Comparing mixing performance of uniaxial and biaxial bin blenders Amit Mehrotra Fernando J Muzzio Department of Chemical and Biochemical Engineering Rutgers University Piscataway NJ 08855 United States a b s t r a c ta r t i c l ei n f o Article history Received 17 February 2009 Received in revised form 30 May 2009 Accepted 14 June 2009 Available online 27 June 2009 Keywords Powder mixing Cohesion Blender Mixer Relative standard deviation NIR Acetaminophen The dynamics involved in powder mixing remains a topic of interest for many researchers however the theory still remains underdeveloped Most of the mixers are still designed and scaled up on empirical basis In many industries including pharmaceutical the majority of blending is performed using tumbling mixers Tumbling mixers are hollow containers which are partially loaded with materials and rotated for some number of revolutions Some common examples include horizontal drum mixers v blenders double cone blenders and bin blenders In all these mixers while homogenization in the direction of rotation is fast mediated by a convective mixing process mixing in the horizontal axial direction driven by a dispersive process is often much slower In this paper we experimentally investigate a new tumbling mixer that rotates with respect to two axes a horizontal axis tumbling motion and a central symmetry axis spinning motion A detailed study is conducted on mixing performance of powders and the effect of critical fundamental parameters including blender geometry speed fi ll level presence of baffl es loading pattern and axis of rotation In this work Acetaminophen is used as the active pharmaceutical ingredient and the formulation contains commonly used excipients such as Avicel and Lactose The mixing effi ciency is characterized by extracting samples after pre determined number of revolutions and analyzing them using Near Infrared Spectroscopy to determine compositional distribution Results show the importance of process variables including the axis of rotation on homogeneity of powder blends 2009 Elsevier B V All rights reserved 1 Introduction Particle blending is a required step in a variety of applications spanning the ceramic food glass metallurgical polymers and pharmaceuticals industries Despite the long history of dry solids mixing or perhaps because of it comparatively little is known of the mechanisms involved 1 3 A common type of batch industrial mixer is the tumbling blender where grains fl ow bya combination of gravity and vessel rotation Although the tumbling blender is a very common device mixing and segregation mechanisms in these devices are not fully understood and the design of blending equipment is largely based on empirical methods Tumblers are the most common batch mixers in industry and also fi nd use in myriad of application as dryers kilns coaters mills and granulators 4 8 While free fl owing materials in rotating drums have been extensively studied 9 10 cohesive granular fl ows in these systems are still not completely understood Little is known about the effect of fundamental param eters such as blender geometry speed fi ll level presence of baffl es loading pattern and axis of rotation on mixing performance of cohesive powders or the scaling requirements of the devices However conventional tumblers rotating around a horizontal axis all share an important characteristic while homogenization in the direction of rotation is fast mediated by a convective mixing process mixing in the horizontal axial direction driven by a dispersive process is often much slower In this paper we experimentally investigate a new tumbling mixer that rotates with respect to two axes a horizontal axis tumbling motion and a central symmetryaxis spinning motion We examine the effects of fi ll level mixing time loading pattern and axis of rotation on the mixing performance of a free fl owing matrix of Fast Flo lactose and Avicel 102 containing a moderately cohesive API micronized Acetaminophen We use extensive sampling to character ize mixing by tracking the evolution of Acetaminophen homogeneity using a Near Infrared spectroscopy detection method After materials and methods are described in Section 2 results are presented in Section 3 followed by conclusions and recommendations which are presented in Section 4 2 Materials and methods The materials used in the study are listed in Table 1 along with their size and morphology Acetaminophen is blended with com monly used excipients and is used as a