Professional Documents
Culture Documents
a b s t r a c t
A precipitated calcium carbonate with Sauter mean diameter of 7.5 m was ltered under conditions of constant rate
and constant pressure in a comparative laboratory investigation. The specic cake resistance to ltration was found
to vary between 1 109 and 1 1011 m kg1 , depending on the applied pressure, and the corresponding lter cake
volume concentrations were between 0.42 and 0.54 (v/v). The calculated specic resistance, from the particle size
distribution data and the KozenyCarman equation is one order of magnitude lower than that measured, even though
the solids were extremely robustly characterised. Practical lter testing rather than design based on size distributions
is known to be essential. However, the conventional approach is to use constant pressure laboratory tests, the results
presented here demonstrate that constant rate ltration is a more reliable method for data acquisition, especially
when determining the lter medium resistance, and readily available laboratory equipment is adequate for use.
2012 The Institution of Chemical Engineers. Published by Elsevier B.V. All rights reserved.
Keywords: Vacuum ltration; Calcium carbonate; Resistance; Cake compressibility
1.
Introduction
1.1.
Filtration
Corresponding author. Tel.: +44 1509 222519; fax: +44 1509 223923.
E-mail address: R.G.Holdich@Lboro.ac.uk (R.G. Holdich).
Received 9 August 2012; Received in revised form 10 October 2012; Accepted 19 November 2012
0263-8762/$ see front matter 2012 The Institution of Chemical Engineers. Published by Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.cherd.2012.11.012
Please cite this article in press as: Mahdi, F.M., Holdich, R.G., Laboratory cake ltration testing using constant rate. Chem. Eng. Res. Des. (2012),
http://dx.doi.org/10.1016/j.cherd.2012.11.012
ARTICLE IN PRESS
chemical engineering research and design x x x ( 2 0 1 2 ) xxxxxx
Nomenclature
A
C
c
(1)
1.2.
1.2.1.
Advantages
A huge number of experimental studies.
Since liquid ow in the cake is additive, the ow rate necessarily increases as the liquid approaches the medium
compressing the lter cake.
Simplicity of obtaining the data.
Disadvantages
Calculations and results have to be treated with caution;
there is often a lack of consistency, especially when different
personnel perform the tests.
At the start there may be experimental difculties, such as
establishing a constant pressure over the lter cake, due to
a period of medium blocking.
There is not enough time to study the initial ltration period
1.2.2.
Advantages
Upper and lower limits on the variation of solids pressure
in the lter cake can easily be obtained.
Industry use is widespread (apart from vacuum lters).
Constant growth rate of the cake is induced with reports of
homogeneous particle incorporation into the growing cake
surface.
The particle packing structure is expected to be independent of cake thickness so that uniform compacts are
formed.
.
Disadvantages
The applied pressure has to be continuously increased with
cake thickness.
In case of investigating gas deliquoring under pressure then
this test rig would need to be modied.
It is common in data analysis to assume that there is no
effect of sedimentation during cake formation. Hence, the
cake resistance and permeability are calculated using equations based on Darcys law for liquid ow through a cake from
Please cite this article in press as: Mahdi, F.M., Holdich, R.G., Laboratory cake ltration testing using constant rate. Chem. Eng. Res. Des. (2012),
http://dx.doi.org/10.1016/j.cherd.2012.11.012
ARTICLE IN PRESS
1.3.
Basic analysis
As ltration proceeds, a porous cake of solid particles accumulates on a porous lter medium; its ltration rate can be
found from
rate of ltration =
driving force
resistance
constants in Eq. (5) can then be found from the intercept and
slope of a linear plot of P against V, noting that dV/dt is a constant, by denition. In the case of compressible cake ltration
equation (5) is still valid, at any increment in time, and the
intercept can be used to determine the lter medium resistance, but the relation will deviate from linearity as specic
resistance changes with the applied pressure over the lter
cake.
dV
AP
=
R
dt
(2)
R dV
A
dt
(3)
(4)
Rm dV
av c
dV
+
(V)
dt
A dt
A2
(5)
2.
2.1.
