ELSEVIER Powder Technology91 (1997) 165- I?2 L Using flow perturbations to enhance mixing of dry powders in V-blenders Dean Brone ~, Carolyn Wightman a, Kevin Conner a, Albert Alexander ~, F.J. Muzzle '~'*, Priscilla Robinson b * Department of Chemical and Biochemical Engineering. Rutgers University, Piscataway, NJ 08855-0909, USA I, Merci: Manufacturing Division. West Point, PA t 9 4 8 6 , ~ , USA Received 13 March 1996; revised 25 August 1996 Abstract Experiments were conducted to compare ndxing performance in a conventional V-blender and in a V-blender that incorporates perturbations of the particle flow by rocking the mixing vessel during rotation. Mixing was investigated using glass beads with sizes from 40 to 800 pm in vessels of approximately one liter volume, Mixture uniformity was assessed qualitatively using two different methods. One method used a transparent mixing vessel to visualize particle flow patterns and assess the state of homogeneity at the mixture's surface during the entire experiment, The second method involved solidification of the mixture by infdtration with a binder inside disposable aluminum mixing vessels. Using this method, it was possible to assess the state of the entire mixture, including its interior structure, by slicing the solidified structure after completion of each experiment, Mixture uniformity was also assessed quantitatively using image analysis to determine the composition of the solidified samples, In all cases, mixing was greatly enhanced in the rocking V-blender compared to the conventional Voblender. Keywo~**: Mixing; Bl~nder; Granular flow I. Introduction Voblenders are widely used in many industries requiring the blending and processing of powders. They arc used both in the laboratory as smalloscale product development units and in manulhcturing as largeoscal¢ production units. Vo blenders are a type of tumbling mixer, consisting of two hollow cylindrical shells (Fig. 1), joined at an angle usually in the ?0%90° range. The mixing vessel is typically connected to a rotating shaR causing a tumbling motion of the powders within the vessel. All commercial Voblenders use a constant-speed tumbling motion to mix powders [I-12]. There have been several reports of slow or incomplete mixing in these devices. (In most cases a system is considered to be well mixed when the standard deviation of samples taken from the mixture is equal to the standard deviation of a random mixture.) Gray [ 1 I found that a mixture of sand and iimenite continued to improve its mixedness even after 1000 revolutions at 24 rpm. Wiedenbaum et al. [2] found that sand and salt particles of similar sizes did not approach a random mixture even alter 5000 revolutions at 24 ~m. Chowhan and Linn [3] found * Conesponding author. Tel.: + 1 908-445 3357; fax: + 1 908,445 .5313; e.mail: muzzio@sol.mtgers.edu, 0032-59t0/97/$17.00 © 1997 Elsevier Science S.A. All rights reserved Pll S0032.5910 (96) 03 23 t .7 Fig. I, Sci~¢mali¢diagram of Ihe V:blender m|xtng ve~el~ that it took approximately i100 revolutions at 24 rpm to obtain a wellomixed system of a cohesive drug with a free flowing ¢xcipient, Cahn ¢t al. 14] needed 10()0 revolutions at 24 rpm to obtain a well.mixed system of same-size CaCO~ and SiO~ particles, Carstensen and Patel [5] found that a system of same.size lactose and corn starch was not sufti° clearly mixed after 500 revolutions at 24 rpm. Harnby [6] found that a mixture of millet and salt exhibited significant segregation after 1000 revolution~ at 33 rpm. Samyn and Mt,rthy [7] found that 118 p,m aspirin and 87 p,m lactose took 60 minutes to mix (the rotation rt,,e waz mot ~,pecificd). The main type of mixing limitation found in these experio ments was slow mixing in the axial direction, especially 166 D Ilnme el al, / Powder T ¢ , ~ o g y 91 (1~7) 165-17,7 between dl¢lls. A ¢onventio,~l V.blender has no mechanism to induce flow in ~ axial direction, hence composition