Mul$scale design models for con$nuous agglomera$on processes
for delivery form manufacture Jim Litster
School of Chemical Engineering Department of Industrial and Physical Pharmacy
Purdue University
OUTLINE
• Overview of CSOPS con$nuous processing plaDorm
• Current status of design models for wet granula$on
• Engineering design models – Concepts and challenges – Case study: Coa$ng or layered granula$on in a mechanical mixer
• Con$nuous granula$on in a twin screw granulator
• Concluding remarks 2
C-SOPS: Broad strokes Center for Structured Organic Particulate Systems • Focus: pharmaceutical product and process
design • Participants: Rutgers (lead), Purdue, NJIT, Univ. of
Puerto Rico • Team: 40 faculty, 80 students and postdocs, 120
industrial mentors • 39 member companies (pharmaceuticals,
equipment, instrumentation, software, process control)
Materials Formation and Characterization A
Design and Scale up of Material Structuring Operations B
Structural Characterization and Modeling C
Integrated Systems Science D
The major conceptual components of product/process engineering
Scien$fic Thrusts
Main Technology Ini$a$ves • Con$nuous Manufacturing of tablets and capsules
– Faster development – Lower cost – Improved quality
• Thin films containing drug nanopar$cles – Poorly soluble drugs – Pediatric and elderly formula$ons – Adjustable dose (for personalized medicine)
• Microdosing-‐based manufacturing – Mul$drug therapies – Diagnos$cs – Personalized medicine – Point of need manufacturing
Test Bed 1:Con$nuous Manufacturing of Tablets
Goal: Fully automated, integrated & robust table4ng opera4on
Feeders
Blender
Tablet Press
Feeding
Blending
Milling
Roller compaction
Tableting
Blender
Feeders
Roller Compactor Mill
Con$nuous oral dosage form manufacture
7
Pharmaceu$cal Product Development
molecule single crystal agglomerates, par4cle domains
granules
powder
compact
Product performance factors • Chemical ingredients • Morphology • Bulk powder proper$es • Proper$es of blends • Processing “unit opera$ons” • Processing history • Dissolu$on dynamics • Bioavailability
Wet/dry granula4on
Milling Compression
Mul4scale System
OUTLINE
• Overview of CSOPS con$nuous processing plaDorm
• Current status of design models for wet granula$on
• Engineering design models – Concepts and challenges – Case study: Coa$ng or layered granula$on in a mechanical mixer
• Con$nuous granula$on in a twin screw granulator
• Concluding remarks 9
Wet Granula$on – Equipment and Rate Processes
Fluid Bed
Impeller
Spray Nozzle
Chopper
Mixer
Tumbling drum
Nuclea4on Consolida4on and Growth Breakage
10
Methodology
Change in Process or Formula4on
Change in Rate Processes
Change in evolu4on of structure
Final Granule Proper4es
Research: Learning Pathway
Required Structure
Control of Rate Processes
Selec4on of Suitable Process and Formula4on
End Goal: Product & Process Design
Hounslow, Agglomeration Symposium 2009
Regime Maps for Rate Processes Hapgood, Litster & Smith, AIChE J, 49, 350-361, 2003
12
Iveson et al., Powder Technol., 117, 83-87, 2001
Influence of primary par$cle proper$es
Influence of primary par$cle proper$es
Influence of primary par$cle proper$es
Wet spot Liquid film
a b c
Influence of primary par$cle proper$es
1
10
100
1000
1E-09 1E-08 1E-07 1E-06 0.00001 0.0001 0.001 0.01 0.1 1 10
Ca
Str*
45-63 micron ballotini 63-90 micron ballotini45-90 micron lactose Simon Iveson Spherical CopperSimon Iveson Irregular Copper Simon Iveson Dendritic Copper63-90 micron ballotini 45-63 micron ballotini fit45-90 micron lactose Spherical Copper fitIrregular Copper fit Dendritic Copper
θγ
σ
cos* ppkdStr =
θγ
εµ
cospdCa
=
Wet spot Liquid film
a b c
Regime Map Approach
17
Problems with this approach
• Large numbers of experiments at various scales required
• Rela$onships do not easily transfer across different types of equipment
• Very simplis$c approach to powder flow and granule mechanics
• Distribu$on of granule a_ributes is not predicted directly
• Impossible to scale keeping all granule a_ributes constant
An Engineering Design Approach
No good model for mechanical dispersion nucleation
Need general approaches to powder flow models (velocity and stress fields)
Little validation of multidimensional models in process equipment
Liquid distribution poorly represented in the PB models
Getting reliable particle scale information a challenge
Few models on how intra-granule structures are built
The “100g” tool kit for complete characterization of real materials
Few of these exist!
