-CHAPTER 9

ABSTRACT passes through four regions: transition from static to dynamic .'r:c-tion; channelling; dispersion; and centritl.;gation. The progress frcmthe region 01 chan~elling to !hat of dispersion results in u-.e drop ~torque. which can be accounted for by media dispersion into super-natant causing a drop in concentration and consequently a sharplowering of viscosity. Because the dispersion region is relativelyuniform and of known concentration. it is especially amenable !oanalysis.

While applications of sti"ed mecia mills for line particle produc-tion have continued to increase, there is a lack of understanding ofoperating behavior and power rec;uirement. Investigations in labo-ratory sIi"ed media mills have been carried out with grinding media.limestone and yttria stabiDzed zlrconia. The torque required to rotateimpellers immersed in dense paI1iculate media with supernatantversus impeller rotational speed displays four regions mar1<ed bysharp transitions: transition from static to dynamic friction: channel-ling; dispersing; and centrifuging. Equations, including dimension-less group co"elations of power ar:d modified Reynolds number,for relating power, Speed. impeller ard tank dimensions ~d design.media size ar:d density, solid concer:tration, and other relevant varI-ables have been established. Scale-up guidelines with respect :0power consumption are also proposed. The best operating condi-tions for grir:ding limestone in the laboratory stl"ed media mill havebeen '.denlified. Complexation by the polymer during uJtrafine grind-ing of zirconia has been found to cause extraction of yttrium intosolution with possibly significant changes in the surface chemicalcomposition of the product

INTRODUCTION

ThfPELLER SPEEDAB st.tti.: to d)1WUic friction rqion

BC channeling reii~ moiia rotation commences near ,

CD media begins to suspend into the JUpa'!1atant

DE media dispersing into available supematmt

EF moiia fully dispersed and unif~ in coocena-atiOQ

FG von~g

G H cen ai fugina

It has been estimated that 1.3-'0 of U.S. electrical power prOCl;C-tion is consu~ed for comminutior:. particularly by fine grinding (Natlcnal Materials AdviSOry Board. 1981 ). Stirred media mills ha..eespecially attracted attention because of the reported high energyefficiency. ability for grinding into the micron and sub-micron range.and reduCed contamination. Stirred media mills have been appliedfor fine particle production in many industries such as mineral. ce-ramic. metallurgical. electronic. pigments. paint and ~. Chemi-cal. big-technology. rubber. agricultural. pharmaceutical. photo-graphic. coal and energy ( Jimenez. 1981: Stehr. 1988; Sharma.Czekai and Texter. 1994). However. detailed operating infonnationappears to be proprietary. and there is a lack of fundamental under-standing of operating behavior and power requirement of thesemills. Therefore. experiments in labOratory vertical stirred mediamills were carried out. The results of tests with media. limestoneand zirconia in our laboratory wig be presented.

Figure 1. Schematic Summary of Regions Displayed by TorCt.eVersus Impeller Speed Curve in Mexia With Supernatant LJquic.

In order to establish relationships between power consumptcnand the process variables. dimensionless groups such as powerand Reynolds number are used. Many researdters (Sadler et aI..1975; Jimenez. 1981; Weit and Schwedes. 1986 ) use the viscos-ity and density of the liquid phase rather than those of sluny ~dmedia combination to calculate power and R~s numbers. Analternative method in the current work considers the mixture of :i<;-uid and grinding media as a non-Newtonian power law liquid. Cc~-bining the power raw equation with the viscosity definition and ~eassumption that the average liquid shear rate is proportional toimpeller speed (Metzner and Otto. 1957). the effective viscosity canbe obtained:

MILL DYNAMICS WITH MEDIA ONLY

Power charaderistics of stirred media mill have been studied byan approach developed from wOr\( on stirred reactors and underdifferent conditions of impeDer Speed, design and dimensions, aOOmedia concentration, size and density (Zheng, Harris andSomasundaran. 1994a; 1994b ). The grinding System is batch witha 4-pin impeller immersed in a media of monosize glass spheres inwater with Supernatant. The schematic summary of torque versusimpeller speed is given in F'lQure 1. This reveals that dte process

48 PROCEEDINGS OF TJoiE XIX IMPC

100IJ. - Ka""N'"(1)

A detailed procedure to determine the effective viscosity hasbeen developed based on the two assumptions that shear rate isproportional to impeilef' speed and shear stress to torque. Relation-ships between effective viscosity ( ~). flow index ( n ). consistency(K). impeller speed ( N ). and concentration have been evaluatedfor a number of design geometries. media patfM:le sizes and soIkjconcentrations ( Zheng. Harris and Somasundaran. 1994a; 1994b). The viscosity value can be incorporated into the Reynolds num-ber:

10

I ~N~.~~~ (2)

Therefore. the power number equation at laminar conditionscan be written as

0.1 , J1

J ,." ""I '..' "".