Material
Suspensions of aqueous calcium carbonate (CaCO3 Fordacal30 manufactured by MINELCO) at different solid concentrations: 9%, 11%, 13%, 15%, 17%, 19% and 21% by volume,
were prepared. In order to avoid too dilute feed suspension,
where sedimentation may inuence the ltration results, 9%
(v/v) was the lowest initial suspension concentration used.
Each suspension was prepared by dispersing the dry powder
in reverse osmosis water to avoid the ionic variation found
with tap water usage. The natural pH of the suspension was
8 and this was constant for all the tests. A Sauter mean
diameter of 7.5 m was measured using a Horiba LA-920-wet
Laser dynamic scattering device. Scanning electron microscope (SEM) images of the calcium carbonate powder show
a rhomboidal shape, indicating that the material is mainly
calcite, see Fig. 1. The values of the D10 , D50 and D90 particle diameters, measured by the Horiba at a variety of different
recirculation pump speeds, are shown in Fig. 2.
2.2.
Methodology
Please cite this article in press as: Mahdi, F.M., Holdich, R.G., Laboratory cake ltration testing using constant rate. Chem. Eng. Res. Des. (2012),
http://dx.doi.org/10.1016/j.cherd.2012.11.012
ARTICLE IN PRESS
chemical engineering research and design x x x ( 2 0 1 2 ) xxxxxx
Fig. 2 Variation of the particle diameters (D10 , D50 and D90 ) over time for six different pump speeds (P.Sp.) 1, 3, 6, 9, 12 and
15 rpm during analysis by laser diffraction.
vacuum. A vacuum gauge (WIKA Instruments Ltd.) was also
connected to the system in order to validate and check the
sensor reading. The ltrate was pumped out using a peristaltic pump (Watson Marlow 401U/D1) and collected in a
vessel placed on an electronic balance (OHAUS SPU601). The
weight of the ltrate and pressure drop in the system, as a
function of time, were recorded using a PC within Labview
software. For the tests three different ltrate pump settings
were used (10, 30 and 50%) of full-scale, which gave rise to
ARTICLE IN PRESS
Table 1 Measured cake heights and determined cake concentrations with 95% condence limits on height
measurements.
In. concentration (v/v)
Errora % ()
9%
5
10
30
20.12
19.31
18.28
0.475
0.494
0.522
4.46
4.45
4.90
11%
5
10
30
24.36
21.21
19.85
0.479
0.550
0.588
2.76
3.17
3.40
13%
5
10
30
31.35
27.76
26.61
0.440
0.497
0.518
4.48
5.06
5.30
15%
5
10
30
36.38
32.33
30.34
0.438
0.492
0.525
2.05
2.30
2.40
17%
5
10
30
44.22
35.48
34.72
0.408
0.508
0.520
2.19
2.74
2.80
19%
5
10
30
45.02
39.30
37.86
0.448
0.513
0.533
1.17
1.34
1.40
21%
5
10
30
51.81
44.68
42.09
0.430
0.499
0.528
1.95
2.27
2.40
95% condence that the concentration is the one provided plus or minus two times the standard deviation based on the cake height
measurement variability.
3.
3.1.
3.1.1.
SEM image
3.1.2.
The variation of the D10 , D50 and D90 particle diameters over
time for six different values of shear elds created by increasing the laser diffraction equipment pump speed from 1 to 15,
is shown in Fig. 2. This work done with and without sonication and the results were the same for both cases. The lack
of variation of these particle diameters shows that there was
no signicant aggregation of the calcium carbonate particles
suggesting that particle clustering is not a problem with this
material under the prevailing conditions. The water used during these tests was collected ltrate.
3.2.
Please cite this article in press as: Mahdi, F.M., Holdich, R.G., Laboratory cake ltration testing using constant rate. Chem. Eng. Res. Des. (2012),
http://dx.doi.org/10.1016/j.cherd.2012.11.012
ARTICLE IN PRESS
chemical engineering research and design x x x ( 2 0 1 2 ) xxxxxx
Fig. 4 Effect of pressure on cake concentration with initial concentrations 921 (% v/v); number in the gure indicates
pump speed 5, 10 and 30 [errors bars provide 95% condence interval].
that the cake is of uniform, or average, concentration throughout its height and is discussed further later.