OUTLINE
• Overview of CSOPS con$nuous processing plaDorm
• Current status of design models for wet granula$on
• Engineering design models – Concepts and challenges – Case study: Coa$ng or layered granula$on in a mechanical mixer
• Con$nuous granula$on in a twin screw granulator
• Concluding remarks 21
Model System: coa$ng in a paddle mixer
Ben Freireich, Jianfeng Li, Carl Wassgren, Jim Litster
• Develop models to track distribu$on of inter-‐par$cle spray coa$ng over $me
• Aim to improve coa$ng uniformity by op$mizing process condi$ons and mixer design
RTD Features
0.001
0.01
0.1
1
10
100
0.0 0.2 0.4 0.6 0.8 1.0
Spray Single Visit Residence Time, t s (s)
Spra
y C
ompa
rtm
ent R
esid
ence
Ti
me
Dis
trib
utio
n, Es(t s
) (s-1 ) • Peak near zero shows
preference of low visit times (short cut)
• Single visit spray zone residence time is approximately exponential
tS
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
0.0 2.0 4.0 6.0 8.0 10.0Bed Single Visit Residence Time, t B (s)
Bed
Com
part
men
t Res
iden
ce
Tim
e D
istr
ibut
ion,
EB
( t B) (
s-1)
DEMCompartment Model
• Single visit bed zone residence time is more complex
• Interesting features
§ Peak at zero: short cut
§ Regularly spaced decaying peaks: recycle
§ Finite width peaks: dispersion
tB
Compartment Model
• Spray region – Well-‐mixed, TS – Shortcut, λS
• Bed region – Dispersion, 1/N, TB
– Shortcut, λB
– Recycle, R
QBQλ
( )1 SR Qλ−
Bed Region
Short cut Recycle
Spray Region
N compartments
SQλ
( ) ( ) ( ) 11 StT
S S SS
E t t eT
λ δ λ−
= + −
( ) ( ) ( )( )
11
1
1 1 111 1 1 !
B
Nn tnT
B B Bn B B
R tE t t eR R T Nn T
λ δ λ−−∞ −
=
⎡ ⎤⎛ ⎞⎛ ⎞ ⎛ ⎞⎢ ⎥= + − ⎜ ⎟⎜ ⎟ ⎜ ⎟+ + −⎝ ⎠ ⎝ ⎠⎢ ⎥⎝ ⎠⎣ ⎦∑
Ts
TB
PB Model
• Par$cle 2-‐D distribu$on n(va, vl, t) – va : seed volume; vl : coa$ng volume – Ini$al condi$on: log-‐normal distribu$on n0 = n(va, 0, 0)
• Growth rate, Gvl = "#↓% /")
– Rate of coa$ng volume increase per par$cle – A power func$on of par$cle volume
• r = 0, size independent growth • r = 2/3, surface area propor$onal growth • r = 1, volume propor$onal growth
25
( )l
rv a lG k v v= +
Growth size dependence
Effect of r with
• 2-‐D distribu$on n(va, vl) – Small par$cles gain less mass of coa$ng with increasing r
26
( )l
rv a lG k v v= +
(a) t = 0, lognormal distribution (b) t = 360 s, r = 0 (c) t = 360 s, r = 2/3 (d) t = 360 s, r = 1
Growth size dependence
• Coa$ng CoV evolu$on – r = 0, CoV ∝ t -1/2
– r = 2/3 or 1, CoV decreases slower – -‐1/2 power law also found in
other 1-‐D models , Mann’s equa$on as an example
• Coa$ng CoV increases with increasing seed mass CoV
27
1E-3 0.01 0.10.1
1
10
Coe
ffici
ent o
f Var
ianc
e of
Coa
ting
Mas
s, C
oV (-
)
Spray Ratio, vl / va (-)
r = 0 r = 2/3 r = 1
seed particle distribution CoV0 = 1
slope = -‐1/2 2 2
CoV , S CC
S Ctσ στ
τ µµ µ
⎡ ⎤⎛ ⎞ ⎛ ⎞⎢ ⎥= = +⎜ ⎟ ⎜ ⎟⎢ ⎥⎝ ⎠ ⎝ ⎠⎣ ⎦
Mann, U., 1983. I&EC Process Design and Development, 22(2), 288-‐292.
