Solid-Liquid Separations
Pharmaceutical API Process
Development and Design
Solid-Liquid Separations
• Filtration Analysis
– Flow in packed beds
– Cake Filtration
– Centrifuges
– Deliquoring
– Washing
• Examples
– Cake compressibility
– Cycle time calculations
Filtration Options
Linear Filtration Centrifugal Filtration
Kevin Seibert (2006), Solid-Liquid Separations in the Pharmaceutical Industry
Filtration Options
Retentate Stream
Backpressure Applied
Feed Tank
Permeate Stream
Cross Flow Filtration
Kevin Seibert (2006), Solid-Liquid Separations in the Pharmaceutical Industry
Mechanisms in Filtration
• Depth filtration
– Particles captured within pore spaces
– Slurries with less than 0.1% solids
• Cake filtration
– Particles bridge pores in medium
– Cake formed on surface of medium
• Cross flow filtration
– Porous tube with cross flow
Driving Forces
• Gravity
– Hydrostatic pressure
– Free filtering materials
• Vacuum
– Downstream pressure below atmospheric
– Rotary drum, moving belt, disc filters
• Pressure
– Pumps or compressed gas
– Plate and frame, leaf
• Centrifugal Force
– Perforated bowl centrifuge, peeler centrifuge
Operating Mode
• Constant pressure filtration
– Vacuum pumps, compressed gas
• Constant rate filtration
– Positive displacement pumps
• Variable pressure, variable rate filtration
– Centrifugal pumps
Cake Filtration
suspension
L, ΔP
filter cake
membrane
Luis Puigjaner (2007), Solid-Liquid Separations
Flow Through Packed Beds
P
v Darcy’s Law
L k
3
k
k1 (1 ) 2 S o2 Permeability
P k1 1 S o2
2
Carman-Kozeny Equation
L 3
P Pressure drop L Bed height
v Superficial velocity Porosity
Liquor viscosity So Specific surface area
Mass Balance
mass of slurry mass of cake mass of filtrate
s V V
w c
1 ms A A
w s 1 L
A Filter area Filtrate density
V Filtrate volume s Solid density
w Dry solids/unit area L Thickness of cake
c Dry solids/unit volume filtrate Porosity
m Mass of wet cake/Mass of dry cake
s Mass fraction of solids in slurry
Cake filtration equation
Rewrite Darcy’s law in terms of specific cake
resistance, filtrate volume, solids concentration
Pc 1 dV Pc Cake pressure drop
v
w k s 1 A dt
P Total pressure drop
dV Pc A Pc A 1 Specific cake resistance
dt w c V k (1 ) s V Filtrate volume
A
w Dry solids/unit area
With medium resistance
c Dry solids/unit volume
dV PA PA filtrate
dt w Rm cV Rm Medium Resistance
Rm
A
: Characteristic parameter of a specific solid/liquid system
Ease of Separation
Ease of Separation Average Specific Cake Resistance (, m/kg
Very Easy 1x109
Easy 1x1010
Moderate 1x1011
Difficult 1x1012
Very Difficult 1x1013
W Leu (1986), Principles of Compressible Cake Filtration
Filtration Analysis
dV P A
Q
dt ave cV
Rm
A
Q = Flow Rate of Eluent
t = time of filtration
P = pressure drop Assumptions:
A = effective area of filtration Constant pressure
μ = viscosity of filtrate Constant area
ave = average specific cake resistance Ignore gravity
c = kg of dry cake per volume of filtrate
V = volume of filtrate
Rm = medium resistance
Parabolic Data Analysis
c 2
t ave V Rm V
2 A2 P AP
driving driving
Rearranging:
t c
ave V Rm
V 2 A2 P AP
driving driving
Plot t/V vs V – Linear
Slope – proportional to average specific cake resistance
Intercept – proportional to medium resistance
Cake Compressibility
Filtrate Flowrate
Incompressible
Highly
compressible
Pressure Drop
Incompressible solids - is independent of pressure
Cake Compressibility
ΔP3, α3
ΔP2, α2
ΔP1, α1
t/V
Compressible solids - varies with pressure
