1、附录:外文资料与中文翻译外文资料：ComparingmixingperformanceofuniaxialandbiaxialbinblendersAmitMehrotraandFernandoJ.MuzzioDepartmentofChemicalandBiochemicalEngineering,RutgersUniversity,Piscataway,NJ,08855,UnitedStatesReceived17February2009;revised30May2009;accepted14June2009.Availableonline27June2009.AbstractThedyn
2、amicsinvolvedinpowdermixingremainsatopicofinterestformanyresearchers;howeverthetheorystillremainsunderdeveloped.Mostofthemixersarestilldesignedandscaleduponempiricalbasis.Inmanyindustries,includingpharmaceutical,themajorityofblendingisperformedusingttumblingmixers,Tumblingmixersarehollowcontainerswh
3、icharepartiallyloadedwithmaterialsandrotatedforsomenumberofrevolutions.Somecommonexamplesincludehorizontaldrummixers,v-blenders,doubleconeblendersandbinblenders.Inallthesemixerswhilehomogenizationinthedirectionofrotationisfast,mediatedbyaconvectivemixingprocess,mixinginthehorizontal(axial)direction,
4、drivenbyadispersiveprocess,isoftenmuchslower.Inthispaper,weexperimentallyinvestigateanewtumblingmixerthatrotateswithrespecttotwoaxes:ahorizontalaxis(tumblingmotion),andacentralsymmetryaxis(spinningmotion).Adetailedstudyisconductedonmixingperformanceofpowdersandtheeffectofcriticalfundamentalparameter
5、sincludingblendergeometry,speed,filllevel,presenceofbaffles,loadingpattern,andaxisofrotation.InthisworkAcetaminophenisusedastheactivepharmaceuticalingredientandtheformulationcontainscommonlyusedexcipientssuchasAvicelandLactose.Themixingefficiencyischaracterizedbyextractingsamplesafterpre-determinedn
6、umberofrevolutions,andanalyzingthemusingNearInfraredSpectroscopytodeterminecompositionaldistribution.Resultsshowtheimportanceofprocessvariablesincludingtheaxisofrotationonhomogeneityofpowderblends.GraphicalabstractThedynamicsinvolvedinpowdermixingremainsatopicofinterestformanyresearchers;howeverthet
7、heorystillremainsunderdeveloped.Mostofthemixersarestilldesignedandscaleduponempiricalbasis.Inmanyindustries,includingpharmaceutical,themajorityofblendingisperformedusing4tumblingmixers,Inallthesemixerswhilehomogenizationinthedirectionofrotationisfast,mediatedbyaconvectivemixingprocess,mixinginthehor
8、izontal(axial)direction,drivenbyadispersiveprocess,isoftenmuchslower.Inthispaper,weexperimentallyinvestigateanewtumblingmixerthatrotateswithrespecttotwoaxes:ahorizontalaxis(tumblingmotion),andacentralsymmetryaxis(spinningmotion).KeywordsjPowdermixing;Cohesion;Blender;Mixer;Relativestandarddeviation;
9、NIR;AcetaminophenArticleOutline1.Introduction2.Materialsandmethods2.1. Nearinfraredspectroscopy2.2. Binblendersusedinthisstudy:uni-axialblender(Blender1),bi-axialblender(Blender2)2.3. Experimentalmethod3.Results4.ConclusionReferences1. IntroductionParticleblendingisarequiredstepinavarietyofapplicati
10、onsspanningtheceramic,food,glass,metallurgical,polymers,andpharmaceuticalsindustries.Despitethelonghistoryofdrysolidsmixing(orperhapsbecauseofit),comparativelylittleisknownofthemechanismsinvolved1,and3.Acommontypeofbatchindustrialmixeristhetumblingblender,wheregrainsflowbyacombinationofgravityandves
11、selrotation.Althoughthetumblingblenderisaverycommondevice,mixingandsegregationmechanismsinthesedevicesarenotfullyunderstoodandthedesignofblendingequipmentislargelybasedonempiricalmethods.Tumblersarethemostcommonbatchmixersinindustry,andalsofinduseinmyriadofapplicationasdryers,kilns,coaters,millsandg
12、ranulators4,5,6,7and8.Whilefree-flowingmaterialsinrotatingdrumshavebeenextensivelystudied9and110,cohesivegranularflowsinthesesystemsarestillnotcompletelyunderstood.Littleisknownabouttheeffectoffundamentalparameterssuchasblendergeometry,speed,filllevel,presenceofbaffles,loadingpatternandaxisofrotatio
13、nonmixingperformanceofcohesivepowdersorthescalingrequirementsofthedevices.However,conventionaltumblers,rotatingaroundahorizontalaxis,allshareanimportantcharacteristic:whilehomogenizationinthedirectionofrotationisfast,mediatedbyaconvectivemixingprocess,mixinginthehorizontal(axial)direction,drivenbyad
