1 76 Solid/Liquid Separation: Scale-up of Industrial Equipment

Figure 4.30 Laboratory test apparatus

The appropriate MF membrane should be chosen for the application. In

the case of wash fluids, membranes with ratings in the 0.05 |im to
0.2 |Lim range were used. The membrane material should be compatible
with the wash fluid and temperature range. Long-term compatibility
tests may need to be carried out to verify this chemical stability.

4.3.2.3 Test procedure

The trial membrane should be flushed free of any preservative fluid
with filtered, deionised water.
The clean water permeate flow should then be measured at various
trans-membrane pressure values. The flux is then determined from a
measurement of the permeate flow rate and the effective area of the
upstream surface of the membrane. For the apparatus shown in Figures
4.18 and 4.30, the TMP is determined from the following expression:

PF+ PR
TMP= — -Pp
2
TMP is usually expressed in bar.

Flux vs TMP To generate flux v^ TMP data, the apparatus should be

filled with the sample and recirculated around the retentate side of the
membrane with the permeate ports closed. The sample should be pre-
filtered (<100 micron) to avoid superficial blockage or damage, as this
would mask the valuable and sensitive data regarding the properties of
the dynamic filter layer. The temperature should be controlled at the
operating temperature of the full-scale process. Record the starting
volume and any volumes added or bled later.
 4 • Membrane filters - microfiltration and ultrafiltration 1 77

An initial crossflow velocity of 1 m s"^ should be set (corresponding to

a typical average TMP of 0.2 to 0.5 bar), with all permeate and
concentrate being recycled back to the feed tank. The test equipment
should be maintained at a constant operating temperature.
Increase the TMP every 45 to 60 minutes, by 0.5 bar increments, and
record flux versus TMP data. Observe if the flux increases with TMP.
If it does not, this may indicate the formation of a gel polarisation layer
or pore fouling.

Flux vs Time Run at a constant (Ix) concentration by re-circulating

permeate to the feed tank for 60 minutes at the optimum TMP
determined above. Note any decrease in flux in 10 or 15 minute intervals
to generate flux vs time data.
A TFF membrane is regenerable by reverse filtration. It may be possible
to assess this during the initial laboratory evaluations. This being the
case, a reverse permeate backwash should be carried out at a pressure
>0.5 bar above Pp to assess any improvement in permeate flux.

Flux vs Concentration Concentrate the fluid in the feed tank to the

desired final concentration, usually a maximum of 40% oil, holding all
other conditions constant, and recording flux at various concentrations.
This is achieved by retaining the permeate in a separate container
rather than returning it to the feed tank.
Collect permeate and retentate samples at various concentrations to
measure permeate quality.
When the desired final concentration has been obtained, the test can be
progressed in continuous mode by setting permeate and concentrate
control valves to give the required VCF. Fresh feed is then continu-
ously added to the feed tank.
When testing is complete, drain the system, flush with clean water, and
clean the system with the appropriate chemicals until the clean water
flux is stabilised at the same conditions. The cleaning chemical used
for this application would typically be an alkaline based detergent,
increasing the temperature usually improves results. The stronger the
alkalinity, and so higher pH, of the cleaning fluid also usually has a
positive effect on cleaning. The recovered permeate flux after
laboratory trials and chemical cleaning should be close to the original
flux measured under the same operating conditions.
178 Solid/Liquid Separation: Scale-up of Industrial Equipment

4.3.2.4 Results and significance of laboratory scale trials

As part of the laboratory evaluation, the permeate quality should be
assessed for contaminant removal and detergent additives levels. Any
possible fluid / chemical problems should have been highlighted. The
permeate flux will be known at normal and concentrated levels
of contaminants from the sample measured in litres/unit area of
membrane per hour, at a known temperature. The temperature factor is
very important as a 1°C change can affect the permeate flux by 3%.
Figure 4.31 below provides a temperature correction factor (TCF). The
recovered permeate flux after cleaning with different cleaning
chemicals may also have been assessed.

0.5 1 1.5
Temperature Correction factor

Figure 4 . 3 1 Temperature correction factor for permeate flows. The graph

compares the temperature to 20°C to give the correction factor

However, many industrial applications tend to exhibit a variety of inlet

conditions. This may result from many factors outside of the
immediate control of the operators. Under these circumstances, the
preliminary laboratory test investigation will give an indication of
sizing for a filtration system unit to see if the application is viable.
These results can only be scaled up to give an indication of how a full
sized system would perform. The required site membrane area can be
only be estimated. As with certain biopharmaceutical applications it
may be prudent to undertake intermediate pilot scale trials.
For example operating a 0.05 m^ membrane over two days of
laboratory trials cannot be confidently scaled up to a 10 m^ membrane
area system that will operate for years. The scaling factor would be
200 fold on two days' work, which for certain industrial applications
would represent a significant commercial risk and site based pilot scale
to scale-up trials should be considered.
 4 • Membrane filters - microfiltration and ultrafiltration 1 79

