Charley Wu Chemical and Process Engineering University of Surrey, Guildford, UK 0 @charleywu @ C.Y.WU@surrey.ac.uk
Introduction Many products are manufacturing through compaction of dry powders, involving powder flow into a confined space. ·Pharmaceutical w; 1p1 q:1:,:1m11 ·Catalyst • Automotive ·Chemical , •Ceramic , ·Magnetic ·Food
~ UNIVERSITY OF Typical Manufacturing Process -i...=;;;... SURREY Upper punch .. I compacts I I Lower punch I .......................... Die Filling Compaction Ejection
Why Die Filling Is Important? Any problem during die filling will have a direct impact on the quality of the final products. Failure during die filling can lead to o Tablets of inaccurate dose! o Products with large weight variation o Products with non-uniform contents that detrimentally affect the functionality o Gears of uneven strength and with weakest links. o Distortion (and complete failure) during subsequent processes, such as sintering. "If your doctor prescribed half a tablet a day, which half would you want to take? "(Malvern Instruments} 2008).
Methodology (Exp. + Modelling) ' SURREY D A combined experimental and numerical approach was employed to understand the die filling process. D A model die filling system was developed. D Die filling behaviour was visualised using a high speed video system. D Quantitative analysis was also performed using o PEPT -> particle velocity o A pressure sensor -> time evolution of deposited mass. o An air pressure sensor -> air pressure build-up D Mechanistic analysis was performed using DEM-CFO
~+ J ~ _F~ J ~ ~ ~ L ~ . t I t L A typical expenmen a se -up SUNUIVERRRSITYEYOF .,J' Air pressure sensor J Iii e:.. 22001 <C .,J ·:ml ...................... . am Mass pressure sensor BO 75 - 70 ~ J I ] 30 32 34 38 38 Time (s) High speed video
PEPT Study ' SURREY (Positron Emission Particle Tracking)
~ L PEPT Study UNIVERSITY OF SllR.REY 30 trajectories of individual particles Spherical microcrystalline cellulose ( Celphere, CP102 ) Initial position of the shoe • .-5' ~~ ~ '• V'i \7 v "' v 7 'V w v" v ·· -. .. • '2> . • . o 0 • . () 0 • o 6 • Die 9 • 0 0 Wu, C.-Y., X.F. Fan, F. Motazedian, J.P.K. Seville, D.J. Parker, A.C.F. Cock s, (2010). A Quantitative Investigation of Powder Flow during Die Filling Using Positron Emission Particle Tracking (PEPT). Proc ee ding s of Process Me ch a ni ca l Engin ee ring. 224(3): 169-175. the In stitution of M ec hanic al En gin ee rs, Pa rt E, Journ al of
DEM-CFO Particle equations of motion: The flow of particles is modelled using DEM. The interaction between particles are rigorously modelled using theoretical contact mechanics: • Hertz-Mindlin-Deresiewicz for elastic particles • JKR for adhesive particles The interaction between air and particles is considered. The flow of air is modelled 8(Epf) ---+ using CFO. at fUU )= 8(Epfu) + '\7. at -Vpf +V·Tf Epfg
~ DEM-CFO Validation D Validation of DEM models is important. D Qualitatitive validation is easy, is it convincing? D Case-to-case quantitative validation is difficult. " . \ • + 8 • • • 0 0 0 • 0 0 • DEM-CFO Experimental Guo Y, Wu C-Y , Thornton C. (2013) Modeling Gas-Particle Two-Phase Flows with Complex and Moving Boundaries using DEM-CFO with an Immersed Boundary Method. A/CHE JOURNAL, 59 (4) , pp. 1075-1087
DEM-CFO with Non-spherical Particles ' SURREY D Multi-sphere -> approximate particle shapes using clumped spheres. D Utilize the rigorous contact laws for modelling particle-particle interaction
~ !(_ SUNUIVERRRSITYEYOF Die filling with real particles (Wu et al. 2016 ) Real crystal DEM approximation
Flow from a stationary feeder In Vacuum InAir 3 , dP =50 µm Ps =1500 kg/m
Flow from a stationary feeder In Vacuun1 InAir
A."'.I"'i'*°"~ I ~ 1~ .-,-.-r-.,-,-,-~-.-.-,-,-.-m.-.-.-r-.-.-.-,.-y-r-c-r-,-,-~-r-r-r-r-.-.-.~ 06~0 o.os~,~-~ ~ ~ I~ Flow from a stationary feeder Normalised 0. 7 0 mass flowrate a ............ ......... .. llf!' ..... .. ................... .A .................... . ... ............... . .. .... ... .. M 0.50 According to Beverloo 0.40 equation, M* is in the range of 0.55-0.65. Normalised 0.30- particle density <I> = Ps 0.20 Pa .A. In vacuum P ln air ]Group 1 • [n vacuum 0 1-n air }Group 2 e In vacuum Archimedes Number Q In air )Group3 T In vacuum } \J Jn air Group 4 0.10 - ) d3 ( ......... Constant in vacuum A = Pa Ps Pa gi p 0.09 - Power law in air 1]2 4 6 8 r 10 10 10 At<P p • This is in excellent agreement with Berveloo constants obtained experimentally (C is in the range of 0.55-0.65, for spherical particle c~0.58, see Seville et al. 1997). Guo Y., Kafui K.D., Wu C.Y., Thornton C. and Seville J.P.K. , (2009), A coupled DEM/CFO analysis of the effect of air on powder flow during die filling. A/CHE Journal, 55 (1 ): 49-62.