Cake Compressibility
ln
ln ΔP
o P
s
Where usually, 0.1 < s < 0.8
Sometimes expressed as:
s
P
o 1
Po
Where o, Po, and s are empirical constants
Medium Resistance
• Typically a linear contributor to overall
cake pressure drop
• May foul if size chosen inappropriately
Run 3
Run 2
Run 1
Increase in
t/V
medium
resistance due to
blinding
V
Experimental Method and Analysis
Laboratory Pressure Filtration Factory or Pilot Plant Filtration
Representative Slurry Sample slurry for laboratory
Volume vs Time Data Volume vs Time data
Cake Size and dry weight Cake size and dry weight
Three-Four Runs at various P’s Different pressures (if possible)
Various Medium Types Different medium types (if possible)
Parabolic Data Analysis Scale Up Calculations
Ave. Specific Cake Resistance Understand geometric
Medium Resistance considerations
Porosity (bulk density) Develop a working model
Liquor viscosity and density Understand equipment specific
Compute ave, Rm issues
Optimize operational strategy
Centrifugal Separations
Constant Pressure
Filtration Centrifugal Filtration
dV P dV Pdriving
Q Q
dt cV R dt cV R
ave m ave m
Ao Ao Ao Alm Am Ao
2h ro rc 2h ro rc
Ao cloth area Am Alm
2 r
ln o
rc
R3 Rc R1 =Ro
P3
Fluid
Pc
Cake + Fluid
P1
Filter Media
Po
Centrifugal Separations
Constant
Centrifugal Filtration
Pressure Filtration
dV Pdriving Pdriving
Q dV
dt cV R Q
ave m dt cV R
ave m
Ao Ao Ao
Alm Am Ao
fluid 2 R32 R12
Pdriving applied pressure Pdriving
2
Driving force and surface area are
2 functions of time, feed profile
c
t ave 2 V Rm V
2 A P AP
driving driving Filtration equation can be
integrated numerically
Cross-Flow Filtration
Backpressure Applied Retentate Stream
Concentrate a dilute two phase
(liquid solid) stream
Wash out a soluble impurity
(diafiltration)
Feed Tank
Permeate Stream Switch solvents for further processing
Scales very easily on filter surface area
Filtration Flux
t
k1t k 2
V
1 dV 1 k2
Permeate Flux
J
A dt A k1t k 2 2
J Filtration Flux
k1 Constant
k2 Constant
V Filtrate volume
Filtration Time A Filtration area
Periodic Operation
Backpressure Applied
Permeate Flux
Filtration Time
Cycle Time Analysis
• Cake formation
• Operation times that depend on cake
thickness
– Washing, deliquoring
• Operation times independent of cake
thickness
– Loading, cake discharge, cleaning
Deliquoring
• Application of vacuum
• Blowing with compressed gas
• Centrifugation
• Compression of the cake
• Complete drainage is not usually achieved
– Final drying with hot gas flow through cake is used
• Kinetics and equilibrium of deliquoring
– Threshold pressure: minimum pressure to achieve reduction in
saturation Pb
– Irreducible saturation: limiting value of saturation beyond which
no reduction in liquid content is possible S
Deliquoring Time
3 x 2 gL P Capillary k Cake permeability
Nc Number
1 2 L Liquid viscosity
S 0.155 1 0.031N 0.49
c Irreducible
Saturation
L Cake thickness
Porosity
4.61 Threshold
Pb Pa Gas pressure
x Pressure
kPb t x Mean particle size
Dimensionless
L(1 S ) Time
Surface tension
Pa Pa Dimensionless
Pa Pressure
Pb inlet Pb outlet Difference
S S
SR Reduced
1 S Saturation
Deliquoring Time
Reduced Saturation SR 1
Dimensionless pressure
difference
Pa
1
Dimensionless time θ
Washing
• Remove contaminants in retained liquor
• Methods
– Displacement washing
– Reslurrying followed by refiltering
• “Perfect” displacement washing
– Wash volume=void volume
– Solute concentration=initial concentration
• Actual washing
