III

~ 0.001 0.01 0.1 1

Volume-based Energy, Ev (Wh/cm-3)Figure 7: Relationship between .:nergy efficiency and volum.:-bas.:d energ~'.

Specific energy (volume or weight based energy) has been used as the basic criterion for the scale-up ofstirred media mills.!.S.17 This specific energy is obtained by integrating average power over experimenttime. from ~'hich the power number is calculated. Modified Reynolds numbers are calculated using

10 I I I 11"'1 I I I , . "'1 '" I I I Illl. c=60% .

v d=lmm~ steel media Y

0 dry grinding. half 4pin0 all otherconditions

0

-C"){e(J

N{

e-C/)~

...

o. I " I I I I 1,1 I I I I , I " I I I , I I" ".

0.001 0.01 0.1 1

Volume-based Energy, Ev ( Wh/cm~3 )Figure 8: Relationship bct""een increase of specific surface area and volume-based energy

Recent Advc:nces in Pow«r R~qmt.s. and Powd«r {Product} Characteristics olStirr«d .\.fedia .\.filling 167

equation (2) and the procedure developed earlier. 12 Finally, d:-nensionless group correlations of power andmodified Reynolds num~r are established for relating the p;:~eters influencing the power consumptionfor media. panicles and liquid system in stirred mills.

J\.fILL PHYSICO-CHE.\.fICA r. ASPECTS

The effecu of chemical additives on grinding have been explained mainly by two kinds of mechanisms. ISOne mechanism is based upon alterations of surface and mechanical properties of individual particles, suchas reduction of surface energy19 and modification of surface hardness,20 while the other considers thearrangement of particles and their flow in suspensions, for example, improvement of pulp fluidiry.2\ Thesemechanisms have been examined for many twnbling milling cases.22-24 However, very few examplesconcern stirred media mills.

It is important to ensure that chemicals used as grinding aiJs do not lead to detrimental side-effects. Inthis regard, ultrafme grinding of yttria stabilized zirconia in polyacrylic acid solutions studied in thislaboratory using a Netzch batch attrition mill showed some important effects.11 Relatively high grindingrates-as measured by production of new surface-were demonsU'ated using zircon media balls of 1.2 mmdiameter. The absence of contaminating materials such as steel grinding media provides an additionaladvantage: high purity product is important in material preparation for high performance ceramicapplications. Interestingly, it was found that the poly acrylic acid added as the ~nding aid caused changesin the chemical composition of the product due to preferential extraction of )"ttrium by the polymer. Thisfmding is shown in Fig. 9 where yttrium concentration in the supernatant is ploned as a function ofequilibrium polymer concentration. It is noted that no dissolved yttrium \\-3S detected in the supernatant inthe absence of polyacr'jlic acid. For both polymers, the yttrium concenU'ation in the supernatant gradually

40N

..EC):]- 30

c0:;:~-c 20Q)uc0uE 10:]

.~-

):0

0 1000 2000 3000 4000

Equilibrium concentration, ppmFigure 9: Amount of >'nrium rclca.sc:d by thc surface of zirconia vs thc equilibrium concenU3tion of po\y~~"\ic 3c:id.\ S

168

c ~~~ :cco. P.cA 15C.CCO&

iQc

S!--->-

0--- 1 0cA/"

I c,C

..I~4'

...,,

~lI

!IJ,

iI,:'..~

I.

05:X) lOCOTinw in mi~..es 1~

CO.\'CLl'SIO.\:r.

PO\\ ~r charact~ristics of stirr~d m~dia mills hav~ b~en stUdied as a function of relevant op~rating

and g~om~tric31 variabl~s. Th~ torqu~ versus s~ed CUf'-e has b~en found to display four "r~gions

marked by sharp transitions: transition from static to dynamic friction. channelling. dispersing. and

centrifuging. Equations. including po\\"er and modified Reynolds numbers. have b~en establishedfor correlating the variables studied. Based on this work. scale-up gui~elines ha\'e been proposed.

po\\er requir~m~nts and product characteristics of grinding limestone in laborator')"-scale Stirredmedia mills have be~n stUdied \\"ith r~srect to several operating v3riabl~s- The beSt conditions forgrinding limestone have been identified. Correlations bet\\"een grinding and energy input for mostconditions ha\ e been established using th~ relationship bet\\een energy efficiency and tht increaseof specific surtaCt arta resptctively as a function of volume-base-rl ..ner..,

169Influenc~ ofSurfactams on Filtl'atio" and Dcwat~I'ing Chal'act~ristS in Vacuum Filtration

which stresses the importance of the effects, panicularly those that are detrimental on downstreamprocesses in the over-all industrial operation.

ACK.VOWLEDGE."1E,VT

This research is supported by the United States Bureau of Mines through the Generic Mineral TechnologyCenter for Comminution under Grant Number G1125249. Gl13S249 and Gll~52~9.

