This item was submitted to Loughborough's Research Repository by the author.

Items in Figshare are protected by copyright, with all rights reserved, unless otherwise indicated.

The simulation, modelling and sizing of pressure filters

PLEASE CITE THE PUBLISHED VERSION

PUBLISHER

© Elsevier

VERSION

AM (Accepted Manuscript)

LICENCE

CC BY-NC-ND 4.0

REPOSITORY RECORD

Tarleton, E.S., and Richard J. Wakeman. 2009. “The Simulation, Modelling and Sizing of Pressure Filters”.
figshare. https://hdl.handle.net/2134/4851.
 This item was submitted to Loughborough’s Institutional Repository
(https://dspace.lboro.ac.uk/) by the author and is made available under the
following Creative Commons Licence conditions.

For the full text of this licence, please go to:

http://creativecommons.org/licenses/by-nc-nd/2.5/
 1

THE SIMULATION, MODELLING AND SIZING OF PRESSURE FILTERS

E.S. Tarleton† ([email protected]) and R.J. Wakeman††

†
Department Chemical Engineering, Loughborough University, Loughborough, Leics., UK.
††
Separation Processes Centre, University of Exeter, Devon, UK.

ABSTRACT

Some preliminary results from a research program aimed at providing computer simulations for
solid/liquid separation processes are presented. The personal computer software package
described models the operation of pressure filters. The example of the diaphragm filter press is
chosen and it is shown how the integration of industrially proven research data with powerful
modular software routines can simulate the cake formation, compression, dewatering and washing
operations. With a knowledge of the intended operating conditions, the software generates
extensive data for the filter cycle and minimises the time and cost associated with preliminary and
pilot scale test work.

INTRODUCTION

Of the many unit operations available to the process engineer, those associated with solid/liquid
separation are perhaps the most difficult to specify. For these systems it would be rare practice to
use fundamental theoretical relationships for either the selection, design or optimisation of
equipment with the result that the specification of chemical plant is rarely performed without
recourse to extensive laboratory and pilot scale tests. Whilst the task of selecting a solid/liquid
separator for a given duty has been considerably simplified by the recent introduction of the
computer software pC-SELECT1,2, its presence in the market place should be regarded in the
context of an initial step. When a particular type of separator has been identified, the separations
technologist is still faced with the prospect of undertaking an extensive test programme to enable
accurate scale-up and the optimisation of size and configuration. These problems are
compounded by the dearth of literature relating to such matters and the apparent lack of ‘standard’
procedures.

The authors have believed for many years that solid/liquid separation technology, whether it be in
the areas of selection, design or optimisation, is best dealt with by interactive computer software.
To be most effective the software must incorporate a well chosen mix of algorithms, expert system
and input information from the user, be based on sound ideas and proven data from experimental
and theoretical research projects and maintain the ability to give advice and education in a
congenial fashion so that any psychological barriers to computer use are avoided.

This paper describes some preliminary results from a software package designed to simulate the
operation of pressure filters. The software incorporates industrially proven methodologies which
are combined with computer code to produce modular routines capable of modelling cake
formation, compression, dewatering and washing. The techniques used to describe the operation
of the diaphragm filter press are shown and information regarding a typical filter cycle is presented.

FILTER CYCLE DESCRIPTION

The versatility of the diaphragm filter press design allows the range of operating configurations
shown in Figure 1.

The slurry is fed to the chambers of the press through suitable porting with either a centrifugal,
diaphragm, single screw or reciprocating pump3 such that filtration can occur at constant
Cite paper as: Tarleton E.S. and Wakeman R.J., 1994, The simulation, modelling and sizing of pressure filters, Filtration and
Separation, 31(4), 393-397. DOI: 10.1016/0015-1882(94)80455-9. Also presented at Filtech Conference, 1993, pp.121-130, Karlsruhe,
Germany.
 2

rate/variable pressure or variable rate/variable pressure depending on the type of pump used. If
the chambers are not completely filled by cake during this primary filtration stage then a secondary
filtration stage may follow where the residual suspension is filtered at constant pressure via
movement of the flexible diaphragm. When sufficient suspension has been filtered for the two
cakes to meet further cake consolidation via the diaphragm (at constant pressure), cake
dewatering by air blowing and/or cake washing can be undertaken in any order to complete the
filter cycle prior to cake discharge.