– Wash liquor tends to proceed through preferential pathways or
cracks in cake
– Concentration of solute in wash liquid depends on mixing and mass
transport
Displacement Washing
c c0 Perfect displacement
1
c/c0 washing
c c0
Actual washing
1
Wash Volume
(no. of void volumes)
Washing Curves
Saturated cake: displacement
followed by mixing and diffusion
1
c/c0
Drained cake:
No displacement stage 1
Wash Ratio
Washing curve for partially drained cakes will be
in between curves for saturated and drained cake
Washing Analysis
• “Perfectly Mixed” washing
kFt
c Concentration at end
c c e L of displacement
washing
F Wash flowrate/area
L Cake thickness
t Time from end of
displacement washing
ln c
Time
Washing Analysis
• Combined mixing and diffusion effects
• Dispersion parameter uL
D
cs uL u Wash velocity
f , , n
co D L Cake thickness
D Axial dispersion
n Wash ratio
Adsorption effects
uL
• Perfect mixing 0
D
cs
en
co
Washing Analysis
Washing curves as a
function of dispersion
1 parameter
uL
0.1 500
D
c/c0
1
Wash Ratio
Washing Time
Washing Time Cake formation time
Wash Ratio
n Wash ratio
t w nt f tw Washing time
tf Cake formation time
Examples
Filtration Analysis Example
5 PSI delta P
140
Weight of Filtrate (g)
120
100
80 25 PSI data
60
40 400
20 350
0 300
0 10 20 30 40 50 250
weight (g)
Time (s) 200
150
15 PSI filtration data 100
50
140
0
120
0 20 40 60 80 100 120
100
time (s)
80
60
40
20
0 Three pressures, same crystal
slurry
0 5 10 15 20
Ti m e ( s )
Kevin Seibert (2006), Solid-Liquid Separations in the Pharmaceutical Industry
Filtration Analysis Example
5 PSI delta P
Cake Filtration
140
Weight of Filtrate (g)
120
100
80
60
40 Cake Deliquoring
20
0
0 10 20 30 40 50
Time (s)
Start up Effects
Filtration Analysis Example
dt/dw vs w for 5 PSI dt/dw vs w for 15 PSI data
0.5 0.25
y = 0.0031x + 0.0473 y = 0.0014x + 0.0362
0.4 R2 = 0.8251 R2 = 0.9334
0.2
dt/dw (s/g)
dt/dw (s/g)
0.3 0.15
0.2 0.1
0.1 0.05
0
0
0 20 40 60 80 100 120 0 20 40 60 80 100 120
Weight (g) w eight (g)
dt/dw vs w for 25 psi data
dt c R
0.45
0.4 y = 0.0011x + 0.0468 ave 2 V m
0.35
R2 = 0.7117
dV PA PA
dt/dw (s/g)
dt c Rm
2 ave 2 V
0.3
d V PA PA
0.25
0.2
V weight of filtrate
0.15
100 150 200 250 300
w (g)
Filtration Analysis Example
dt/dw vs w for 5 PSI
1.5
y = 0.006x - 0.127
dt/dw (s/g)
1
R2 = 0.4515
0.5
0
0 50 100 150
Cake Deliquoring
-0.5
Weight (g)
Start up Effects
Filtration Analysis Example
dt/dw vs w for 5 PSI
0.5 y = 0.0031x + 0.0473
0.4 R2 = 0.8251
dt/dw (s/g)
0.3
0.2
0.1
0 Slope
0 20 40 60 80 100 120
Weight (g)
Intercept
Slope Intercept
dt c Rm
2 ave 2 V
d V PA PA
Filtration Analysis Example
Slope 0.0031 s/g2
Viscosity 8.94E-04 kg/m-s
c 61.12 kg/m3
A 0.002 m2 ave c
slope 2
PA 2
ΔP 34474 N/m2 (5 psi)
Density 1.0 g/cm3 Alpha = 0.782E+10
Filtration Analysis Compressibility
o P
ln
s
ln ln o s ln P
ln ΔP
P alpha ln(p) ln()
5 0.785E+10 1.609438 22.78378
15 1.06E+10 2.70805 23.08412
25 1.39E+10 3.218876 23.35515
Filtration Analysis Compressibility
ln ln o s ln P s 0.342
ln (alpha) vs ln (dp)
Slightly compressible
23.4 y = 0.342x + 22.215
23.3 R2 = 0.9686
Expect:
ln (alpha)
23.2
23.1 Some effect of
23
22.9
pressure on
22.8 filtration flux
22.7
1.5 2 2.5 3 3.5
Likely acceptable
filtration in
ln (dp) centrifuge
Series1
Filtration Analysis Scale Up
Filtrate volume as a function of time at
several pressures
Understand the relationship between specific
cake resistance and pressure (compressibility)
Fully characterized liquid / solid system
(physical properties etc.).
