Scale-up of high area filters for biological fluid microfiltration requires accounting for multiple factors to ensure reliable scaling. Key factors include variability in membrane and device properties, process conditions, and non-membrane pressure losses. High area filters have increased productivity but scaling is more complex. Proper device design and narrowing the performance range of small-scale devices improves scaling accuracy. Accounting for pleat density, height, and support permeability is important. High area filters scale linearly for plugging streams but non-linearly for streams where surface caking occurs. A scaling tool with identical pleat structure confirms expected performance.

1. Scale-up of High Area Filters for
Microfiltration of Biological Fluids
Points to consider for reliable scale-up
Sal Giglia, Manager, Filtration
Applications R&D
MilliporeSigma

2. Advantages of High Area Filters
• Compared to standard area filters, filters that
contain more effective filtration area have:
– Increased productivity per device
– Smaller filter footprint
– Improved filtration economics
• However, scaling from small discs to high area
density configurations can be more challenging
Filtration Area

3. Elements of Sterile Filter Scaling
A comprehensive understanding of all scaling elements is needed to take full
advantage of high area devices.
Scaling tool
hydrodynamics
Variability in scaling
tool and cartridge
properties
Variability in process
conditions
Membrane fouling
mechanisms
Non-membrane
pressure losses
Pleat hydrodynamics
1
23
45
6

4. Scalability Definition
• Operating Conditions
– Flow rate or pressure
– NFF or TFF
• Filtration endpoint
– Process time
– Maximum flux decay or membrane plugging
– Minimum flow rate
3000 X
Ratio of full size device to scaling tool of the quantity; fluid volume processed
per unit filtration area, at a specified set of operating conditions

5. Small-scale Device Design: OptiScale®-25 device
Inlet
Vent
Underdrain
Membrane
Porous support
3.5 cm2 Effective
Filtration Area (EFA)
Device design is critical for accurate determination of intrinsic membrane properties.
Outlet

6. Membrane/Device Performance
Variability Effect on Scalability
70 80 90 100 110 120 130
Performance Value
Frequency
• Scaling factor must account for
possible range of both small-
scale and large-scale
performance

7. Narrowing Performance Range of Small-scale
Devices Improves Scaling Reliability
70 80 90 100 110 120 130
Performance Value
Frequency
• OptiScale®-25 devices contain membrane that
represents the center of membrane property
distribution, resulting in minimized scaling
uncertainty*
• Reducing range of small-scale device to ±5% of
mean reduces scaling uncertainty by 30%
*Method for improved scaling of filters
US 8387256 B2

8. External Pressure Losses
• Scaling from discs to pleated devices should account for all membrane
external pressure losses.
• Non-membrane pressure losses include those from the housing inlet and
outlet ports, and any other plumbing between the measured pressure
source and fluid outlet.
)(** membranenontotalp PPLAQ 
Minimize!
l/min m2 l/min-psi psi psi

9. 0
1
2
3
4
5
6
7
0 20 40 60 80 100 120 140 160 180
Flow Rate (LPM)
ShellP(psi)
1-inch T-Line Connection
1-inch In-Line Connection
Cartridge Housing Pressure Loss
• Select the housing and
plumbing connections that
can accommodate the
process flow rate
10”0.2μmPES@10psid
10”0.1μmPES@10psid

10. Process Condition Variability
• Process variability allowance is highly application
dependent
• Typical values are:
- 10-20% for low fouling streams
- 30-60% for high fouling streams1
1. H. Lutz, “Rationally defined safety factors for filter sizing,” J. Membr. Sci., 341 (2009) 268-278.

11. Cartridge Design Factors Relevant to Scalability
• ΔP losses in the housing-sleeve
annulus and in the cartridge core
- Both of these are small in 10”
cartridges and have negligible
impact on scalability
• ΔP losses within the pleat structure
are potentially significant

12. Pleat Design Factors Related to Scalability
DenseSemi-DenseLoose
Upstream Porous Support
Membrane
Downstream Porous
Support
ΔP losses in the upstream and downstream
non-woven pleat supports depend on:
• Pleat density
• Pleat height
• Pleat geometry
• Porous support lateral
permeability

13. Pressure Drop in Substrate Lateral
Direction Reduces TMP Across Membrane
Membrane
Downstream porous support
Upstream porous support
Flow pattern in
a pleat

14. Pleat Resistance Equations
01
11
u
dx
dPk

02
22
u
dx
dPk

Darcy’s law
(upstream support)
Darcy’s law
(downstream support)
2
212
)]()([
b
xPxPL
dx
du p 

Mass balance
across dx
b1
b2
L
x
Sal Giglia and David Yavorsky, “Scaling from Discs to Pleated Devices,” PDA J Pharm Sci Technol
July/August 2007 61:314-323.