(i) & (ii) Cake Formation

Most filtration processes are described by separations technologists through the general filtration
equation

dtf α c μ μRm
= av2 c Vf + (1)
dVf A Δp AΔ p

where αav is the average specific cake resistance and the effective concentration of the solids in
the feed (cc) is given by

ρMs
cc = (2)
1 − mMs

In the general sense both the pressure and rate can vary and it is necessary to impose the pump
characteristics on the governing equations. Hence equation (1) can be rearranged to

1 ⎛ A 2 Δp ⎞
Vf = ⎜ − ARm ⎟ (3)
cc α av ⎝ μQ ⎠

where αav is a function of the pressure applied to the solid/liquid mixture through the pumping
action. To account for the potential compressibility it is common practice to relate the local specific
resistance (α) and porosity (ε) of filter cakes to the applied pressure over a specified range through
constitutive equations such as

α = α 0 Δp n (4)

ε = ε0 Δp n (5)

Combining the integrated form of eqn. (4) (to give an average cake resistance) with eqn. (3) and
performing a mass balance on the solids in the cake yields the two equations governing cake
formation during the primary filtration phase

1 ⎛ A 2 Δp ⎞
Vf = ⎜ − ARm ⎟ (6)
cc α 0 (1 − n ) Δpcake ⎝ μQ
n
⎠

⎛ ⎛ ρs ⎞⎞
⎜ Vf Ms ⎜ ( m − 1) + 1⎟ ⎟
h = 2⎜ ⎝ ρ ⎠⎟ (7)
⎜ ρ ⎟
⎜ A ρ (1 − mMs ) ⎟
s

⎝ ⎠

Cite paper as: Tarleton E.S. and Wakeman R.J., 1994, The simulation, modelling and sizing of pressure filters, Filtration and
Separation, 31(4), 393-397. DOI: 10.1016/0015-1882(94)80455-9. Also presented at Filtech Conference, 1993, pp.121-130, Karlsruhe,
Germany.
 3

When unfiltered suspension remains in the filter chamber(s) a secondary filtration phase may
follow the initial primary filtration. If the secondary phase is required then further filtration at
constant pressure is initiated via the diaphragm, with filtrate draining through one filtration surface
in each chamber. Modifying eqn. (1) to account for the effective increase in resistance to filtration
due to the presence of a filter cake on the filter medium at the start of the filtration gives the
governing equation

tf − ttr α av cc μ μ ⎛ α av c pVp ⎞
= (Vf + Vtr ) + ⎜⎜ + Rm ⎟
⎟
(8)
Vf − Vtr ( A 2 ) Δp
2
( A 2 ) Δp ⎝ ( A 2 ) ⎠

Thus, if the pump characteristics relating Δp and Q and the constitutive equations relating α and ε
(and hence m) to Δp are known from laboratory scale tests, cake formation can be modelled.

(iii) Cake Consolidation

The cake consolidation is performed at constant pressure through use of the filter diaphragm and
is analysed via an empirical relationship relating the consolidation ratio

ho − h
Uc = (9)
ho − h∞

to a dimensionless consolidation time defined by

(n )
2
Cc tc
Tc = 1 (10)
ω02

such that

4Tc π
Uc = (11)
(1+ ( 4T )
1 2ν
π)
ν
c

where Cc is the consolidation coefficient and ν the consolidation behaviour index. Values for ν and
the other scale-up constants in the constitutive equations for the consolidation phase can be
obtained through laboratory tests with, for instance, a piston press (as previously described4) and
related through eqns. (9)-(11) to give expressions for cake height and filtrate volume during the
consolidation period(s).

Consolidation is assumed to end when the flow rate of liquid from the filter falls below a pre-
determined level.