differences betm:en shells may persist for long times. Two approaches have been ~ in previous studies to solve this p r o b e . The first involved changing the geometry of the mixer. Cahn et al. [ 4] found that a V-blender of slightly diffenIat geometry, formed by changing the alignment of the ~ i l s with ~ t to ore ~ t h e r . could produce axial flow. effect was most pronounced when the two shells were aligned 18~ with respect to one another, creating a geometry i d e n t ~ m an inclined drum mixer. A similar approach is to a %blender with s~lls of different lengths (P~ K Cross° lRowTM B ~ r , Patterson Kelley, East Stroudsburg, PA). An axial flow from one shell to the other ~Cr~ due to the different limes necessary for particles to tumble to the center of the blender from the e n ~ of the shells. The second approach involves modification of the operation of the Vblender, for example by the pre~blending of components prior to loading the V-blender or by di~harging and reloading the componentsrepeatedly during the mixing process. Both geometric and operational approaches have drawbacks. Chang~ ing the gcometw of the mixing yes.Is can involve the costly pu~hase of new equipment. Relying on methods which involve ~veral imermediate processing steps can lead to complicated tim¢oconsuming proceduresand i~producible results, In thb paper, flow perturbations of con~lled amplitude frequency ~ u~d to enhance axial mixing in the Vo t~nder ves~el, Such perturbations ~ introduced by ~ k i n g th~ device with re:~pcet Io its axis, ~ u c i n g a conv~tiv¢ axial flow, resulting in I~ge aceelera|ions of the mixin,gproco ¢~, Mixing ~rformanc¢ i~ qualitatively and quantitatively compared for a standard V-blender and a V-blender of improved design. 2. Experimental A custom-designed mixing apparatus (Sincro Motion, Valley Forge, PA) was builtto examine the effectsof wellcontrolled flow perturbationson mixing processes inside a partially filledV-blender (Fig. I). Rotational and rocking motions (Fig. 2) were independently controlled using two stepping motors (Arrick Robotics, Hurst, TX) interfaced to a computer (Gateway 2000, North Sioux City, SD). One motor was direcdy linked to the drive shall and enabled the mixing vessel to rotate,. The other motor was linked to the frame that h o ~ the drive shaft by means of a screw shaft. It enabled the frame to rotate partially around a pivot and imparted a vertical rocking motion on the mixing vessel. A motor-control computer program was developed to enable precise independent control of both the rotational and rocking frequencies, During a given experiment, the rotation rate, mixing time and number of rotationsper rocking cycle were specified,with the remaining parameters necessary to control motors then being calculated. Each rocking cycle con° sisted of a lop downward tilt, followed by a rise back to the horizontal position and a 10P tilt in the opposite direction (Fig, 2), A rocking cycle was complete when the mixer returned to the horizontal position. Two types of experiments were used to demonstrate the mixing enh~cements obtained by the application of rocking motion, TI~ experimental techniques used in this study have been d¢~fibed elsewhere I I 3]; only a brief description is l-- ,it m~lea I .10° (a) (b) (c) (d) Fig, .~, V-blenderk~ldingleehnkiuesllowinB( a ) plungerinse.iou iluoone end, l'ollo~edby ~b) loading one color of heads into the lowershell, After adljnslingthe plunger1oalign the interfacealongthe cenlerlineoflhe vessel, (¢) heads oi" a difl'e~enlcolor a~e ¢~l'ully poured onto the initial layer, A{lerIlllinB tile vessel uprighl, I d) a ntable verlicalinterfileeis obtained, p~sented here, In the lirst experiment, direct visualization o1" the mixin~ processes was achieved u~ing c(.muercially av~lilo ~lble Plexiglas m i x i n g vessels (P~lller,,~oll Kelley, E~tsl Stmudsbur~, PAr thin were 3 inche,~ in diameter, 6 incl~e~ lens ~tntl h~ld ~tn 84 '~ angle between ~hell~ (Fig, I), These Plexi~las ve~el~ were lilled inside a Ill inch diameter Plesio ~l~is cylinder, which wlt~ ~u~pended on top of two roller~ {lnd held in place li'om above hy a third, freely rotating, roller (Fig. 2). Red and blue 600 p.m glass ~ d s (Jaygo. Union. N J) were ~ e d in the direct visualization experiments. The beads had a density of 2.55 gtcm ~ with the lbllowing size distribution data (supplied by the manufacturer): dp. io,~.