2 9 0 Solid/Liquid Separation: Scale-up of Industrial Equipment

Cake moisture can be mechanically and thermally reduced on certain

types of vacuum filter. Cake compression is available for most drum
and belt filters, whereas thermal drying is more suited to belt filters
due to their configuration.
Excluding any mechanical and thermal drying mechanisms, and
assuming that for any given system the cake permeability and particle
nature remain constant, then the only remaining variable is drying
time. A plot of drying time against cake moisture will indicate the
potential benefits of extending the drying time for a given cycle.
Figure 6.33 is a typical cake moisture profile. The available drying time
is usually an operating constraint of the filter unit, more so with a drum
filter than with a horizontal belt, and so a practical approach has to be
adopted. In this example, there is no significant benefit in extending the
drying time beyond the knee of the graph since there is no appreciable
reduction in moisture with increasing time. Cake moisture content
should be low enough to ensure effective cake discharge and meet any
operational and process constraints (e.g. as feed into a dryer, or
'stability' for landfill). In fact, higher cake moistures may be perfectly
acceptable as part of the solids handling system, especially if the solids
are to be re-slurried for further processing.

F i g u r e 6 . 3 3 Cake moisture as a function of drying time

Cake moisture decreases rapidly during the initial stages of the drying
cycle as the interstitial fluid is drawn out of the cake structure. The
bulk of the fluid is displaced between the moment the cake surface just
falls 'dry' and when the air flow breaks through. Thereafter, the cake
moisture reduces slowly with time as more voids within the cake
structure are gradually evacuated by the air flow.
 6 -Vacuum filters 291

6.5.4 Filter cake washing

In some applications cake washing is an integral part of the filtration
process. It is particularly important when the purity of the solids needs
to be maintained, or the filtrate is valuable. The quantity and applica-
tion of wash liquor needs to be carefully assessed to ensure that
underwashing, overwashing or ineffective washing does not occur. The
filter needs to be able to remove the wash liquor and reduce the cake
moisture content to an acceptable level prior to discharging the solids.
This must be achieved during the normal dewatering and drying part of
the overall filtration cycle.
Cake washing is the process by which the interstitial liquor within the
filter cake is displaced, thereby recovering more valuable filtrate or
increasing the purity of the solids. It can be assumed that displacement
is by plug flow, but this assumption is only valid if no cake cracking
occurs during the wash. In these instances, the wash liquor would be
drawn through the cracks in the filter cake, thus bypassing the majority
of solids and making the wash ineffective and wasteful of wash liquor.
The wash efficiency is a measure of the residual solute in the filter
cake after washing when compared with the cake in its unwashed
condition. It can be expressed as:

n.=^^^oo (20)

For many applications that use cake washing, the mother liquor may
contain trace salts. In such cases it is easy to measure the washing
efficiency as it can be correlated to the final salt content of the filter
cake. Where the filter cake is reslurried prior to another processing
operation, the washing efficiency can be monitored on line by
measuring the electrical conductivity of the slurry.
Effective washing is not just about determining the volume of wash
liquor required to obtain a certain purity of solid or recovery of filtrate.
The fundamental requirement of all cake washing applications is the
even distribution of wash liquor over the filter cake thereby ensuring a
good wash efficiency in the allowable time. This can be achieved by
gentle wash sprays, mists or weir boxes, although in reality the
washing efficiency that can be achieved under laboratory and pilot
conditions is usually greater than can be achieved on full-scale filters,
particularly for drum filters. Unless re-slurry washing is required, it is
imperative that the wash liquor does not disturb the structure of the
filter cake, especially on rotary drum filters.
292 Solid/Liquid Separation: Scale-up of Industrial Equipment