(iv) Cake Dewatering (by air blowing)

The dewatering phase is analysed through a combination of empirical and theoretical relations.
These facilitate the calculation of the time required to achieve a given filter cake moisture content
and the air flow rate through the cake during the dewatering period and have been described in
detail previously5-7. Two parameters of prime importance in dewatering are the pressure required
to initiate dewatering (pb) and the irreducible saturation of the filter cake under consideration (S∞).
Although these should be measured experimentally, the former can be determined with sufficient
accuracy from

Cite paper as: Tarleton E.S. and Wakeman R.J., 1994, The simulation, modelling and sizing of pressure filters, Filtration and
Separation, 31(4), 393-397. DOI: 10.1016/0015-1882(94)80455-9. Also presented at Filtech Conference, 1993, pp.121-130, Karlsruhe,
Germany.
 4

4.6 (1 − εav ) σ
pb = (13)
εav xav

and correlations exist (for relatively large, granular materials) which allow the estimation of S∞

Ncap =
3
εav 2
xav ( ρgh + Δpd ) (14)
(1 − εav ) hσ
2

( −0.49
S∞ = 0.155 1 + 0.031Ncap ) (15)

With data for pb and S∞ available, and assuming that (i) no dewatering of the cake occurs through
evaporation of the mother liquor, and (ii) the displacing fluid is saturated instantly as it enters the
cake, a dimensionless saturation (SR, termed the reduced saturation) can be related empirically to
a dimensionless time (θ)

S − S∞
SR = (16)
1 − S∞

1 b3
⎛ 1⎛ 1 ⎞⎞
θ = ⎜⎜ ⎜ − 1⎟ ⎟⎟ (17)
⎝ b2 ⎝ SR ⎠⎠

where b2 and b3 are known empirical constants. The dewatering time (td) is then estimated from

θεav μh 2 (1 − S∞ )
td = (18)
Δpd kav

Allowing for losses in the feed pipes to the filter and the effects of plant elevation, the air flow rate
required to achieve the desired moisture content can be calculated6,7.

(v) Cake Washing (by displacement)

The technique used in the analysis of the washing phase(s) employs the dispersion model7-10. Use
of the model requires a superficial wash velocity

Δpw
uo = (19)
μw ( α av ρs h (1 − εav ) + Rm )

and the product Re.Sc to be calculated

uo xav
Re.Sc = (20)
εav D

The ratio DL/D can then be evaluated according to the filter cake height

DL
= 0.707 + 55 (Re.Sc )
0.96
0 ≤ h ≤ 0.1 m (21)
D

Cite paper as: Tarleton E.S. and Wakeman R.J., 1994, The simulation, modelling and sizing of pressure filters, Filtration and
Separation, 31(4), 393-397. DOI: 10.1016/0015-1882(94)80455-9. Also presented at Filtech Conference, 1993, pp.121-130, Karlsruhe,
Germany.
 5

DL
= 0.707 + 1.75 (Re.Sc ) h > 0.1 m (22)
D

and the dispersion number is then given by

Dn =
(Re.Sc ) h (23)
xav ( DL D )

With a knowledge of the number of wash ratios to be applied the washing time (tw), effluent solute
concentration (c) and fractional solute recovery (F) can be estimated from known, experimentally
proven, correlations when allowances for cake saturation and scale are included7.

AN EXAMPLE OF FILTER SIMULATION

The above methodologies enable flexible simulations to be developed. When suitably applied
these can model most combinations of cake formation, consolidation, dewatering and washing to
allow preliminary sizing and scale-up of equipment to be performed. Although the example of a
diaphragm press has been chosen, the procedures can be readily adapted to model the
performance of other filtration devices such as tube filter presses and rotary vacuum filters. The
data shown below illustrate how information associated with the entire filter cycle can be produced
for a set of typical operating conditions through software simulation techniques.

Example Problem

A diaphragm filter press has been identified as being potentially suited to the task of producing a
nominal 5 te of dry particulate solids from an 8% w/w slurry. Following the formation stage(s), the
homogeneity of the cake is to be improved through consolidation with the diaphragm. The
resultant compact is then to be dewatered by air blowing until a moisture content of 20% is
achieved and subsequently washed until the solute concentration in the cake is reduced from 30 to
2.5 kg m-3. The feed is delivered to the filter press via a centrifugal pump (see Table 1 for the
characteristic curve of the pump) and the feed suspension characteristics have been
experimentally determined (see Table 2).