The uncertainty bars around the mean values in Fig. 4 represent two times the standard deviation around the mean
concentration value, based on the ve height measurements
used to calculate that concentration. Hence, the condence
level in the cake concentration is again 95% within the range
illustrated by the bars. Fig. 4 shows three different trends,
which depend on pump speed and initial suspension concentration. Pump speed determines the ltration ow rate and,
consequently, the pressure drop in the lter cake, which has
the expected power-law type of relation with the cake concentration. The cake volume concentration was used to calculate
the dry cake mass per unit volume of ltrate (Holdich, 2002),
using Eq. (6).
c=
1
(1 s)/(s) (1 C)/(Cs )
sM
VF
(7)
3.3.
Specic resistance by constant rate and constant
pressure testing
(6)
The two linear equations represented by Eqs. (5) and (4) represent the conventional method of analysis for determining
specic resistance from constant rate and constant pressure
ltrations respectively. Examples of the results from these
tests are shown in Figs. 5 and 6 respectively. In Fig. 5 three
different starting concentrations are illustrated: 11%, 15% and
21% (v/v). There is a very slight intercept above the origin on
the pressure axis, indicating a small but nite lter medium
resistance, in accordance with Eq. (5). It is relatively easy to
determine the intercept from the cut-away graph and linearity of these data plots is good, after about 10 mL of ltrate and
up to about 120 mL, after which the ltration is substantially
completed. For comparison, Fig. 6 shows a typical set of data
achieved using the constant pressure ltration equipment, at
a constant feed concentration of 11% (v/v), and three different constant applied pressures: 0.2, 0.4 and 0.6 bar. Again the
linearity is good, but as is typical of constant pressure ltration the intercept values are highly scattered and the trend
with pressure (for a constant value of Rm ) is not as would be
expected from Eq. (4). It did not help to plot the initial data
on an expanded scale, as it was very randomly scattered, as is
often the case in constant pressure ltration.
Both sets of data, together with a number of other tests,
were used to determine the relation between specic resistance and pressure. The log plot for the constant rate tests is
Please cite this article in press as: Mahdi, F.M., Holdich, R.G., Laboratory cake ltration testing using constant rate. Chem. Eng. Res. Des. (2012),
http://dx.doi.org/10.1016/j.cherd.2012.11.012
ARTICLE IN PRESS
Fig. 5 Measured ltration pressure during ltration at three starting concentrations: 11%, 15% and 21% (v/v) with 10 rpm
pump speed.
shown in Fig. 7, where the specic resistance and pressure values plotted were determined by the end point of the constant
rate ltration data: i.e. the last recorded values for specic
resistance with pressure using the rearranged form of Eq. (5):
av =
A
cV
PA dt
dV
Rm
(8)
Fig. 6 Constant pressure ltration of calcite at 11% (v/v) starting concentration and three total applied pressures for
analysis by parabolic rate law.
Please cite this article in press as: Mahdi, F.M., Holdich, R.G., Laboratory cake ltration testing using constant rate. Chem. Eng. Res. Des. (2012),
http://dx.doi.org/10.1016/j.cherd.2012.11.012
ARTICLE IN PRESS
Fig. 7 Effect of pressure on cake ltration resistance, initial concentrations from 9 to 21 (% v/v); numbers in the gure
indicate pump speeds 5, 10 and 30 rpm.
Constant rate
Constant pressureb
a
b
0 (m kg1 )
Co (v/v)
6.78 10
5.53 109
0.19
0.18
0.25
0.08
In Fig. 8 the specic resistance at each time interval is calculated, according to Eq. (8), and this is compared to the specic
resistance calculated using Eq. (1), where the cake forming
pressure is used. In order to determine the cake forming pressure the lter medium resistance was assumed to be constant
and equal to the value determined from Fig. 5. This gave a
value of 1.43 109 m1 . The agreement between the values of
specic resistance calculated by Eqs. (1) and (8) is very good.
Clearly, as an experimental technique it would be possible to
36KC
3
2
(1 C) xsv
s
(9)
where K is the Kozeny constant, a value of 5 was used as is con2 is the Sauter mean diameter of
ventionally assumed, and xsv
the particles. It is fairly common for the text book value of the
Kozeny constant to provide values of permeability, and hence
Fig. 8 The effect of pressure during constant rate ltration test on the specic cake resistance during ltration of the 21%
(v/v) feed solids at 10 rpm pump speed and theoretical resistance by KozenyCarman.