Growth size dependence
• Coa$ng CoV evolu$on – r = 0, CoV ∝ t -1/2
– r = 2/3 or 1, CoV decreases slower – -‐1/2 power law also found in
other 1-‐D models , Mann’s equa$on as an example
• Coa$ng CoV increases with increasing seed mass CoV
28
2 2
CoV , S CC
S Ctσ στ
τ µµ µ
⎡ ⎤⎛ ⎞ ⎛ ⎞⎢ ⎥= = +⎜ ⎟ ⎜ ⎟⎢ ⎥⎝ ⎠ ⎝ ⎠⎣ ⎦
Mann, U., 1983. I&EC Process Design and Development, 22(2), 288-‐292.
Model Development
29
Collision dynamics • collision energy, E • collision velocity, v • collision rate, β
DEM Simulation
Compartment model
PB Simulation
Par4cle Scale Process Scale
Flow characteris$cs • posi$on, x
• velocity, # • RTD, E(t) • solid frac$on, ν
Physical kernels
Flow in/out composition
Sub-‐model Scale
Flow characteris$cs • posi$on, x
• velocity, # • RTD, E(t) • solid frac$on, ν
Coalescence/breakage model
Possible compartments for HSWG
0 200 400 600 800 1000 1200 1400 1600 1800 20000
5
10
15
20
25
Impeller speed [mm/s]
% o
ccup
ancy
Velocity Distribution
Tran et al, WCPT5, 2006
OUTLINE
• Overview of CSOPS con$nuous processing plaDorm
• Current status of design models for wet granula$on
• Engineering design models – Concepts and challenges – Case study: Coa$ng or layered granula$on in a mechanical mixer
• Con$nuous granula$on in a twin screw granulator
• Concluding remarks 31
Con$nuous granula$on
• Be_er opportunity for “regime separated granula$on” – Independent tuning of different granula$on rate processes
• Be_er opportunity for in line measurement, regulatory control and real $me process op$misa$on
Twin Screw Granulator (TSG)
33
Eurolab 16 mm TSG
Liquid Inlet
Powder Inlet
Conveying Elements
Kneading Blocks
Twin Screw System Variables
34
• Powder Feed rate • Formula$on Composi$on
• Liquid Feed rate • Granula$ng Liquid Composi$on • Liquid Feed addi$on method
• Screw Speed • Shak Length • Screw Configura$on
Distribu4ve
Forward Feed screw
Reverse Feed screw
Conveying Elements Mixing Elements
0o Offset
90o Offset 60o
Dispersive
Kneading Blocks
Comb Mixing Elements
0
500
1000
1500
2000
2500
3000
0.1 0.2 0.3 0.4 0.5
d 50 (µ
m)
L/S Ratio
Raw Material A1ributes
Granule Proper$es and Growth Behavior
0
2
4
6
8
10
12
0.1 1 10 100 1000 10000
% V
olum
e
Particle size (µm)
Pharmatose Impalpable Supertab 30GR
Size (µm) 10 102 103 104 10 102 103 104 0
0.2 0.4 0.6 0.8 1
1.2 1.4 1.6 1.8
10 102 103 104
fmi (
lnx)
Increasing L/S Ra;o L/S = 0.15 L/S = 0.3 L/S = 0.45
0 0.2 0.4 0.6 0.8
1
10 102 103 104 10 102 103 104 10 102 103 104
Size (µm)
fmi (
lnx)
Binder Distribu;on Method
Dry Binder 1:1 Dry: Liquid Binder Liquid Binder
Pharmatose Impalpable Supertab
35
Mechanis$c Studies
36
Screw Element Characteriza$on
Granule A_ributes (size, shape, density,…etc)
Liquid Distribu$on
Residence Time Distribu$on
Popula$on Balance Model Development and Process
Op$miza$on
0
0.2
0.4
0.6
0.8
1
1 10 100 1000 10000
fmi (lnx)
Size (microns)
Granule Size Evolu$on Along the TSG
37