Determine if the separation can be achieved, the required filtration time, the filter area and other
performance indicators.

Solution via Simulation

When the computer simulation package is used to model the above problem a wealth of data
related to both the individual phases and the total filter cycle is produced. Figures 2-4 give
examples. The data suggest that the required objectives of the separation can be readily achieved
with the diaphragm press and pump specified. The filter area and total cycle time required are
estimated to be 380.3 m2 and 7678 s respectively with the filtration, consolidation, dewatering and
washing phases projected to end at times corresponding to 2028 s, 4056 s, 4707 s and 7678 s.

During the period of the primary filtration the cakes in each of the filter chambers progressively
grow to a combined height (noting that cake formation occurs on two surfaces in each chamber) of
34 mm whilst the pressure in the chambers increases from 49.7 to 94 kPa as the cakes form. At
the end of the primary filtration period (tf = 1138 s) un-filtered suspension still remains in the
chambers and a secondary filtration at constant pressure is required. Within this period the cakes
in each chamber continue to form and eventually join to achieve the maximum allowable height of
40 mm (i.e. 2x20 mm) after 2028 s. At the end of the cake formation stages the chamber and cake

Cite paper as: Tarleton E.S. and Wakeman R.J., 1994, The simulation, modelling and sizing of pressure filters, Filtration and
Separation, 31(4), 393-397. DOI: 10.1016/0015-1882(94)80455-9. Also presented at Filtech Conference, 1993, pp.121-130, Karlsruhe,
Germany.
 6

moisture contents are identical and equal to 52.1% with the total solids, liquid and solute masses in
the filter estimated at 5092, 5534 and 166 kg respectively.

Following the cake formation phases the consolidation, dewatering and washing phases are
sequentially initiated. During the period of consolidation the cake moisture and solute masses are
reduced to 27% and 57 kg respectively whilst the cake height is also reduced to 21 mm and is
assumed to remain at that value for the subsequent dewatering and washing phases. The solute
mass and moisture content of the cake are further lowered to 38 kg and the desired 20%
respectively at the end of the dewatering period using an overall air rate of 1.8x10-5 m3 m-2 s-1. In
the washing period, 3.5 wash ratios (based on the unsaturated cake) are used to produce the
required solute concentration of 2.5 kg m-3; the fractional recovery for the washing period alone
being 88%.

At the end of the filter cycle it is estimated that 5092 kg of solids, 1891 kg of liquid and 4.7 kg of
solute remain in the filter press. The fractional recovery of solute for the whole cycle is calculated
to be 97.2%.

CONCLUDING REMARKS

The above considerations give an indication of the manner in which computer simulations can
significantly improve the understanding of the operations which can occur during the filter cycle.
Each of the theories used to model the cake formation, consolidation, dewatering and washing
phases has been proven in pilot scale and industrial tests. It is apparent that the modelling
procedures adopted for the simulation of the diaphragm filter press can be readily adapted to
include optimisation routines and extended to encompass other types of separation equipment.
When fully developed, such computer software should provide a suite of programs which are
sufficiently powerful to minimise the number of costly and time consuming experiments which need
to be performed to assess the suitability of equipment for a particular duty.

NOMENCLATURE

A Total area of filter medium (m2)

a Index in eqn. (5) (-)
b2, b3 Coefficients used in eqn. (17) (-)
Cc Modified consolidation coefficient (m2 s-1)
c Instantaneous solute concentration in wash effluent (kg m-3)
cc Effective feed suspension concentration (kg m-3)
D Molecular diffusivity of solute (m2 s-1)
DL Dispersion coefficient (m2 s-1)
Dn Dispersion number (-)
e Voids ratio (-)
F Fraction of recoverable solute removed from filter cake (-)
g Acceleration due to gravity (m s-2)
h Total filter cake thickness (m)
k Filter cake permeability (m2)
Ms Solids mass fraction in feed suspension (kg kg-1)
m Ratio mass wet cake to mass dry cake (kg kg-1)
Ncap Capillary number (-)
n Index in eqn. (4) (-)
n1 Number of drainage surfaces per filter chamber (-)
pb Breakthrough pressure (Pa)
Q Flow rate of liquid from filter (m3 s-1)
Re Reynolds number (-)
Cite paper as: Tarleton E.S. and Wakeman R.J., 1994, The simulation, modelling and sizing of pressure filters, Filtration and
Separation, 31(4), 393-397. DOI: 10.1016/0015-1882(94)80455-9. Also presented at Filtech Conference, 1993, pp.121-130, Karlsruhe,
Germany.
 7