10 100 1000REYNOLDS NUMBER

N~. C [ !J.~~ r (3)

where C is a design ~ dependent on impeIerl1ank geometrywhich can be detemined by using the power and Reynolds numbercorrelation evaluated using Newtonian liquids ( Zheng. Harris andSomasundaran. 1994a ). For 1t1e f\JII d"lSpersion condition. the aver.age density can be calculated from the volumetric concentration, c.usi~g the equation

P - CPa + (1-C) P, (4)

Power and Reynolds number can be calculated using the vis-cosity and density determined from equation (1) and (4). Fig. 2 pre-sents a summary of plots of power number vs. Reynolds number fortests with 4-pin impellers tor different impeller speed and dimen-sions. media size. concentration and density: the correlation is ap-proximated by a single straight line of slope -1 for the wide range ofvariables studied. Based on the above results. cenain scale-upguidelir:es with respect to power consumption are proposed: usinga smail mill :0 determine the vahJes of C. K. n and a parameters andassuming that they remain constant in scale-up. power consump-tion for larger units can be calculated using the equation: FIgUre 2. Power NumberVesus Reynolds Number tor 4-pin Impel-

ler tor Different Conditions. The Order of the Tabular Data is D(cm);T(cm); d(cm); c(%) by VohJme. All Refer to Glass Media Exceptthe Last Entry for Steel Balls.

P . CKa,"'N""'DJ (5)

MILL PERFORMANCE IN GRINDING LIMESTONE around the center of the impeller are stirred causing partides con-tained to be groond. while beyond the impeller pins the solids re-main almost stationary and partides Slay unground. Also wet grind-ing is more energy efficient than dry grinding (c=100~.).

Power requirement and product characteristics lor stirred mediamilling 01 limestone have been studied with respect to a variety 01variables (Zheng. Harris and Somasundaran. 1995). Lower stirringspeed has been lound to give better energy effICiency lor grindinglimestone without any supernatant. This finding is in accordancewith ~. results of Mankosa. Adel and Yoon ( 1989 ) lor grindW1g coaland Gao and Forssberg ( 1993 ) lor grinding dolomite. However. lorthe case of grinding with supernatant. the best energy efficiencyoccurs at the stirring speed corresponding to the lowest point intorque versus speed curve. in the .onset of dispersion" region ( seeFig. 3 ).

The ratio of media to particle volume lor the best energy effi-ciency is 2.8 lor grinding limestone using glass beads as media.This OOIrespondS to the yom in the media packing being just occu-pied by the limestone particles. This critical ratio of lliling is given by

R. - 1. ".t_- , (6)t. \ I . t. Iand confWmed in this case (tm.Q.4 and '0=0.47 give R.2.8)

Effect of total solids ( media and partides ) concentrations byvolume on grinding limestone is shown in Fig. 4. The solid concen-tration for the best energy efficiency corresponds to the value atwhich there is minimal supernatant liquid in the system. AI the very

high concentration ( ~ ). only the solids ( media and particles )

The optimum ratio of media to feed particle size is found to be12:1 lor the case of grinding limestone using glass beads. Further-more. glass media is more energy efficient for grinding limestonethan steel (Zheng. Harris and Somasundaran. 1995 ). The bestratio may depend on media density and min«aJs ground. The op-

"3:

;2

POWER AND OPERATING BEHAVIOR IN STIRRED MEDIA MILLS 49

0.8

0.7

0.1

0.5

0..

0.3

0.2

M,

eC.I

C\J,

e-

rn<3

o.

Figure 3. Effect of Impeller Speed with the Supernatant (Condi-tions: O=6.5cm; T-11.8cm; t=15min; c-60%; R-3; d-2.05mm;V=150 cm3).

Figure 5. Relationship Between Energy Efficiency and Volume-based Energy (All Other Conditions: 0: 6.5 - 10 cm; T: 8.5 - 11.8cm; N: 500- 1500 rpm; t: 5 - 15 min; c: 65 - 800/0; R: 1 - 4; V:60 - 150 cm3; d: 2 - 4.4 mm; Glass Beads and 4-pin Impeller).

Figure 6. Relationship Between Increase of Specific Surface Areaand Volume-based Energy (All Other Conditions: the Same asThose in Figure 5).