=490 l~m. dp, s~)~,~= 600 Ixm. d~. ~)~,= 7 ~ ia.m. The total loading for each experiment was 50% of the vessel volume. Typical ioadings are in the range 40--70%. Loading affects mixing performance (for example, excessive loading can lead to the development of unmixed regions). The effect of variable loading will be discussed in detail in a future publication. The vessels were loaded axially, with one color being loaded into each shell. This was done one color at a time. First a plunger was inserted into one end of the twinshell (Fig. 3(a)). A measured amount of red beads was added into the other end of the shell and the plunger depth was then adjusted slightly until the level of the beads was at the centerline of the vessel (Fig. 3(b) ). A measured amount of blue heads was then carefully added on top of the layer of red beads (Fig, 3(c) ) so that the red-blue interface between the heads was maintained along the centerline of the twin° shell. Finally the vessel was carefully turned upright (Fig. 3(dr ) and the ends of the twin-shell were closed with caps. A photograph of the initial condition tbr an experiment is shown in Fig, 4, The second type of experiment was designed to f,acilitate examination of the structure of the mixture throughout the entire volume of the powder bed, At the end of the mixing experiment, the structure of the mixture was preserved by inliltrating the voids b~tween p~,'ti¢les with ~l polymer ~olulion, which was allowed to cros~°link, yielding ~l ~olidified luon~)lith 1131, Thi~ monolith w~)~ ~ub~equentiy ~liced t() reveal Ihe internal ~tructure ot' th,~ mix|ure, 'i'he~e ~olidilic~)o li~m experinlent~ were carried oul in ~,u~lOllVill~|deiiltllnmllli1 Iwino,~hell,~ (American Alun~inunh Mounl~lin~ide, N J) th~l had similar dime,skins (except I~)r a OW eagle betwcci'~ Fig. 4. Inili;d condition shown for 600 I~), red and blue I~ds in the Plexiglas mixing ve,;~el. D i#r,m¢ et aLI P~wder Techm~/ogy 91 ¢1~7) 165-172 t~ -1, bmde~ tlmd OOA 0,$" I l~,,~llil tinfoil Fig. 6. ,~'~ngtlic diagrmn of the image analysis system, I I i lll"l~llli~llll ~)I I i i I I IY! I I ! I I I hf, 5, Schemalic diagrams of (a) Ihe infiltration~yslem and (b) the slicing pa,~¢mu ~ mth~~lidifica~ioncxp~in~.mt~, shells) and were loaded in the ~me manner as the Plexiglas ~lls. Red and green 66 ttm glass be~s ( Potters Industries. Par. sippany. NJ) were used in the solidification/slicing experiments~The beads had a density of 2.46=2.49 g/cm s with the following size distribution data (supplied by the manufaclurer): d~ ,,~- 30 tim, dr,~ ' 6 6 tim. dr,,n.~ ~ 103 tim. A ~ r the mixing run was completed, the twin=shell vessel wa~ c~fully removed from the mixing apparatus and wa~ then pla~edinto an infiltration apparatuswhere it wa~ held in a ~cure horizontal position. The infiltration apparatus, ~hown in Fig+ ~ (a), consisted of a fluid re~rvoir, a pump, and tubing connected to a no~.