How do we scale up to understand
plant time cycles?
Filtration Analysis Scale Up
c 2
t ave V Rm V
2 A2 P AP
driving driving
Parameters: Rm, - known from scaled down experiments
All else known
Assuming: Same slurry composition, same filter medium, 25 psi
Filtration Time
Kg
Product 50 100 200 300 400 500 750 1000
2 m2 Filter 6m 25 m 1.6 h 3.7 h 6.5h 10 h 23 h 41 h
4 m2 Filter 1.5m 6m 25 m 55m 1.6 h 2.6 h 5.8 h 10 h
Cycle Time Analysis Example
Base Scale-Up Options
A (2x) B C D E F G
Batch Size (kg) 11.0 50.0 100.0 100.0 100.0 200.0 200.0 200.0
FD Area (m2) 0.2 0.6 0.6 2.0 4.0 0.6 2.0 4.0
Mother Liquor Volume (L) 86.0 390.9 781.8 781.8 781.8 1563.6 1563.6 1563.6
Mother Liquor Filtration Time (Hr) 3.3 5.0 10.0 3.0 1.5 20.0 6.0 3.0
Wash 1 Volume (L) 50.0 227.3 454.5 454.5 454.5 909.1 909.1 909.1
Wash 1 Filtration Time (hr) 7.3 11.0 22.0 6.6 3.3 43.9 13.2 6.6
Wash 2 Volume (L) 50.0 227.3 454.5 454.5 454.5 909.1 909.1 909.1
Wash 2 Filtration Time (hr) 20.0 30.3 60.6 18.2 9.1 121.2 36.4 18.2
Wash 3 Volume (L) 50.0 227.3 454.5 454.5 454.5 909.1 909.1 909.1
Wash 3 Filtration Time (hr) 5.7 8.6 17.2 5.1 2.6 34.3 10.3 5.1
Cake Depth (ratio to base) 1.5 3.0 0.9 0.5 6.1 1.8 0.9
Filtration Cycle Time 36.21 110 109.7 32.9 16.5 219.5 65.8 32.9
Milling Time per slurry mill (hr) 3.0 27.3 27.3 27.3 27.3 54.5 54.5 54.5
Total Time Cycle (hr) 39.2 137.0 137.0 60.2 43.7 274.0 120.4 87.5
Filtration and wash times for scale-up options based on constant flux
(L/M2H)
References
• W. Leu, Principles of Compressible Cake Filtration, in
Encyclopedia of Fluid Mechanics (N.P. Cheremisinoff,
ed), Gulf, 1986.
• A. Rushton, A. S. Ward, R. G. Holdich, Solid-Liquid
Filtration and Separation Technology, VCH, 1996.
• A. Rushton, Batch filtration of solid-liquid suspensions, in
Handbook of Batch Process Design (P.N. Sharatt, ed),
153-192, Springer, 1997.