Figure 4. Effed of Total SoIkj Ca1cen1raOOn (~: 0-6.5 an:T-11.8an; t.15min; N-1000rpm; R-3; d-2.05mm; Va150an3). in the correlations, which implies that grinding may not be too de-

pendent on the design of this type of impeller ( Zheng, Harris a~dSomasundaran. 1995 ).timum range of media to feed size suggested by Conley ( 1983 ) is

7:1 to the maximum 20:1. For the case of grinding coal using steelballs. the optimum ratio is reported to be 20:1 by Mankosa. Adel andYoon (1986).

The current work is leading to the following conclusions: spe-cific energy ( volume or weight based energy) may be used as thebasic criterion for scale-up of stirred media mills. and specific en-ergy can be obtained by integrating average power over experimenttime. from which power number is calculated. Modified Reynoldsnumbers are calculated using the similar procedure developed ear-lier ( Zheng. Harris and Somasundaran. 1994a ). Finally, dimen-sionless group correlations of power and modified Reynolds num-ber may be established for relating the parameters influencing thepower consumption for media. particles and liquid system in stirredmills.

The torque is found to depend principally on the impeller diam-eter but hardly on the tank diameter. The lower the ratio of tank toimpeller diameter. the finer the product. However. the maximumenergy efficiency may be obtained at the highest ratio of tank toimpeller diameter ( Zheng. Harris and Somasundaran. 1995 ).

The correlation between grinding and energy input is found tobe described by the equations derived from energy efficiency andincrease of specific surface area respectively as a function of vol-ume-based energy for most cases studied ( see Fig. 5 and 6 ). How-ever, there are some conditions, such as lower solid concentration.smaller media size. higher density and dry grinding, which are notincluded in the correlations probably due to different grindingmechanisms. The results using the half 4-pin impeller are included

Mill PHYSICO-CHEMICAL ASPECTS IN GRINDINGYTTRIA STABILIZED ZIRCONIA

Ultrafine grinding yttria stabilized zirconia in polyacrylic acidsolutions has been studied in this laboratory using a Netzch batch

50 PROCEEDINGS OF THE XIX IMPC

ACKNOWLEDGMENTI - -- -, -~ -~i -

I :tJl .:.:;= ., ~t I -"' ., .! ! . .

.-~&..;;.""' ..;;.""'-,~,..' O./0

.~ ~

This partial research has been SL;~.oorted by the Cec~ment ofInterior's Mineral Institute Program administered by :~e UnitedStates Bureau of Mines through the Generic Mineral Te-::!'1nologyCenter for Comminution under Grant NumberG11452~9 '.'Je thanKProfessor P. Somasundaran for helpful discussion.

--.:=

-.:Dr

Symbols Used./,.

~/)0

a I~ ~

r! ./- , .: '~;'r ' .. ~ ~

I '..: ~

/./_°'0 SJ) ::;m ,~ ?:OJ ~ :n 3:0 Ja:Q

::=.0, I 'o,.' ~. ,n ~

Figure 7. Amount of yttrium Released by the Surface Zirconia vsthe Equilibrium Concentration of Polyacrytic Acid.

attrition mill ( Lartiges and Somasundaran. 1992 ). Reiative-ly highgrinding rates. as measured by production of new surface - weredemonstrated using zircon media balls of 1.2 mm in diameter. Theabsence of contaminating materials such as steel grinding mediaprovides an additional advantage: high purity product. important inmaterials preparation for high performance ceramic applications.Interestingly. it was found that polyaCtyfic acid added as grinding aidcaused changes in the chemical composition of the product due topreferential extraction of yttrium by the polymer. This finding isshown in Fig. 7 where yttrium concentration in the supernatant isplotted as a function of equilibrium polymer concentration. It isnoted that no dissolved yttrium was detected in the supernatant inthe absence of polyacrylic acid. For both molecular weights. the yt-trium concentration in the supernatant gradually increases with theresidual polymer concentration and then reaches a plateau. Rea-sons for this are not yet known even though complexation betweenpoly acrylic acid and yttrium can be expected to playa role.

c Volumetric concentration of solidsC Impeller geometry factor cor-.stantd Media particle diameterD Impeller diameterE Energy inputEf Energy efficiency ( d . Ds. Ev )Ev Volume-based energy ( = E. V )K Consistency coefficientn Power law indexN Impeller rotational speedNp Power number ( . P/(pN305 ) )NRe Reynolds number ( = N02p.'~ )P PowerR Ratio of media to particle volumeS Specific surface area~ Increase of specific surface areat Grinding timeT Tank diameterV Volume of ground materiala Impeller shear rate constantEm Media packing porosityEp Particle packing porosity~ Viscosityp Average media densityP1 Density of liquidPs Density of solid particlet Torque

CONCLUSIONS REFERENCES

(1) Power characteristics of stirred media have been studied asa function of relevant variables. The torque versus speed curve hasbeen found to display four regions marked by sharp transitions:transition from static to dynamic friction, channelling, dispersing,and centrifuging. Equations, including power and modifiedReynolds numbers, have been established for correlating the stud-ied variables. Based on this won<., scale-up guidelines have beenproposed.