~te. The iniiltra~ion medium was a come mercially available mixture of SO alcohol 40, water, octyho crylamide, ~rylates ~nd butylaminomethacrylat¢copolymer (R~ve~, C h e ~ g h ~nd~, G~nwich., CT). The ~ i u m was ~ ~ slowly onto the mixture to avoid trapping air in the system. The no~zle wa~ placed at the centerli~ el ° the v e ~ l , touching the vessel wall, allowing the infiltration m~ium to seep gently into the powder b ~ , pushing the air out slowly t h ~ the open ends of the vessel. Repealed experiments using both horizontally l a y e ~ and vertically lay~d initial conditions ~owed that the inliltration p ~ e s s did not cause disturhences to the mixture, The embedded mixtures were allowed to dry t ~ 14 days. The solidified ~ructures were then removed from the y e s , Is and sliced using a handsaw. First the mixing ves,~l was sliced along the centeHi~ while the vc~el still contained the m[xo turn. The two shell.s ~ r e then cut along the top surft~c of the ~didified beads, Al~er b~iefly heating the ~ l l s , the intact sol~iik~d structures were det~hed from the shell walls and removed fl~cn the vessels, T ~ structures were then slkx,d in halt\inch intervals along the axis of re[alien (Fig, $(b)). ~ ' h ex~ment resulted in about 14 ~tions. ~ntitatlvc mixing data were obtain~ from the slices ~sing image analysis [ 131. Mixture s~tions were scquen- tinily recorded as digital 8-bit gray-scale in~lges, and analyzed using numerical algorithms. The image analysis equipment setup used in these ex~riments is shown in Fig. 6. A 6510 CCD monochrome camera (Cohu, San Diego, CA) with a 55 mm f12,8 telecentric video lens (Edmund Scientific. Barrington, NJ) was mounted vertically above the image. Sufficiently uniform illumination of the field of view was attained by a fiber optic ringlight (Vo~pi, Auburn, NY ). This light source was supplied by a 150 W halogen bulb hound in an lntralux 6000-I controller (Volpi, Auburn, NY). For mixtures of red and green glass beads, a sharp-cut filter (R-60. Newport, Irvine. CA) was used to attenuate the shorter wavelengths (690 nm or less) while transmitting the longer wavelengths. Such a filter maximized grayoscale cone trust of the red aM green components. The red component became the brightest, the green component the darkesL Bad~ slice was ~anned with the aid of a programmable xy table (Untdex Aerotech. Pittsburgh, PA) operated ~mo|ely by a coulpul~f+~ I~h¢ video signal was digitally displayed as an 8~ bit image (256 gray levels) on an RSoI?0 picture monil~r (Sot~y Trinitron m~!¢l No, PVM~I342C. Sony, Tokyo. Japan). The outpu! signal from the monitor was ~nt to an MV20 im~le processing board (Datacube, Danve~, MA) where the signal was converted t~m analog to digital, "~e image was also displayed as a gray-scale image on it Sun workstation (Sun, Mountain View. CA), where image pr~Co essin8 software (developed at the Center for Computer Aids for Industrial Productivity, Rutgers Unive~ity, Piscataway, NJ) handled the. video signal, data retrieval, and storage. During acquisition, a slice of the mixture was partitioned into ,~parate fields of view, ¢~h 5 mm by 6.7 mm. Fig. 7 (a) shows a sketch of a slice with the~ field subdivisions. I~ch field contained ~p~ximately I0 000 particles and was digo i t i ~ into 480 × $12 pixels, with each pixel pos~ssing a gray level on a scale from 0 to 255. During pr~:essing, each field of view is further subdivided into regularly spaced regions (hereafter called 'patches'). The !~al composition was measur~ for each of the~ small patches by computing the mean gray-level intensity of the pi~els in the patch. 3"he patens became the ~m~llest areas evaluated in the experiments. Therefore, the patch si~e determined |he scale of examination in the mixing analysis. Although averaging m m , p mmmmmmmmmmmm !immnmmmmmmmlm mmmmllmmm lmmm/mmmm iImll ilmm L,m m mmm m !m mmk m l mmmw Y mmmmmmr mmi,:)-- imm c) ~1 p] 100 90 80 70 60 5O 40 0 within patches 'blurted' some of the im~lgo detail, the largo number of patches used in the analysis ( I0~10 °*per sli¢~) gave a detailed characterization of the composition statistic~ of the entire mixture, A typical slice is shown in Fig, 7(b), and the reconstructed image using the moan gray-scale value of the patches is shown in Fig. 7 (c). The reconstructed image retained a largo amount of contrast and detail found in the actual mixture. These data and the raw pixel values of the field image were written in separate liles for postoprocessing and analysis. 