. ~ ~ 0.70~ 06~0 . . ' SURREY Flow from a stationary feeder ........ . .... ........... .. . .... . ...... .. .. ... ........ .. ... .. ... .. ............... ....... ........... . ~ .... ~ .... .. 0.50 0.40 Air inert 0.30 Air sensitive 0.20 • In v acuum !:;::. Jn air }Gro up I • In vacuum 0 In air }G rou p 2 e In vacuum 0 ln air }Gro up 3 T In vacuum } . Gro u p4 '7 1 v n atr .. .. . .. .. Constant in vac uum Power law in air - 0.08 3 10 Ar" </J p Guo Y., Kafui K.D., Wu C.Y., Thornton C. and Seville J.P.K. , (2009) , A coupled DEM/CFO analysis of the effect of air on powder flow during die filling. A/CHE Journal , 55 (1 ): 49-62.
Flow from a moving feeder (Fine sand, Vshoe=300 mm/s) (MCC, Vshoe=50 mm/s)
, Flow from a moving feeder In Air ,.-~ In Air ~-=- V=35 mm/s V=70 mm/s
~ ~ ' SURREY Flow from a moving feeder D There is a critical filling speed In v acuum • during die filling, above which 0 In air - Fitting cu rve in vac uum the die cannot be completely 0.8 - Fitting cur ve in air filled. ) 1 68 8 =(99 .33/v 0.6 shoe ..... / D The critical filling speed is a 0 ~ ;..., function of powder properties, .......... .......... 0.4 and process system I I parameters I /c ru v el \ i 0.2 11 9 8 = (4 7 .26/v ) shoe D For a given process system, the critical filling speed is 50 100 150 200 dominated by powder Shoe speed v 1 (nun/s) S lOe properties. This can also be used to assess powder flowability.
l\fr1~ Flow from a moving feeder . .\.ng.J.e of repose Good flowability 0 C11t1cal Ftllu1c Sp et-! d • nu11 ~ • 0 200 Poor flowability CP30:' Fl o\'\' Ftu1 ction 50 1.6 Flo\\· R~1te 1 g ~ 1 '-
oL-~-L-~ ' SURREY Flow from a moving feeder • MCC PH102 e MCC PH101 ... MCCDG 200 + Mannitol D Mix 1 ,,,-..... 8 Mix2 rfj Mix3 ] 150 - Eq.(9) ......_,. > (.) 100 ... . 50 • 0 10 20 30 40 Flow index rljJ (mm)
~ :~ ~ ~a A Mathematical Model for Die Filling • Experimental 400 40 • Experimental Model -- 30 300 ... - - Model •M> - l:ID 20 200 Mt ·- Mt lg 2 10 100 0 0 100 D 200 300 400 500 0 200 400 600 800 Fiming speed (mm/s) Filling speed (mm/s) (a) (b) • Experimental • Experimental 60 40 - Model • Model so - 3D - - Ga 40 illa ..,. ..., 2D 30 ftl ftl ::!: 20 ::!: • 10 10 0 0 D 50 l!.00 200 250 150 200 0 400 600 Flllng speed (mm/s) Filling speed (mm/s) (c) (d) The variation of the deposited mass with the filling speeds for a) Silibeads 300; b) Cenopheres 500; c) Mannitol and d) Alumina 4.
~ A Mathematical Model for Die Filling 300 S300 • COE Experimental 250 -- Full model e GBLSmm - 200 QI) -...... E 150 - u > I C500 100 Mann 1to9 so MCCOG 0 11 1 .. E+os 1 .. E+o8 1.E+o3 1.E+o4 1.E+06 1. E+07 1.E+o9 1.E+ lO (=Ar.<J>) The critical filling speed obtained in the closed die experiments as a function of~ <I>= PP Ar= PaPsgd; rr ') Pa
Conclusions D Powder into a confined space depends upon powder properties, die geometry and filling conditions. D The influence of air presence can be significant. D DEM-CFO is capable of capturing the major features during die filling. D Critical filling speed could be used to characterise powder flowability. D Based on air sensitivity classification obtained by Guo et al. (2010), a model was developed to predict the deposited mass and the critical filling speed.
Acknowledgements EPSRC IFPRI AstraZeneca Sanofi Pfizer Dr. Yu Guo Dr. Chunlei Pei Dr. Serena Schiano Mr. Joesry El Hebieshy Ms. Anastasiya Zakhvatayeva Dr. Colin Thornton Dr. Ling Zhang EPSRC Engineering and Physical Sciences Research Council
Recommend
More recommend