tracer to evaluate the degree of homogeneity achieved as a function of number of revolutions Acetaminophen is one of the drugs most widely used in mixing studies and Avicel and Lactose are commonly used pharmaceutical Powder Technology 196 2009 1 7 Corresponding author Tel 1 732 445 3357 fax 1 732 445 2581 E mail address muzzio sol rutgers edu F J Muzzio 0032 5910 see front matter 2009 Elsevier B V All rights reserved doi 10 1016 j powtec 2009 06 008 Contents lists available at ScienceDirect Powder Technology journal homepage excipients In the interest of brevity their SEM images are not included in this paper but can be found in Handbook of Pharmaceutical excipients 2 1 Near infrared spectroscopy Acetaminophen homogeneity was quantifi ed using near infrared spectroscopy A calibration curve was constructed for a powder mixture containing in average 35 avicel PH 102 62 lactose and 3 acetaminophen Near infrared NIR spectroscopy can be a useful tool to characterize acetaminophen Samples are prepared by keeping the ratio of Avicel to lactose randomized in order to minimize effects of imperfect blending of excipients during the actual experiments on the accuracy of the results The Rapid Content Analyzer instrument manufactured by FOSS NIR Systems Silver Spring MD and Vision software version 2 1 is used for the analysis The samples are prepared by weighing 1 g of mixture into separate optical scintillation vials Kimble Glass Inc Vineland NJ using a balance with an accuracy of 0 01 mg Near IR spectra are collected by scanning in the range 1116 2482 nm in the refl ectance mode Partial least square PLS regression is used in calibration model development using the second derivative mathematical pretreatment to minimize the particle size effects As shown in Fig 1 excellent agreement is achieved between the calibrated and predicted values 2 2 Binblendersusedinthisstudy uni axialblender Blender1 bi axial blender Blender 2 Due to its widespread use a cylindrical blender 1 with a capacity of 30 L is chosen as a reference blender in the study As shown in Fig 2 this blender has a circular cross section and tapers at the bottom It can be used with or without baffl es which are mounted on a removable lid In this study all the experiments are conducted without the use of baffl es Mixing performance in this device is used to provide a base line for evaluating the mixing performance of a newly developed blender 2 with a capacity of 40 L which is also cylindrical in order to determine the effect of dual axis of rotation on mixing performance The blender shown in Fig 2 b has two axis of rotation The spinning rate of precession relative tothe central axis of symmetry is geared tobe halfof thatof therateofrotationaroundthe horizontal axis 2 3 Experimental method Two types of initial powder loading used in the experiments top bottom loading and side loading which are schematically repre sented in Fig 3 To avoid agglomeration the API acetaminophen was delumped prior to loading it into the blender by passing it through a 35mesh screen In order tocharacterize mixingperformance a groove sampler was used to extract samples from the blenders at 7 5 15 30 60 120 revolutions The thief was carefully inserted in the bin and a core was extracted at each point of insertion each stab minimizing perturbation to the powder bed remaining in the blender Approxi mately 7 samples are taken from each thief stab and a total of fi ve stabs are used at each sampling time as shown in Fig 4 so a total of 35 samples are taken at each sampling point The experimental plan used in this study is as follows Fill level blender 1 60 Fill level blender 2 60 70 80 Loading pattern blender 1 side side loading top bottom loading Loading pattern blender 2 side side loading top bottom loading Speed blender 1 15 rpm 20 rpm 25 rpm Speed blender 2 rotational spinning 15 7 5 rpm 20 10 rpm 30 15 rpm Sampling time blender 1 blender 2 7 5 15 30 60 120 revolutions 3 Results The homogeneity index used is the RSD where C is the concentration of each individual sample C is the average concentra tion of all samples and n is the total number of samples obtained at a given sampling time RSD S C where S ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffiffi ffi ffi ffi ffiffi ffiffi ffiffi ffiffi n P C2 P C 2 n n 1 s We examine the effect of fi ll level on mixing performance Previously there have been studies on the effect of fi ll level in the Bohle bin blender Gallay bin blender and V blender and double