Rm Filter medium resistance (m-1)

S Filter cake saturation (-)
SR Reduced saturation (-)
S∞ Irreducible saturation (-)
Sc Schmidt number (-)
Tc Dimensionless consolidation time (-)
t Time (s)
U Consolidation ratio (-)
uo Superficial wash velocity (m s-1)
V Cumulative volume of liquid produced from filter (m3)
xav Mean particle size in feed suspension (m)

Greek symbols
α Local specific cake resistance (m kg-1)
αav Average specific cake resistance (m kg-1)
α0 Specific cake resistance at Δp = 1 kPa (m kg-1)
Δp Pressure applied to solid/liquid mixture (Pa or kPa)
Δpcake Pressure drop over filter cake (Pa)
ε Local filter cake porosity (-)
εav Average filter cake porosity (-)
ε0 Filter cake porosity at Δp = 1 kPa (-)
θ Dimensionless time (-)
μ Liquid viscosity (Pa s)
ν Consolidation behaviour index (-)
ρ Liquid density (kg m-3)
ρs Solids density (kg m-3)
σ Liquid surface tension (N m-1)
ωo Volume of solids produced per unit area of drainage surface (m3 m-2)

Subscripts
c consolidation phase
d dewatering phase
f filtration phase
o start of phase
p end of previous phase
tr transition from primary to secondary filtration phase
w washing phase
∞ at equilibrium

REFERENCES

1. E.S. Tarleton and R.J. Wakeman, Solid/Liquid Separation Equipment Simulation & Design: pC-
SELECT - Personal computer software for the analysis of filtration and sedimentation test data
and the selection of solid/liquid separation equipment, Separations Technology Associates,
1991.

2. R.J. Wakeman and E.S. Tarleton, Solid/liquid separation equipment simulation and design - an
expert systems approach, Filtration and Separation, 28(4), 268-274, 1991.

3. I.M. Young, in Solid/liquid Separation Equipment Scale-up, 2nd Ed., D.B. Purchas and R.J.
Wakeman (Eds.), pp.446-484, Uplands Press and Filtration Specialists, London, 1986.

4. R.J. Wakeman, M.N. Sabri and E.S. Tarleton, Factors affecting the formation and properties of
wet compacts, Powder Technology, 65(1-3), 283-292, 1991.
Cite paper as: Tarleton E.S. and Wakeman R.J., 1994, The simulation, modelling and sizing of pressure filters, Filtration and
Separation, 31(4), 393-397. DOI: 10.1016/0015-1882(94)80455-9. Also presented at Filtech Conference, 1993, pp.121-130, Karlsruhe,
Germany.
 8

5. R.J. Wakeman, Vacuum dewatering and residual saturation of incompressible filter cakes,
International J. Mineral Processing, 3, 193-206, 1976.

6. R.J. Wakeman, An improved analysis for the forced gas deliquoring of filter cakes and porous
media, J. Separation Process Technology, 3, 32-38, 1982.

7. R.J. Wakeman and E.S. Tarleton, Modelling, simulation and process design of the filter cycle,
Filtration and Separation, 27(6), 412-419, 1990.

8. R.J. Wakeman, Transport equations for filter cake washing, Chemical Engineering Research
and Design, 64, 308-319, 1986.

9. R.J. Wakeman and G.J. Attwood, Developments in the application of cake washing theory,
Filtration and Separation, 25, 272-275, 1988.