3, Results and discussion 3. Io Direct visualizatim! Experim~)nts were conducted using the initial condition shown in Fig, 4. The effect of the rocking motion on mixing in a V-blender is shown in Fig. 8. The photographs were taken while the vessels were in motion to avoid perturh~tions of the granular flow Ih~n would occur if the vos~el stopped, Th~ fill level appears greater than 50% due to the dynamic antic of repose of the beads which is equal to 300 at 16)rpm, Fig, ~(a) displays the mixture after live minutes of pure rotation at 16 rpm. A comparison between Fig, 8(a) and the initial condio tion shows that only a minimal degree of mixing has occurred, Even after 50 minutes of mixing without rocking (Fig, 8(b) ), complete mixedness is not achieved, To put thi~ into proper context, it should be noted that common practice in industry is to mix particles t~r only a few minute~, Clearly, this indicates that mixtures produced in this manner may bc less homogeneous than desired, Recent work in fluid mechanics has shown that steady Ilows or time.periodic flows witll a single fluetuati~m th:o quency often create permanently unmixed region~. Liu ctal, [ 14,15] recently showed that the introduction of flow perturbations with multiple frequencies prevented the formation of such unmixed regions, generating significant mixing i~~i~ii~ i'~!ili iiiii!i!!iii!!ii, iilii'iiii ~li,!~i¸i!il¸~ii: ii!~i~ii !~ii '!'~!:i ~!¸i~i ~¸i,i:!' iii~~i ii ] r~i~ ~Ia ratio o( ~ i~--voluti~I~¢ i~¢ki~i~c~cI~~ d a ~ (a) ampllt~le o( ~ flY., (b) (c) D. Bnme~,£a~. , / Powdt+rTechnology91 ~~997,~165-/'~ enhancements. Wightman et al. [ 16 ! found that this approach enhanced mixing in granular flows as well. Using a cylindrical mixer, they found that rocking perturbations disrupted recirculation patterns, especially when the rotational motion of the vessel was perturbed by a rocking motion at a different frequency than the rate of rotation. As shown in Fig. 8(c), this method is also effective for enhancing mixing in Vblenders. The figure shows the state of the mixture after 5 minutes of mixing at 16 rpm, with a rocking rate of 5.09 rocking cycles per minute (corresponding to a ratio of revolutions per rocking cycle) and an amplitude of :t: 10°. By using a rocking frequency with an irrational ratio relative to the rotation frequency, the periodicity of the rotational motion was disrupted (in fact, the net result is an overall aperiodic flow). As a result, mixing was enormously enhanced, and the system became completely homogeneous in just 5 minutes. (A detail,cd study of the effects of rocking frequency and amplitude will be the subject of a future publication.) The interior mixing patterns were examined l'or a similar set of conditions as in Fig. 8 using the solidification/slicing technique described above. Results shown in Fig. 9 reveal the mixing structure along the axis ~f rotation, Fig. 9(a) shows an experiment carried out for a rotation rate of 16 rpm and a total mixing time of 10 minutes without rocking. As shown in the photograph, the composition of each slice along the axis of rotation varies greatly from one end of the structure to the other, However, note that the variability within each slice is very small. Each slice by itself looks fairly well mixed, except fi)r the larger slices near the initial interface I~etween the ~d and green beads (at 1he cenler of the mixing vessel) which show signilicant heterogeneity, Fig. 9(h) show~ ~lnolherexperiment perl'ollned under the S~lUeconditions, but with a total mixing time of 20 minules. 'Rlere i~ slill a dis~+ cernibl¢ i~mount of h~lerogeneily in the ~y~teul. Fig+ 9(e) shows an experiment caffied out adding rocking motion, The experiment ~:orresponds to a rotation rate of 16 rpm, the same total mixing time as in Fig, 9(a) ( 10 rain), a ratio of ~ revolutions per rocking cycle and a rocking ampli= rude of ± I0°~ The entire structure is now extremely well mixed, It is only upon careful inspection that one can determine which end of the mixer was initially red and which one was initially green, in this case the composition is essentially the same between slices as well as within each slice. 