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1 10 100 1000 10000
fmi (lnx)
Size (microns)
0
0.1
0.2
0.3
0.4
0.5
1 10 100 1000 10000
fmi (lnx)
Size (microns)
0
0.1
0.2
0.3
0.4
0.5
1 10 100 1000 10000
fmi (lnx)
Size (microns)
Kneading Sec;on (Reverse)
90° 60°
30°
3KB 5KB 7KB
L/S Ra4o = 0.15 L/S Ra4o = 0.4
0
0.2
0.4
0.6
0.8
1
1 10 100 1000 10000
fmi (lnx)
Size (microns)
0
0.1
0.2
0.3
0.4
0.5
0.6
1 10 100 1000 10000 fm
i (lnx)
Size (microns)
Granula$on in Conveying Elements
0
0.1
0.2
0.3
0.4
0.5
0.6
10 100 1000 10000
Dye Co
nc (m
g/g sample)
Sieve Size (microns)
Conveying_1
Conveying_2
0
0.2
0.4
0.6
0.8
1
1 10 100 1000 10000
fmi (lnx)
Size (microns)
L/S=0.15
0
0.2
0.4
0.6
0.8
1
1 10 100 1000 10000
fmi (lnx)
Size (microns)
L/S=0.4
0
0.2
0.4
0.6
0.8
1
1 10 100 1000 10000
fmi (lnx)
Size (microns)
L/S=0.2
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45
10 100 1000 10000 Sieve Size (microns)
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45
10 100 1000 10000
Dye Co
nc (m
g/g sample)
Sieve Size (microns)
39
Screw Configura$on and LD Reverse Angle
90° Offset 30° Offset
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45
10 100 1000 10000 Sieve Size (microns)
3KB 5KB 7KB Conveying Nominal Dye Conc
60° Offset
3D-‐Granule Shape Characteriza$on
40
7KB30R
3KB90
3KB30R
7KB90
Conveying
Mechanis$c Studies
41
Screw Element Characteriza$on
Granule A_ributes (size, shape, density,…etc)
Liquid Distribu$on
Residence Time Distribu$on
Popula$on Balance Model Development and Process
Op$miza$on
0
200
400
600
800
1000
1200
1400
0 200 400 600 800Granu
le size parameter (m
icrons)
Time (sec)
d10d50d90
42
In-‐line Monitoring of Twin Screw Granulator
L/S ra4o: 0.15-‐0.35
New genera4on of integrated, intensified & intelligent crystalliza4on systems with dras4cally improved flexibility, predictability, stability & controllability.
• Seed addition • Cooling profile • Antisolvent • Growth/nucleation
modifiers
Manipulated inputs:
Development of the Crystallisa$on & Plant-‐wide Process Informa$cs Systems (CryPRINS & Plant-‐wide PRINS)
§ Integrated, intensified & reconfigurable plant § Batch versus continuous manufacturing
Nagy&Braatz, Handbook of Ind. Cryst., 2012; Nagy&Braatz, Annu. Rev. Chem. Biomol. Eng., 2012
Take home messages
• Primary par$cle proper$es are very important in determining downstream delivery form processes and product a_ributes
• Mul$scale and compartmental approaches have a lot of promise to develop predic$ve design models for granula$on processes
• Con$nuous granula$on offers significant poten$al improvement in both design and opera$on
• Integra$on of con$nuous manufacture through to delivery form is the future
Acknowledgments • Students & Postdocs
– Ben Freireich, Jianfeng Li, Arwa El Hagrasy
• Collaborators – Carl Wassgren, Jeff
Hennenkamp (GSK), Ma_ Burke (GSK), James Cartwright (GSK)
• Funding sources – NSF Goalie, ERC-‐CSOPS,
Procter & Gamble, GSK