Conley, A.F.. 1983. "Attrition Millirg of Industrial Mirerals', Con-ference Proceeding of the Ultrafine Grinding and Separa:ion of In-dustrial Minerals. Malghan, S.G. and Somasundaran P. eds. NewYork, SME-AIME, pp. 37-48

Gao, M.W. and Forssberg, E., 1993. "A Study on the Effect ofParameters in Stirred Ball Milling", Int. J. Miner. Process., 37, pp.45-59

Jimenez, J.L.S., 1981, "A Detailed Study on Stirred Ball MillGrinding", Ph.D. Dissertation, Universrty of Utah, Salt Lake City

Lartiges. Bo. and Somasundaran. P.. 1992. "Ultrafine Grindingof Yttria Stabilized Zirconia in Polyacrylic Acid Solutions". Specialsymposium Proceedings: Comminution - Theory and Practice. SMEAnnual Meeting. Phoenix. AZ. Feb. 24-27

(2) Power requirement and product characteristics of grindinglimestone in stirred media mills have been studied with respect toseveral variables. The best conditions for grinding limestone havebeen identified. The correlation between grinding and energy in-put for most conditions have been established using the relationshipbetween energy efficiency and increase of specific surface arearespectively as a function of volume-based energy.

(3) Physico-chemical aspects in ultrafine grinding of yttria sta.bilized zirconia in polyacrylic acid solution in stirred media mills havebeen studied. Effect of poly acrylic acid as grinding aids has beenfound to cause extraction of yttrium into solution with significantchanges in the chemical composition of the product. and complex-ation between polyacrytic acid and yttrium is proposed to be thereason for the above effect.

Mankosa, M.J., Mel G. T. and Yoon, A.H.. 1986, .Effect of Me-dia Size in Stirred Ball Mill Grinding of coar. Po'Mier Technology,59, pp. 255-260

Mankosa, M.J., Adel, G.T. and Yoon. A.H.. 1989. 4Effect ofOperating Parametef$ in Stirred Ball Mill Grinding of CoaI8, Powder

POWER AND OPERATING BEHAVIOR IN STIRRED MEDIA MILLS

Int. J. Miner. Process.. 22. pp. 431-444

51

Technology. 59. pp. 255-260

Metzner. A.B. and Otto. 1957. R.E.. A.I.Ch.c. Journal. 3. 3 Weit, H.. and ~es, J.. 1987, "'Scale-up of Power Consumptionin Agitated Ball Mills", Chern. £ng. Tec:hm.. 10. pp. 3S8-:C4

NationaJ Materials Advisory Board. 1981, .Comminution & En.ergy Consumption", Report 364, Washington, DC Zheng. J., Harris. C.C. and Somasundaran. P.. 1994a. .Power

Consumption of Stirred Media Mills., Preprint Number 94-118. SME'Annual Meeting. Albuquerque. New Mexico. Febru2l"f 14-17Sadler, L.Y.III. Stanley, D.A.. and Brooks. D.A., 1975, "Attrition

Mill Operating Characteristics", Powder Technology, 12, pp. 19-28Zheng. J.. Harris. C.C.. and Somasundaran. P.. 1994b. .Power

Characteristics of Stirred Media Mills". First International ParticfeTechnology Forum. Denver. USA. Part 2. August 17-19. pp. 135-141

Sharma. R.. Czekai, D.A., and Texte. J., 1994. . ElectrokineticCharacterization of Attrition Products Derived from Zirconia-SilicateMilling Media", First International Par1icJe Technology Forum, Den-ver, USA, Part 2, August 17-19, pp. 42-47

Zheng, J.. Harris. C.C.. and Somasundaran. P.. 1995. "A Studyon Grinding and Energy Input in Stirred Media Mills". Preprint Num"ber 95-175. SME Annual Meeting. Denver. Colorado. March 6-9

Stehr. N.. 1988. "Recent Developments in Stirred Ball Milling"