3. 2. Quantification of the mixing results A quantitative comparison of the mixing results was per= formed for pure rotation and rotation supplemented with rocking. Results were obtained Ibr a mixing time of 10 minutes at 16 ~m. Fig. 10(a) shows the axial composition profiles° which were obtained by plotting the average corn° position (wt.% red) of each slice versus the position of the slice. The most salient feature of the graphs is the large disparity in composition between the left and right halves of the twin-shell tbr the case of pure rotation. Each location in the 17t tO ) 0 0 0 O With Rocking 0 0 O Without Rocking O q~ d~ 60 SoI I 0 E E o t.) UI 0 O O 0 O 4o r~ D 0 0 0 0 0 0 ~0 2O 0 0 10 0 41 .I 0 I Axial Position (inches) With Rocking 0'07i + Witheri Rocking oo~ 00~ 00~ . 003 = 00~ 00+ 0 ~0 ~0 ~o 40 ~0 60 70 BO ~0 ~00 Composition (red bead wt~) Fig, I0, QB~iBIiltIIiv¢¢ol)ipari~oli~ ~f ~,p~rililt~fll~ with ~tild 9.ilhoiii Itt~klit~ for BiiMIIB'~ ftlli|led ill l t;} I1}|it fi)f ~ll |iti~il)~ li|itg Of I0 liltiitilg~ ~howtBg (=11 Ih¢ I|iClm ~li¢t~¢Oliipo~llion~ and (h) Iht~ov~rcdl ¢OlllpO,~il|on~ Ich hall' of the shell (originally red) contains at Ica~t 70f/, rod particles, All slices in the right halfol'the shell (originally green) contain between 20% and 40% red parti~le~, The data clearly suggest that each ~hell behavc~ a~ a ~eparate entity, with slow flow of particles across the central symmetry plane, which is the main barrier to mixing. On the other hand, whcl~ rocking is used, the system quickly becomes considerably more homogeneous, and the composition of each slice approaches 50% in just a few minutes. The experiment further demonstrates that axial flow perturbations greatly enhance mixing. Once particles cross the boundary ITem one ~hell Io the other, they become well mixed with the rest of the ~y~tcm relatively quickly The overall composition probability density functions for both mixtures are shown in Fig. 10(b). The normalized di~= tributions were computed by dividing the composition range into 100 intervals and counting the fraction of patches that have compositions in each interval. The distribution for the pure rotation case is bimodal, again demonstrating the sig° nificant amount ot' heterogeneity between the two shells. The left peak in the distribution corresponds to the predominantly green right shell, while the right peak corresponds to the t~ D. llnm¢ ¢~al. I Powder Tech,n~ogy 91 f 1997) 165-172 4. C o n d m i o m Fig+ I l+ I . H ~ comli~iow~used to delermi.e ~ , r i s t i ~ wiil~ d ~ and (b) belweea slices. mixing times (a) ~inantly red left shell. The case with rocking, on the o~her ~ d . has a nearly-Gaussian distribution typical of wellmixed sDtems. While the means of ~ mixtures are es~ntinily the same (52% red). the s ~ deviation of the ca~ with pure rotation is 23.9% compared with 6.9% for the case with rocking. Based on evidence from both the visualization and the ~lidification experiments, it is ~ a r e n t that once the beads cross the boundary between a given pair of adjacent slices, they become mixed within the slice relatively q~ickly, The data a l ~ suggest the. to a good approximation, one can ~ r f i M mixing in a conventional V-blender in terms ofthree characteristic mixing times (based on the three modes of mixing): it) within slices (convective mixing due to mtat i ~ ) , (it) between slices (convective mixing due to tumbling from the center to the outer part of each cylindrical ~bell ) and (iii) between ~ells (dispersive mixing across the cefltertifle between ~hells), Qualitative experiments were per= formed to determine the older of magnitude of the~ char= ~teri~ti¢ time~ by varying initial ¢ondition~ and oh~rvin$ the miking prt~e~, Quantitative data discussing the eft~¢t of varying initial ¢ondition~ will be re~oed in a tuture puhli= eatlofl, Vi.