14、ispersiveprocess,isoftenmuchslower.Inthispaper,weexperimentallyinvestigateanewtumblingmixerthatrotateswithrespecttotwoaxes:ahorizontalaxis(tumblingmotion),andacentralsymmetryaxis(spinningmotion).Weexaminetheeffectsoffilllevel,mixingtime,loadingpatternandaxisofrotationonthemixingperformanceofafree-fl
15、owingmatrixofFastFlolactoseandAvicel102,containingamoderatelycohesiveAPI,micronizedAcetaminophen.WeuseextensivesamplingtocharacterizemixingbytrackingtheevolutionofAcetaminophenhomogeneityusingaNearInfraredspectroscopydetectionmethod.AftermaterialsandmethodsaredescribedinSection2,resultsarepresentedi
16、nSection3,followedbyconclusionsandrecommendations,whicharepresentedinSection4.2. MaterialsandmethodsThematerialsusedinthestudyarelistedinTable1,alongwiththeirsizeandmorphology.Acetaminophenisblendedwithcommonlyusedexcipientsandisusedasatracertoevaluatethedegreeofhomogeneityachievedasafunctionofnumbe
17、rofrevolutions.Acetaminophenisoneofthedrugsmostwidelyusedinmixingstudies,andAvicelandLactosearecommonlyusedpharmaceuticalexcipients.IntheinterestofbrevitytheirSEMimagesarenotincludedinthispaper,butcanbefoundinHandbookofPharmaceuticalexcipients,2.1. NearinfraredspectroscopyAcetaminophenhomogeneitywas
18、quantifiedusingnearinfraredspectroscopy.Acalibrationcurvewasconstructedforapowdermixturecontaining(inaverage)35%avicelPH102,62%lactoseand3%acetaminophen.Nearinfrared(NlR)spectroscopycanbeausefultooltocharacterizeacetaminophen.SamplesarepreparedbykeepingtheratioofAviceltolactoserandomizedinordertomin
19、imizeeffectsofimperfectblendingofexcipientsduringtheactualexperimentsontheaccuracyoftheresults.TheRapidContentAnalyzerinstrumentmanufacturedbyFOSSNIRSystems(SilverSpring,MD)andVisionsoftware(version2.1)isusedfortheanalysis.Thesamplesarepreparedbyweighing1gofmixtureintoseparateopticalscintillationvia
20、ls;(KimbleGlassInc.Vineland,NJ)usingabalancewithanaccuracyof0.01mg.Near-IRspectraarecollectedbyscanningintherange1116-2482nminthereflectancemode.Partialleastsquare(PLS)regressionisusedincalibrationmodeldevelopmentusingthesecondderivativemathematicalpretreatmenttominimizethePartiClesizeeffects.Asshow
21、ninFig.1,excellentagreementisachievedbetweenthecalibratedandpredictedvalues.Fig.1.NearInfrared(NIR)spectroscopyvalidationcurve.Theequationusedtopredictacetaminophenconcentrationisvalidatedbytestingsampleswithknownamountsofacetaminophenconcentration.Theyaxisrepresentstheconcentrationcalculatedfromthe
22、equationandthexaxisrepresentstheactualconcentration.Thusaperfectlystraightlineat45owouldrepresentthebestcalibrationmodel.Eachpointonthegraphrepresentsasinglesample.Theconcentrationofacetaminophenexaminedhererangesfrom0to8%.2.2. Binblendersusedinthisstudy:uni-axialblender(Blender1),bi-axialblender(Bl
23、ender2)Duetoitswidespreaduse,acylindricalblender1withacapacityof30Lischosenasareferenceblenderinthestudy.AsshowninFig.2,thisblenderhasacircularcrosssectionandtapersatthebottom.Itcanbeusedwithorwithoutbaffles,whicharemountedonaremovablelid.Inthisstudyalltheexperimentsareconductedwithouttheuseofbaffle
24、s.Mixingperformanceinthisdeviceisusedtoprovideabase-lineforevaluatingthemixingperformanceofanewlydevelopedblender2withacapacityof40L,whichisalsocylindrical,inordertodeterminetheeffectofdualaxisofrotationonmixingperformance.TheblendershowninFig.2(b)hastwoaxisofrotation.Thespinningrateofprecessionrela
25、tivetothecentralaxisofsymmetryisgearedtobehalfofthatoftherateofrotationaroundthehorizontalaxis.Fig. 2. Pictorial representation of (a) bin blender 1 and (b) bin blender 2 showing the corresponding axis of rotation.Fig. 2.2.3. ExperimentalmethodTwotypesofinitialpowderloadingusedintheexperiments:top-b
26、ottomloadingandside-loading,whichareschematicallyrepresentedinFig.3.Toavoidagglomeration,theAPI,acetaminophen,wasdelumpedpriortoloadingitintotheblenderbypassingitthrougha35meshscreen.Inordertocharacterizemixingperformance,agroovesamplerwasusedtoextractsamplesfromtheblendersat7.5,15,30,60,120revoluti
27、ons.Thethiefwascarefullyinsertedinthebin,andacorewasextractedateachpointofinsertion(eachstab)minimizingperturbationtothepowderbedremainingintheblender.Approximately7samplesaretakenfromeachthiefstab,andatotaloffivestabsareusedateachsamplingtime,asshowninFig.4soatotalof35samplesaretakenateachsamplingp