4.3.2.5 Pilot trials

The purpose of the pilot trial is to further investigate the application
over a longer time period with a larger membrane area to more
accurately size a full MF tangential flow filtration system. To ensure
long-term trouble free operation of a system the pilot trial will help to
give accurate data that can be scaled up. Where scale-up is significant,
the laboratory tests can be seen as a feasibility study and site piloting
will then give accurate sizing data.
Further advantages of a site trial over the laboratory work are that the
trial will be carried out under the normal process conditions with
constant ingression of contaminants and any temperature variations
will be seen.
The equipment used to trial at site should ideally be in the form of a
fully automated test rig. Telemetry would be advantageous, as this
would enable the rig to be set up and monitored remotely over a period
of weeks. The use of manual laboratory equipment will often be too
time consuming and expensive to operate on site. For example, most
MF systems will typically require a reverse pulse backwash every 15
minutes for long-term permeate flux maintenance.
Pilot testing is necessary to prepare the system design and con-
figuration, to make propositions for system size and to evaluate
economic payback factors.
The membrane type, crossflow velocity and rating should have been
chosen from the laboratory bench test work. The trial rig system
operation should be the same as the full size system with all the flux
maintenance regimes operating, backwashing, chemical cleaning, and
if used, reverse crossflow.

Pilot system components Each type of pilot unit can have different
components, which should be chosen according to the application to be
tested. An illustration of a site trial unit for industrial TFF evaluations
is given in Figure 4.32 with a flow schematic diagram given in Figure
4.33. It will be noted that the design of the site pilot unit is necessarily
very much more complex than the simple laboratory bench scale units
discussed previously. The pilot rig often resembles full-scale equipment
in that it will have temperature control, instrumentation and automation
consistent with extended operating regimes.

Membrane filtration module: Modules with a membrane filtration

surface area of between 0.2 and 3.0 m^ are usually sufficient for scale
pilot testing.
180 Solid/Liquid Separation: Scale-up of Industrial Equipment

F i g u r e 4 . 3 2 Pilot scale test equipment

X
VMl

Retentate - ^ ^ o

iJ—LF. -L-KD
Perm e a t e ^ i^^

CBF17OO>

l b <c
IJJ_.J
CIP

m
f A Hut WdiH

Tank -ixH 0) I I C O M w^itt

-©^
A^
u
f-«ed

< ^

S I Check valve p % Automated corjtrc?! vcslve © Temperatuf« g o u ^ ^ ^ ? VaiiabJe speed drfve pump

5 9 Merribraise © Pressure gauge ^ < M o n u d byfterfly I * l Eledro magnetic ( ^ Litjuid detetior

valvo valve Howmeter

Figure 4 . 3 3 Pilot scale flow schematic

 4 • Membrane filters - microfiltration and ultrafiltration 181

Feed and re-circulation pump: Typically, feed pumps are low flow,
high pressure pumps while re-circulation pumps are high flow, low
pressure pumps. The re-circulation pump in the tangential flow filtration
system should provide the necessary velocity to reduce fouling, with
enough pressure head to overcome the channel/fibre pressure drop.
Pumps should be capable of generating the desired range of pressures
and flow rates to be tested. Variable speed drives provide an added
dimension of flexibility to the pilot test process.

Flowmeters'. The main types of flow meters used in pilot units are
electro-magnetic. These must be precise, reliable and versatile.

Flow control valves and level switches: Flow control valves are
especially used for pilot tests intended to run continuously over long
periods of time. At a minimum, these should allow the control of
permeate flow and bleed flow (concentrate leaving the system). Level
switches are needed to prevent pump failure if the tank level
unexpectedly drops too low or to help maintain a certain level in the tank.

Temperature and pressure gauges: Accurate and reliable measurement

of temperature and pressure values is very important in pilot testing as
these two parameters may have a significant influence on the filtration
performance. Glycerine-filled pressure gauges are recommended for
better accuracy. Direct recording analogue gauges must be capable of
accurately reading within 0.1 bar. Pressure transducers have high
accuracy readings and are readily adapted to data logging.

Back pulse: The back pulse set up is usually arranged to generate a

pulse 0.5 bar higher than the system trans-membrane pressure, at 10 to
30 minute intervals to ensure effective backwashing.

Feed tank: The tank should be of sufficient volume to allow for the
whole range of concentrations to be evaluated or to provide sufficient
residence times in situations where it is also used as a reaction vessel.
The materials of construction of the feed tank and its components,
usually stainless steel, must be compatible with the operating temper-
ature, pressure and chemical nature of the process sample. The tank
should have a lid and an anti-vortex device if necessary, and a sloping
or conical bottom to allow more complete drainage.

Heat exchanger: A multi-tube heat exchanger is generally placed in the

filtration loop to maintain the required temperature over the entire run
time and during in-place cleaning/sterilizing. Small volume recirculating