cone blender 11 13 All the aforementioned blenders have only one axis of rotation therefore the objective of this study is to examine how dual axis impact mixing performances at high fi ll levels To avoid repetition studies for fi ll level are not conducted for bin blender 1 Results available from a previous study using MgSt as a tracer showed that mixing in a uni axial blender slowed down quite dramatically as the fi ll level exceeded 70 Moreover results for acetaminophen can be assumed to be similar to those obtained in previous work by Muzzio et al 11 13 for a single axis rectangular bin blender 11 which have shown that even after few hundred revolutions homo geneity achieved with a 80 fi ll level is very poor as compared to 60 fi ll level To examine the effect of fi ll level in a dual axis blender experi ments were performed in blender 2 with the top bottom loading pattern for a rotational speed of 15 rpm and with spinning speed of 7 5rpm Thefi lllevelsexaminedare60 70 and80 respectivelyand samples are taken after 7 5 15 30 60 120 revolutions Typical results are shown in Fig 5 which shows the RSD vs number of blender revolutions As expected for non agglomerating materials the curves Table 1 Materials studied in this paper NameSize and morphologyVendor City State Fast Flo Lactose 100 sphericalForemost farms Newark NJ Avicel PH 102 Microcrystalline cellulose 90 needle likeFMC Rothschild WI Acetaminophen 40 needle likeMallinckrodt St Louis MO Fig 1 Near Infrared NIR spectroscopy validation curve The equation used to predict acetaminophen concentration is validated by testing samples with known amounts of acetaminophen concentration The y axis represents the concentration calculated from the equation and the x axis represents the actual concentration Thus a perfectly straight line at 45 would represent the best calibration model Each point on the graph represents a single sample The concentration of acetaminophen examined here ranges from 0 to 8 2A Mehrotra F J Muzzio Powder Technology 196 2009 1 7 Fig 2 Pictorial representation of a bin blender 1 and b bin blender 2 showing the corresponding axis of rotation Fig 3 Schematicoftheloadingpatternusedinthestudy Intop bottomloading Avicelisloadedfi rstintotheblenderfollowedbyLactoseontopofitandfi nallyAcetaminophenisuniformly sieved over In side side loading avicel is placed at the bottom and then Acetaminophen is only sieved only in half part of the blender and is sandwiched between lactose and Avicel Fig 4 a Thief sampler b top view of the sampling position scheme Fig 5 Mixing curves for different fi ll levels in blender 2 The RSD of acetaminophen is plotted as a function of number of revolutions The loading pattern in top bottom and the blender rotational speed is 15 rpm with spinning speed of 7 5 rpm 3A Mehrotra F J Muzzio Powder Technology 196 2009 1 7 show a rapidly decaying region The slope of the curves in this region in semi logarithmic coordinates is used to defi ne the mixing rate The curves then level off to a plateau that indicates the maximum degree of homogeneity that is achievable in the blender for a give material Similar to previous studies with other tumbling blenders we ob servethat blending performance is adverselyaffected byincreasing fi ll levels As shown in Fig 5 the curve for 80 fi ll performs more poorly than those for 60 and 70 fi ll as fi ll level increases RSD curves decay more slowly signifying a slower mixing process However the effect is not as pronounced as in other bin blenders and after about only 100 revolutions the same plateau the same asymptotic blend homo geneity is achieved for all three fi ll levels Next the effect of rotational speed is investigated in the blender 1 with one axis of rotation and is compared to the blender 2 with dual rotation axis Experiments were conducted for both blenders with top bottom and side side loading Experiments were performed at 60 fi ll level and the rotation speeds considered for blender 1 are 15 rpm 20 rpm and 25 rpm respectively As shown in Figs 6 and 7 whenplotted as a function of blender revolutions there is not much of an effect of rotation speed on the homogeneity index RSD of acetaminophen at 60 fi ll level It is observed that mixing perfor manceat 20 rpmand25 rpmis slightly better than at 15 rpm however the differences in the performance of the blender under different speeds are probably too small to be signifi cant RSD curves decay with the same slope indicating similar mixing rates In the study reported