Once the filter cake has formed, the solids start the dewatering and
drying part of the overall filtration cycle. It is during this period that
cake washing takes place. The wash cycle, where used, forms part of
the dewatering and drying cycle. Application of wash liquor is
normally delayed until there is no visual surface liquid on the filter
cake - the cake changes from a shiny, gloss finish to matt. At this
point, the filter cake has fully formed, and washing can start. It is
imperative that the wash liquor is applied as soon as the cake surface
falls dry in order to prevent cake cracking. The wash liquor is applied
to the surface of the filter cake from where it is drawn through the cake
structure by the vacuum, thereby displacing the mother liquor. The
volume of wash liquor must be sufficiently great to ensure that the
desired solids purity or filtrate recovery can be achieved.
After the wash liquor has been applied, the filter cake will resume the
dewatering and drying process prior to discharge. The volume of wash
liquor and the duration of its application must be regulated so that the
solids discharge characteristics are not compromised.
On rotary drum filters, cake washing occurs up to just before top dead
centre, so that the wash liquid runs back down the drum surface away
from the solids discharge point. Excessive quantities of wash liquor
could fall back into the filter trough and dilute the feed suspension
which is detrimental to the operation of the filter. The maximum
quantity of wash liquor for a filter is the volume that can be drawn
through the filter cake during the allocated wash cycle.
Similarly, on horizontal belt filters, cake washing usually starts after
the solids lose the surface liquid. Horizontal belt filters, by virtue of
their design, can provide extensive washing cycles. On many instal-
lations, the wash cycle can be set up to provide, for example, multi-
stage counter-current washing and reflux washing - this cannot be
done with rotary drum filters.
Precoat drum filters need relatively small volumes of wash liquid when
compared with other drum filters, and only require sufficient wash to
displace the mother liquor from the outer 'skin' before it is peeled
away from the precoat bed. Mist and atomising sprays are the most
convenient ways to provide a gentle, even wash on precoat filters.
It is possible to estimate the theoretical wash efficiency and wash
liquor volume during the scale-up tests, and this is discussed briefly in
Section 6.4.2, the Test Leaf Procedures. In the majority of rotary drum
filter applications, the wash efficiency is hindered by the mechanics of
applying a wash liquor to a rotating, curved surface. The normal
response is to increase the wash liquor flow rate but this often causes
 6 • Vacuum filters 293

operational difficulties. The method of application and wash rates

should be carefully considered to ensure optimum washing efficiency.
Conversely, washing on horizontal belt filters is very efficient and
even, thorough washing can be readily achieved. For further advice on
cake wash, it is recommended that the equipment supplier is consulted.

6.5.5 Time and mechanical degradation of agglomerated particles

The use of flocculants and coagulants to agglomerate fine particles is
common in many solid/liquid filtration systems. The main benefit of
using these additives is the increase in the apparent particle size by
bonding many smaller particles together. This reduces the effective
specific surface and, in turn, increases the overall cake permeability in
line with equation (10). However, long-term storage of agglomerated
suspensions and/or mechanical agitation and pumping can break the
agglomerates back down to their constituent parts, which has an
adverse effect on the filterability of the suspension.
The correct selection and use of flocculants and coagulants is usually
carried out by the manufacturers - more information on this is given in
Chapter 2. For filtration processes, the use of low molecular weight
(LMW) polymer flocculants is preferred since the floes have a greater
resistance to shear (Moss and Dymond, 1968). High molecular weight
(HMW) polymers are more suitable for sedimentation processes as
they provide better flocculation and faster settling times. Floes formed
with HMW polymers also have a tendency to trap moisture within the
structure, increasing the overall moisture content of the filter cake.
To assess the stability of the floes within the system it is necessary to
repeat the filtration tests at regular time intervals, although it is
unlikely that laboratory conditions could fully mimic the actual full-
scale process conditions and therefore these tests would provide
indicative information only.