~aali~ation experiments in which the initial condi= lion i~ c o m p o ~ of horit,oNal laye~ of d i f t ~ n t colors (Fig, I l i a ) ) ~how that mixing within slices ~ c u ~ within ~0 revolutions, ~pending on the thickness of the layers. Visuali~atlon e x ~ r i ~ n t s with vertical laye~ o~lhogonal to the axis of rotation (Fig. I I(b) ) show that mixing between slices ~ u r s within I ~ revolutions (when tracer particles ~ t i o ~ d ~ the outside or one of the shells, the par° tides t ~ e f~.m 80 to I~ revolutions to ~ h the inner slice, heal' the t'enterline of the ves~l ), Mixing between the ~ells, however, takes at least I000 revolutions. Hence, it is the slow flow of ~ti¢les along the axis, especially near the centerline of the vessel, thai limits the mixing proe~,~s in a ctmventional V=blender, The eff~t of the ~ k i n g motion is to add a con+ vecti~ flow to the system in the axial d i c t i o n . The esults ~nted ~ d~ow that a V.blender s ~ h ~ the one ~ r i ~ , d in this ~ y , e o d o ~ with a m~hanism for gun. crating axial eonv~tive motion across the ~enterline of mixing ~ l ' s , will di~lay e n h a ~ mixing c h ~ t e r i s t i c s ~ ovexaU characteristic mixing time of ab~mt IIYO rev~d,uti~ms. The effects of flow perturbations on the mixing rate in a V-blen~r were examined. Rocking motion perturbs the rotational particle flow by adding a time-dependem convective flow in the axial direction. Mixing is greatly enhanced by such flow perturbations. On a laboratory scale, same size particles are mixed 2-10 times faster using a rocking Vblender than in a convemional V-blender. The approach for enhancing mixing described here can be applied directly to small-scale laboratory equipment. Size and weight concerns may prevent the application of rocking to large-scale equipmenL However, axial flow perturbationscan be i m ~ in a different manner to yield similar enhancetaunts. One method of accomplishing this is to use a rotating ribbon attached to an intensifier bar (which is typically pres. ent in large-scale equipment), Efficient mixing would be accomplished by repeatedly reversing the d i c t i o n of rotation of the ribbon, creating an axial flow of variable direction across the center boundary and enhancing mixing in the same manner as if rocking had been applied. Refermces Ill J~Gray. Dry solidsmixingequipment,Chem..EnR,Pn)gro, 33 t 1957) l.+lS+$, WetdeM~um. R.,C Ccn~n and DP., Miller, Mi~ing of solid~In a Iwio4hetl bl~dcr, C¢~m+A~, 70 (1%3) 39° Z V,Chowl~n~ ~,i~,Lift#, Mi~in~of p h ~ u | i ~ l ~lid~, Powder r¢¢t~+@.,~J (1970) ~_~'L in Ih~ 141 DS, C~n, T~W, H~.~lya ~ DW~ Fue~l~a., Blender~ r y Ioi N Hmltby, A ¢on~l~,~ of Ihe l~rtotn~tt~ of t~du~ul~i mdid.~ I?I J,C~ ~mpI+ ~¢IMKS Mushy, I~l~ttI~tl|~ in I~wd~r bl~ndi~ ~d uoble0dt~, 8..PiPPin,&~d,,63 ( 19~4~ 371 of dry blettdi~g lSl II~F,I+~Ada~ ~d A.,G B ~ . Art a ~ n l 191 F. L~ ae~dJA Hersey. Mixt~8 ~ffom~e of a V,blend~r, lnst. C.~m, i'~f. $.t~p. ~r. No. 65, ( 1981) SI/C ItOl C,F, tla~ood, K Woi~k,~ E L~el~ke~d C Sw~t~m. TI~ petq~ olf~~mtmtto~mt~r ford~ pow~, P ~ t e r T¢c~eeM. t t { 197~) ~++9., lltl J,C. Willim~ ~ M l-,Kh~, Themixiog mini~gr~8~lto,of p~k;ul~le ~olld~of diffetc-mpimt¢lesi~. 11,¢ fl, emietd l:M~I~¢er. ]69 (1973) 19. 1121 A. Kauhp,~, Mixt~ of ~l|d~, le~l FJ~g, Chem. Fand~,~., I (1~2) 104 ll,~IC W i g ~ , F.I..M~##~lo~md]..W|lder,Aq~t|lai~e imageanalysi~ melhod for ¢~er~flng m|xlu~s oil"gr~tp.ul~rm a t ~ . t),~ler 1 I++IM. t.~. F.L Mu~.to ~ R.L Pesktn~Qaaottfi~lio~ of mining to ~ o d k C~I¢ flows~C/e,e.o~$~,lid.tl'%wIals. 4 (6) (19~.4) 869~ ~93. l~.+l M..Lie, FJ Mu~;~to~d RL ~kln, '1~ uruetureof the stretching fieldi# eb.a~¢c~it:yflows.AICh~'$.. 40 (~,) (1994) 1273=1286. II(+IC W i g ~ , PR. M~t, FJ. Mu~;~.io,RJL Rtm~nand E.K.Glea~on. stl~-te~e of mixture of particles getteraledby thne.Mel~ndeni l~ws, P ~ d e r reHmoto. ~4 ( 1995~213-240.