28、oint.Fig. 3.Fig. 3. Schematicoftheloadingpatternusedinthestudy.Intop-bottomloading,AvicelisloadedfirstintotheblenderfollowedbyLactoseontopofitandfinallyAcetaminophenisuniformlysievedover.Inside-sideloadingavicelisplacedatthebottomandthenAcetaminophenisonlysievedonlyinhalfpartoftheblenderandissandwic
29、hedbetweenlactoseandAvicel.Fig. 4.Fig.4.(a)Thiefsampler(b)topviewofthesamplingpositionscheme.Theexperimentalplanusedinthisstudyisasfollows: Filllevel:blender1-60% ,Filllevel:blender2-60%,70%,80% 1.oadingpattern:blender1side-sideloading,top-bottomloading 1.oadingpattern:blender2side-sideloading,top-b
30、ottomloading Speed:blender1-15rpm,20rpm,25rpm Speed:blender2rotational/spinning:15/7.5rpm,20/10rpm,30/15rpm Samplingtime:blender1,blender2-7.5,15,30,60,120revolutions3. ResultsThehomogeneityindexusedistheRSD,whereCistheconcentrationofeachindividualsample,C二istheaverageconcentrationofallsamplesandnis
31、thetotalnumberofsamplesobtainedatagivensamplingtime.Weexaminetheeffectoffilllevelonmixingperformance.PreviouslytherehavebeenstudiesontheeffectoffilllevelintheBohlebinblender,GallaybinblenderandV-blenderanddoubleconeblender111,12and113.Alltheaforementionedblendershaveonlyoneaxisofrotation,thereforeth
32、eobjectiveofthisstudyistoexaminehowdualaxisimpactmixingperformancesathighfilllevels.Toavoidrepetition,studiesforfilllevelarenotconductedforbinblender1.ResultsavailablefromapreviousstudyusingMgStasatracershowedthatmixinginauni-axialblendersloweddownquitedramaticallyasthefilllevelexceeded70%.Moreover,
33、resultsforacetaminophencanbeassumedtobesimilartothoseobtainedinpreviousworkbyMuzzioetal.Illand13,forasingleaxisrectangularbinblender11J,whichhaveshownthatevenafterfewhundredrevolutionshomogeneityachievedwitha80%filllevelisverypoorascomparedto60%filllevel.Toexaminetheeffectoffilllevelinadualaxisblend
34、er,experimentswereperformedinblender2withthetop-bottomloadingpatternforarotationalspeedof15rpmandwithspinningspeedof7.5rpm.Thefilllevelsexaminedare60%,70%and80%respectivelyandsamplesaretakenafter7.5,15,30,60,120revolutions.TypicalresultsareshowninFig.5,whichshowstheRSDvs.numberofblenderrevolutions.A
35、sexpectedfornon-agglomeratingmaterials,thecurvesshowarapidlydecayingregion.Theslopeofthecurvesinthisregion,insemi-logarithmiccoordinates,isusedtodefinethemixingrate.Thecurvesthenlevelofftoaplateauthatindicatesthemaximumdegreeofhomogeneitythatisachievableintheblenderforagivematerial.Fig.5.Mixingcurve
36、sfordifferentfilllevelsinblender2.TheRSDofacetaminophenisplottedasafunctionofnumberofrevolutions.Theloadingpatternintop-bottomandtheblenderrotationalspeedis15rpmwithspinningspeedof7.5rpm.SimilartopreviousstudieswithothertumblingblendersWeobservethatblendingperformanceisadverselyaffectedbyincreasingf
37、illlevels.AsshowninFig.5,thecurvefor80%fillperformsmorepoorlythanthosefor60%and70%fill;asfilllevelincreases,RSDcurvesdecaymoreslowly,signifyingaslowermixingprocess.However,theeffectisnotaspronouncedasinotherbinblendersandafteraboutonly100revolutions,thesameplateau(thesameasymptoticblendhomogeneity)i
38、sachievedforallthreefilllevels.Next,theeffectofrotationalspeedisinvestigatedintheblender1withoneaxisofrotationandiscomparedtotheblender2withdualrotationaxis.Experimentswereconductedforbothblenderswithtop-bottomandside-sideloading.Experimentswereperformedat60%filllevelandtherotationspeedsconsideredfo
39、rblender1are15rpm,20rpmand25rpmrespectively.AsshowninFig.6andFig.7,whenplottedasafunctionofblenderrevolutions,thereisnotmuchofaneffectofrotationspeedonthehomogeneityindex(RSD)ofacetaminophenat60%filllevel.Itisobservedthatmixingperformanceat20rpmand25rpmisslightlybetterthanat15rpm,howeverthedifferenc
40、esintheperformanceoftheblenderunderdifferentspeedsareprobablytoosmalltobesignificant.RSDcurvesdecaywiththesameslope,indicatingsimilarmixingrates.Inthestudyreportedhere,thefilllevelisonly60%,andalltherotationalspeedsareenoughtoachievehomogenization.Theaforementionedstudieswereconductedat85%filllevel.