here the fi ll level is only 60 and all the rotational speeds are enough to achieve homogenization The aforementioned studies were con ducted at 85 fi ll level For such a high fi ll level at low speeds a stagnant core is known to occur at the center of many blenders requiring higher shear stress per unit volume to achieve homogeniza tion Moreover the fl ow properties of MgSt are known to be strongly different than those of most materials and are known to have a deep Fig 6 Mixing curves for top bottom loading experiments with 60 fi ll level RSD is plotted as a function of number of revolutions Dotted lines correspond to experiments in the blender 1 while solid lines represent data points from the blender 2 Fig 7 Mixingcurvesforside sideloadingexperimentswith60 fi lllevel RSDisplottedasafunctionofnumberofrevolutions Dottedlinescorrespondtoexperimentsintheblender1 while solid lines represent data points from the blender 2 4A Mehrotra F J Muzzio Powder Technology 196 2009 1 7 impact on the fl ow properties of the mixture as a whole Furthermore MgSt is famously known to be a shear sensitive material Thus an expectation that lubricated and unlubricated blends would show similar behavior with respect to shear is probably unwarranted Subsequently experiments were performed using the blender 2 at three rotation speeds 15 rpm 20 rpm and 30 rpm and as explained before the corresponding spinning speeds were 7 5 rpm 10 rpm and 15rpm Filllevelconsideredfor both side side andtop bottomloading was 60 Again it was observed that varying rotation and spinning speeds didnot makemuchdifferenceinmixingrate As showninFigs 6and7 mixing curves for blender 2 vary only slightly with rotation speed For the top bottom loading pattern it appears that mixing improves slightly when rotation speed is increased the plateau is slightly lower for higher rotation speeds indicating an improvement in the levels of asymptotic homogeneity but no signifi cant changes with speed are observed in side side loading pattern The blending performance of both blenders is compared at different rotation speeds for both side side and top bottom loading patterns To make a fair comparison the fi ll level was kept as 60 for both blenders a condition for which both blenders achieve effective mixing at long enough times Due to geometric similarity of the two blenders this comparison help evaluate the effect of spin rotation with respect to the central symmetryaxis on mixing performance As shown in Fig 6 the mixing curves for the blender 2 lie below those for the blender 1 for each rotation rate indicating faster mixing Note that the fi nal RSD asymptote reached for both blenders is also different with the blender 2 showing a lower asymptote better fi nal mixed state presumably due to a lesser effect of the slow mixing mode in the horizontal direction than blender 1 Similar results were obtained for the side side loading pattern as displayed in Fig 7 The RSD curves for the blender 1 for all the three rotation rates lie above the blender 2 It is therefore confi rmed that spinning a blender in direction perpendicular to the rotation axis helps in enhancing mixture homogeneity however for the materials examined here the rotation rate does not have much effect on mixing performance Finally a comparison is madebetweenthe two loading patterns for both blenders Again to achieve a fair comparison all experiments are performed at 15 rpm and 60 fi ll level As evident in Fig 8 in both Fig 8 Comparisonbetweenthemixingcurvesof theblender2andtheblender1fortop bottomandside sideloadingpattern Dottedlinescorrespondtoexperimentsintheblender1 while solid lines represent data points from the blender 2 Experiments are performed at 15 rpm with 60 fi ll level Fig 9 A typical mixing plot with RSD plotted against number of revolutions The two solid lines emphasize on the two distinctive mixing regimes 5A Mehrotra F J Muzzio Powder Technology 196 2009 1 7 blenders top bottom loading gives a more rapid decay of the RSD indicating faster homogenization as compared to side side loading pattern However for both loading modes blender 2 achieves faster homogenization As reported in previous studies all the RSD curves in this paper exhibit a common trend with respect to time characterized by an initial period of rapid homogenization due to convective mixing fol lowed by a period of much slower homogenization typically con
免责声明：
1. 《外文翻译--混合性能单轴构件式搅拌机 英文版》内容来源于互联网，版权归原著者或相关公司所有。
2. 若《86561825文库网》收录的文本内容侵犯了您的权益或隐私，请立即通知我们删除。