6.5.6 Air flow rate and applied vacuum

The air rate through a vacuum filter is dependent upon the permeability
of the filter cake during the dewatering phase. The flow of air through
the cake during its formation phase can be assumed to be zero - only
liquid passes through the cake at this stage of the cycle. Once the cake
has formed and begins to dewater the air displaces the moisture in the
cake structure. The air flow through the cake increases from zero to an
asymptotic maximum value as shown in Figure 6.34, and can be
measured during a leaf test by attaching a suitable rotameter to the
2 9 4 Solid/Liquid Separation: Scale-up of Industrial Equipment

exhaust side of the vacuum pump. The curve shows the difference
between the dewatering and drying phases.

START OF
CAKE CRACKING

i •'" .^,_.^..-.» f

BREAK THROUGH

Figure 6 . 3 4 Typical air flow profile through a filter test leaf

The exact shape of the curve will be a function of the filter cake
permeability, and may be influenced by cake cracking. Cake cracking
is a characteristic of the solid/liquid mixture and the way in which the
filter cake forms. Cracking does not always occur, but is usual during
the dewatering/drying phase and the total air flow must be based on
this condition. At the point of cake cracking, the air flow through the
test leaf makes a step increase as the air bypasses the filter cake, taking
the route of least resistance. In many instances, cake cracking coincides
with the start of the drying process so no step increase in air flow is
apparent. The total air flow requirement for the dewatering/drying
phase of the filter cake is the mean value of the curve, although the
maximum value is always the more conservative option. The air flow
rate through the filter cake is usually expressed in m^ mr^ hrK
Whatever the type of filter, there are likely to be some minor air losses
as a consequence of incomplete seals at either the filter medium, the
rotary valve or interconnecting pipe work on vacuum connections.
These potential losses must be allowed for when sizing the vacuum
pump. Good maintenance on the equipment should keep these air
losses to an acceptable minimum. It is not uncommon to add at least
10% to the mean air flow figure from the test work. This also allows
for the attenuation of the air between the source (i.e. vacuum pump)
and the filter surface, as well as the additional air volume associated
with the interconnecting pipe work.
The vacuum pump should be selected in conjunction with the pump
supplier, who should be made aware of anything that could influence
 6 • Vacuum filters 295

the choice of pump. For example, any temperature variations in the

feed sample may not significantly affect filterability, but could alter the
vapour pressure and therefore applied vacuum. Similarly, the location
(altitude) of the proposed installation relative to the test laboratory
may influence the pump size. At higher altitudes, the vacuum pump
size must be increased for the same operating capacity.
The applied vacuum depends upon the filter cake structure and mean
particle size. Large, crystalline solids, forming a high permeability
filter cake usually have a lower applied vacuum of between -0.2 and
-0.3 barg. This is also true for precoat beds formed from diatomaceous
earth or perlite. For fine particle suspensions, like pigments and metal
oxides/hydroxides, the applied vacuum is usually much higher at
between -0.6 and -0.8 barg. For most applications the normal
operating vacuum is less than -0.7 barg and this is often the basis for
laboratory tests.

6.6 Laboratory and pilot scale test units

There are many benefits to conducting filter leaf tests, especially in

terms of the ease of operation, flexibility when investigating the
characteristics of several samples and the need for only small test
volumes. The main limitation is that the tests are only a snapshot of the
filtering process. In many instances, further experimentation is required
to produce additional data to validate the filtration process and, under
these circumstances, laboratory scale (see Figure 6.35) and pilot scale
(see Figure 6.36) filtration equipment is necessary.

6.6.1 Laboratory scale trials

Laboratory scale trials are the next step up from leaf tests, and are
often used to process small batches of suspension over several hours.
Here, the mechanical handling of the suspension in relation to a
vacuum filter can be investigated, for example, the method of feeding
the suspension to the filter and the application of wash liquor. Larger
volumes of filtrate and filter cake can be collected for further analysis
and product validation. In addition, these extended trials can highlight
any potential problems with filter medium compatibility, such as
blinding. The laboratory scale filters can be set up with all the operating
parameters of a full-scale unit, such as vacuum levels, cycle times and
filter medium. In general, scale-up factors for laboratory equipment
would be the same as for the standard filter leaf tests.