SY.WBOLS USED

c Volumeh"ic conc~ntration of solidsC Impeller geometry factor constantd Media particle diameterD Impeller diameterE Energy inputEf Energy efficiency (- OSlEv)Ev Volume-based energy (- EN)K Consistency coefficientn Power law indexN Impeller rotational sp~ed~p Power number~Rc Re~-nolds numberP Po\verR Ratio of media to particle volumeS Specific surface areacS Increase of specific surface areat Grinding timeT Tank diameterV Volume of ground material3. Impeller shear rate constant~ Media p&Cking porosity~ Particle packing porosity!.1 Viscosityp Average media dens it)'I)! Densit)" of liquid1)$ Dens it), of solid particlet Torque

REFERE.VCES

I.2.

3.~.

N:ltional ~tateri31s Adviso~' Board: Comminution Ii; En~' Consumption. Report 364. Washington. DC. t 98 I.J. L. S. Jimenez: A Dctailing Study on Stirred Ball ~til\ Grinding. Ph.D. Dissertation. Univcrsi~ of L't31-- SaltLak~ Ci~'. 1981.~. Stehr: Int. J. ~tir.cr. ?rocess.. t988. 22. pp. ~3t~~~.R. Shanna. D. A. Czebi. and J. Texte : First Intem:ltion:1l Particle Technology Forum. R. Davics. R. Pfeffcr. 7\{.Roco and B. Ennis. cd.. Part 2. pp. ~2~7. AIChE. ~e\.. York. 199~.

.70 ."(in~ral Processing: Recent Ad\'ances and Future Trends

S. Q. Q. Zhao. J. W. Laporte. J. M. Leithren and J. J. Harrington: First International Particle TechnolOg)' Forum.R. Davies. R. Pfeffer. M. Roco and B. Ennis. ed.. Part 2, pp. 142-147. AICh£. N~v York. 199~.

6. C. C. Harris. E. M. Schnock and N. Arbiter: Miner. Process. and Technol. Revi~'. 1985. 1. pp. 29"i-34S.7. J. Zhcn&. C. C. Harris and P. Somasundaran: Miner. and mctallur&ical Process.. 1995. 3~. pp. 3+-40.8. J. Zheng. C. C. Harris and P. Somasundaran: First International Particle Technolog)' Forum. R. Davies. R. Pfeffer.

M. Roco and B. EMis. ed.. Part 2, pp. 13.s-141. AICh£. New York. 1994.9. L. Y. Sadler III. D. A. Stanle)' and D. R. Brooks: Powder Techno\.. 1975. 12. pp. 19-28.10. H. "'eit and J. Schwedes : Chern. Eng. Techno\.. 1987. 10. pp. 39&-404.11. A. B. Metzner and R. E. Ono: A.I. Ch. E. Journal. 19S7. 3. 3.12. J. Zhcng. C. C. Harris and P. Somasundaran: SME Annual Meeting. 1995. Denver. Colorado. Preprint 9.s-17S.13. M. J. Mankosa, G. T. Adel and R. H. Yoon: Powder Techno\.. 1989. S9. pp. 25.s-260.14. M. W. GIO and E. Forssberg: Int. J. Miner. Process.. 1993,37. pp. 4S-S9.IS. R. F. Conley: Conference Proceedina of the Ultratine Grinding and Separation of Industrial Minerals. S.G.

Malghan and P. Somasundaran ed.. pp. 37-48. SME-AIME, New York. 1983.16. M. J. Mankosa, G. T. Adel and R. H. "oon: Powder Techno\.. 1986. S9. pp. 25.5-260.17. H. Weit, J. Schwedes andN. Stehr: 6d1 European S~-mposium Comminution. pp. 709-723. Number!. 1986.1 S. B. Larti&es and P. Somasundaran : Special S~.mposium Proceedings: Comminution-Theory and Pnctice. S~IE

Annual Meeting. 1992, pp. S8.5-S98. SME-AI~(E. New York.19. P. A. Rehbinder: physik, 1931.72.191.20. A. R. C. Westwood and D. L. Goldheim: J. Appl. Phys.. 1968. 39. 3~OI.21. R. R. Klimpel: Reagents in Mineral Technolog)'. P. Somasundaran and B. ~L ~Ioudgil cd.. ~larc=1 Dekker. Inc..

New York. 1987. pp. 179-194.22. P. Somasundaran and I. J. Lin: I and EC Processes Des. Dev.. 1972. 11.321.23. H. EI-Shall and P. Somasundaran : Po..vder Techno\.. 198.J. 38. pp. 2i.s-293.24. R. R. Klimpel and W. Manfro~. : Ind. Eng. Chern. Precess Des. Dev.. 1978. 17. S 18.2S. A. Foissy. A. EI Anar and J. M. Lamarche: J. Colloid and Interface Sci.. 1983. Vol 96. No.1. pp. 27.5-287.26. 26. L. Burlamacchi. M. F. Ottaviani. E. M. Ceresa and M. Visca: Colloids and Surfaces. 1983. \.01. 7. pp. 16~-

181