15. Effect of Pleat Height and Pleat Density on
Calculated Scaling Factor
• Long pleat lengths and high pleat density allow for
higher area within a device but can result in
reduced permeability scalability
• Effects are more pronounced for high permeability
membrane
• Tradeoff between high area density in a device
and efficient utilization of the contained
membrane area0.70
0.75
0.80
0.85
0.90
0.95
1.00
0 0.5 1 1.5 2
Pleat Height (cm)
ScalingFactor
Semi-
dense
Dense
0.2 μm PES

16. High Area Sterile Filters
10” Millipore
Express® SHC and
SHRp cartridges

17. Performance Comparison between Standard and
High Area Cartridges
Flow rate, clean
water
Total throughput, plugging
stream
For non-plugging streams, high area
offers minimal advantage. However,
for plugging streams high area has
double the throughput compared to
standard area.
Why the difference?
10” Millipore Express®
SHC cartridges

18. Series Resistance Model for Device Flow Rate
• As membrane plugs, membrane resistance
becomes dominant; support and housing
resistance is constant
• Disc test measures resistance of membrane only
• Therefore, as membrane plugs, throughput of
disc and device converge

19. Optimal Design of Pleated Filters
• Optimal filter design and selection
should consider the application
conditions:
- High density pleat patterns
may not be advantageous for
high flow, low plugging
applications
- No single filter will be optimally
designed for all applications
Optimal device area is dependent
on the application

20. Scalability of Standard Area Cartridges
0.2 and 0.1 μm PES filters
Semi-dense pleat pattern
Permeability Throughput Capacity
Near linear scalability for standard area cartridges

21. Scalability of High Area Cartridges
Cartridge throughput initially lags disc throughput, but as membrane
plugs throughput values tend to converge.
Constant pressure
filtration
Millipore Express®
SHC-HA 0.5/0.2 PES
membrane
1.5 g/l soy peptone
Throughput vs. Time Flux vs. Throughput

22. Scaling Factor Increases with Filtration Time
Constant pressure
filtration
Millipore Express®
SHC-HA 0.5/0.2 PES
Membrane
Good agreement between measured and predicted throughput

23. Scalability at Constant Flow Operation
Constant flow filtration
Millipore Express ® SHC-HA
0.5/0.2 PES Membrane
Flow resistances of disc and high area cartridge devices converge as
membrane plugs, resulting in near linear scalability at filtration endpoint.

24. Effect of Filtered Particle Size on Scalability
• Streams that primarily plug the internal structure of the membrane tend to scale
linearly since the membrane is the dominant resistance.
• For very large particle sizes that form a cake on the surface of the membrane,
available volume on the surface of the membrane is less in a dense pleated
structure than in a disc, which may result in non-linear scaling.
• For streams where caking is a predominant fouling mechanism, prefiltration is
recommended.
Small particles Large particles

25. Scalability with a Large Particle (>> pore size) Stream
0
0.002
0.004
0.006
0.008
0.01
0.012
0.01 0.1 1 10 100 1000
LogNormalizedVolumeFrequency
Particle Size (um)
1g/L EMD Soy in DMEM
0.3g/L Sigma Whey in PBS
0.1g/L Hy-Soy T in DMEM
Most Soy T particles > 10 µm
In dense pleat packs, very large particles may be inhibited from accessing the
membrane surface, resulting in diminished scalability.

26. Pleat Fouling Simulations
Support Material
Membrane
Disc
Semi-
dense
pleat
Dense
pleat

27. Scaling Tool Containing M-pleat Structure
• Scaling tool contains identical pleat
structure as pleated cartridge
• After initial screening with Optiscale®-25
devices to assess stream filterability, a
scaling tool containing the M-pleat structure
can be used to confirm expected cartridge
performance
• Pleated scaling tool scales linearly with
cartridge for all streams
10 cm2 Effective Filtration Area (EFA)

28. M-Pleat Cup Scaling Tool Scalability Data
Constant pressure
filtration of 2 g/L soy
peptone with SHC-HA
M-pleat cup scaling tool and 10-inch pleated cartridge show identical filtration behavior.

29. M-Pleat Cup Scaling Tool Scalability Data
Constant pressure
filtration of 0.3 g/L soy
T with SHC-HA
M-pleat cup scaling tool and 10-inch pleated cartridge show identical filtration behavior.

30. Scalability Factor Comparison
• High area devices scale linearly to discs in
plugging streams except where caking is a
predominant fouling mechanism. To
mitigate this effect, prefiltration is
recommended.
• Pleated scaling tool can be used to
confirm M-pleat performance.
OptiScale®-25 Device M-Pleat Cup Device

31. Summary of System Sizing Components
• Proper accounting of all factors related
to scaling will ensure proper system
sizing
- Eliminates risk of filtration area
shortfall
- But without undue excess area
- Allows for maximizing advantage
of high area devices
Theoretical
Minimum
Pleat effects
Process
Variability
Membrane
Variability
Plumbing/ fitting
losses
Roundup
1.0
~1.3-1.6
Minimum filtration area
based on filter trial data
Support resistance and stream
accessibility
Allowance for variability in fluid
characteristics and process conditions
Allowance for range of filter properties
Flow resistances external to cartridge
Integer number of devices

32. Summary
• Accurate scaling to pleated devices is possible with the use of good
scaling tools and proper accounting of the important factors that can
impact permeability and throughput performance.
• A theoretical understanding along with modeling of device flow
hydrodynamics and membrane fouling mechanisms facilitates reliable
sizing of filtration systems.
• High area pleated devices can be advantageous in terms of filtration
footprint and economics, but the plugging characteristics of the streams
should be considered.

33. Acknowledgements
Joe Hersey
Michael Lynch
Ryan Sylvia
Songhua Liu

34. Questions?