41、Forsuchahighfilllevel,atlowspeeds,astagnantcoreisknowntooccuratthecenterofmanyblenders,requiringhighershearstressperunitvolumetoachievehomogenization.Moreover,theflowpropertiesofMgStareknowntobestronglydifferentthanthoseofmostmaterials,andareknowntohaveadeepimpactontheflowpropertiesofthemixtureasawh
42、ole.Furthermore,MgStisfamouslyknowntobeashearsensitivematerial.Thusanexpectationthatlubricatedandunlubricatedblendswouldshowsimilarbehaviorwithrespecttoshearisprobablyunwarranted.tofhoNModln9 O%BFig. 6. Mixing curves for top-bottom loading experiments with 60% fill level. RSD is plotted as a functio
43、n of number of revolutions. Dotted lines correspond to experiments in the blender 1, while solid lines represent data points from the blender 2.Fig. 6.Fig.7.curvesforside-sideloadingexperimentswith60%filllevel.RSDisplottedasafunctionofnumberofrevolutions.Dottedlinescorrespondtoexperimentsinthewhiles
44、olidlinesrepresentdatapointsfromthe2.Subsequently,experimentswereperformedusingtheblender2atthreerotationspeeds:15rpm,20rpmand30rpm,andasexplainedbefore,thecorrespondingspinningspeedswere7.5rpm,10rpmand15rpm.Filllevelconsideredforbothside-sideandtop-bottomloadingwas60%.Again,itwasobservedthatvarying
45、rotationandspinningspeedsdidnotmakemuchdifferenceinmixingrate.AsshowninFig.6andFig.7,mixingcurvesforblender2varyonlyslightlywithrotationspeed.Forthetop-bottomloadingpatternitappearsthatmixingimprovesslightlywhenrotationspeedisincreased(theplateauisslightlylowerforhigherrotationspeeds,indicatinganimp
46、rovementinthelevelsofasymptotichomogeneity),butnosignificantchangeswithspeedareobservedinside-sideloadingpattern.Theblendingperformanceofbothblendersiscomparedatdifferentrotationspeedsforbothside-sideandtop-bottomloadingpatterns.Tomakeafaircomparison,thefilllevelwaskeptas60%forbothblenders,aconditio
47、nforwhichbothblendersachieveeffectivemixingatlongenoughtimes.Duetogeometricsimilarityofthetwoblenders,thiscomparisonhelpevaluatetheeffectofspin(rotationwithrespecttothecentralsymmetryaxis)onmixingperformance.AsshowninFig.6,themixingcurvesfortheblender2liebelowthosefortheblender1foreachrotationrate,i
48、ndicatingfastermixing.NotethatthefinalRSDasymptotereachedforbothblendersisalsodifferent,withtheblender2showingalowerasymptote(betterfinalmixedstate,presumablyduetoalessereffectoftheslowmixingmodeinthehorizontaldirection)thanblender1.Similarresultswereobtainedfortheside-sideloadingpattern,asdisplayed
49、inFig.7.TheRSDcurvesfortheblender1forallthethreerotationrateslieabovetheblender2.Itisthereforeconfirmedthatspinningablenderindirectionperpendiculartotherotationaxishelpsinenhancingmixturehomogeneity;however,forthematerialsexaminedhere,therotationratedoesnothavemucheffectonmixingperformance.Finally,acomparisonismadebetweenthetwoloadingpatternsforbothblenders.Again,toachieveafaircompar