10. R.J. Wakeman and G.J. Attwood, Simulations of dispersion phenomena in filter cake washing,
Chemical Engineering Research and Design, 68, 161-171, 1990.

Cite paper as: Tarleton E.S. and Wakeman R.J., 1994, The simulation, modelling and sizing of pressure filters, Filtration and
Separation, 31(4), 393-397. DOI: 10.1016/0015-1882(94)80455-9. Also presented at Filtech Conference, 1993, pp.121-130, Karlsruhe,
Germany.
 9

FIGURES AND TABLES

Figure 1: Possible operations in a diaphragm filter press.

Cite paper as: Tarleton E.S. and Wakeman R.J., 1994, The simulation, modelling and sizing of pressure filters, Filtration and
Separation, 31(4), 393-397. DOI: 10.1016/0015-1882(94)80455-9. Also presented at Filtech Conference, 1993, pp.121-130, Karlsruhe,
Germany.
 10

0.045

0.040

0.035

0.030
Cake thickness (m)

0.025

0.020

end of consolidation
0.015
end of filtration (1)

end of filtration (2)

end of dewatering

end of washing
0.010

0.005

0.000
0 2000 4000 6000 8000
Cycle time (s)

Figure 2: Total cake thickness per chamber vs. filter cycle time.

100

chamber

80
Moisture content (%)

60
cake

40
end of filtration (2)
end of filtration (1)

20
end of wash.
end of con.

end dew.

0
0 2000 4000 6000 8000
Cycle time (s)

Figure 3: Chamber and cake moisture contents vs. filter cycle time.

Cite paper as: Tarleton E.S. and Wakeman R.J., 1994, The simulation, modelling and sizing of pressure filters, Filtration and
Separation, 31(4), 393-397. DOI: 10.1016/0015-1882(94)80455-9. Also presented at Filtech Conference, 1993, pp.121-130, Karlsruhe,
Germany.
 11

200

Mass of solute in cake (kg) 150

100

end of filtration (2)

end of filtration (1)

50

end of washing
end of con.

end dew.
0
0 2000 4000 6000 8000
Cycle time (s)

Figure 4: Mass of solute in cake liquor vs. filter cycle time.

Flow rate (m3 s-1) Pressure (kPa)

0 360
0.022 300
0.03 250
0.035 200
0.037 150
0.0385 100
0.0395 50
0.04 0

Table 1: Pump characteristics.

Cite paper as: Tarleton E.S. and Wakeman R.J., 1994, The simulation, modelling and sizing of pressure filters, Filtration and
Separation, 31(4), 393-397. DOI: 10.1016/0015-1882(94)80455-9. Also presented at Filtech Conference, 1993, pp.121-130, Karlsruhe,
Germany.
 12

Filter chamber thickness 80 mm

Filter medium resistance 3x1011 m-1
Max. cake thick. on each filter surface 20 mm
Density of solids 2500 kg m-3
Density of filtrate and wash 998 kg m-3
Viscosity of filtrate and wash 0.001 Pa s
Surface tension of filtrate and wash 0.07 N m-1
Solids concentration in feed 8% w/w
Mean size of solids in feed 10 μm
Secondary filtration pressure (if needed) 600 kPa
Consolidation index 3
Consolidation and washing pressure 600 kPa
Dewatering pressure 400 kPa
Barometric pressure 100 kPa
Irreducible cake saturation 0.26
Viscosity of air 1.8x10-5 Pa s
Solute concentration in feed 30 kg m-3
Solute concentration in wash 0.5 kg m-3
Solute diffusivity 1x109 m2 s-1
Constitutive eqns. for filtration α = 6x109Δp0.6; e = 3 - 0.1log(Δp)
Constitutive eqns. for consolidation e = 1.63 - 0.55log(Δp); Cc = 1x10-8Δp0.1

Table 2: Values used in simulation.

Cite paper as: Tarleton E.S. and Wakeman R.J., 1994, The simulation, modelling and sizing of pressure filters, Filtration and
Separation, 31(4), 393-397. DOI: 10.1016/0015-1882(94)80455-9. Also presented at Filtech Conference, 1993, pp.121-130, Karlsruhe,
Germany.