Please cite this article in press as: Mahdi, F.M., Holdich, R.G., Laboratory cake ltration testing using constant rate. Chem. Eng. Res. Des. (2012),
http://dx.doi.org/10.1016/j.cherd.2012.11.012
ARTICLE IN PRESS
chemical engineering research and design x x x ( 2 0 1 2 ) xxxxxx
Fig. 9 The percentage of total ltration resistance due to the lter medium as a function of total applied ltration pressure
and time.
specic resistance, that are an order of magnitude lower resistance (higher permeability) than are measured. It is notable
that this occurs even here when using very stable and well
characterised solids, as evidenced by the data illustrated in
Fig. 2. Hence, there is a signicant need for laboratory testing when designing ltration systems and the constant rate
system, described here, provides a convenient and thorough
technique for such analysis. By changing the Kozeny constant
to a value of 33 it was found possible to match the specic
resistance data illustrated in Fig. 8, but the text book value of
5 is signicantly in error.
The conventional technique for laboratory testing is a constant pressure ltration. Supercially, this has the attraction
of providing a xed and known pressure drop so that a limited
number of experiments, typically four, may provide values of
specic resistance and cake concentration as a function of
pressure; to provide the data for Eq. (1) and its concentration equivalent equation. However, the results are often very
scattered and one reason for this is due to the lack of constant pressure conditions. Even if the total applied pressure
remains constant, the pressure drop over the lter cake will
vary during an experiment and hence so will the pressures
that should be used in Eq. (1); where cake forming and not
total pressure should be used. At the start of the constant
pressure ltration the total pressure drop is over the lter
medium and zero pressure is over the lter cake. During the
initial stages the proportion of the pressure drop over the cake
rises from zero to a nite amount. After a short while it is usually assumed that the total pressure drop is now acting over
the cake but between the start of the ltration, and the end, the
pressure forming the cake will be changing. Hence, even during constant pressure ltration the cake forming pressure is
varying. The point of varying resistance over the lter medium
is illustrated for the constant rate ltration data investigated
here in Fig. 9. At the start the percentage of the total pressure drop due to the lter medium is 100%, but this quickly
drops down to a negligible amount. The inset graph shows
that this is a period during which the applied pressure rises to
30 mbar, equal to 200 s. Hence, over the rst 200 s the cake concentration would be expected to be changing rapidly. During
constant pressure ltration this stabilisation period is likely
4.
Conclusions
Please cite this article in press as: Mahdi, F.M., Holdich, R.G., Laboratory cake ltration testing using constant rate. Chem. Eng. Res. Des. (2012),
http://dx.doi.org/10.1016/j.cherd.2012.11.012
10
ARTICLE IN PRESS
chemical engineering research and design x x x ( 2 0 1 2 ) xxxxxx
is not so prevalent. However, it is possible to induce ltration by means of a simple laboratory peristaltic pump and to
very precisely monitor the mass of ltrate with time using
a balance. Using this procedure reliable data points at the
start of the ltration can be determined, providing a reliable value for the lter medium resistance. The robustness
of this data is superior to what is normally obtained from the
alternative constant pressure ltration data. Even in the case
of constant total pressure ltrations, the pressure difference
going to form the lter cake will vary; as the proportion of
the total pressure over the lter medium varies from 100%
at the start of the ltration to a much lower value towards
the end of the ltration. The actual pressure forming the cake
can only be calculated knowing the lter medium resistance,
which is notoriously difcult to achieve when using a constant pressure ltration with small samples of slurry and lter
areas.
Once it has been reliably established, the medium resistance can be assumed to remain constant during the ltration
and the values of ltration rate at any instance in time will
provide the pressure drop over the lter medium from Darcys
law. Hence, by deduction from the total pressure drop the
value of pressure over the lter cake can be calculated at any
instance in time. Likewise, it is possible to deduce the average
specic resistance to ltration at any instance and this can be
plotted against the prevailing cake forming pressure. Hence,
a constant rate ltration provides much more detailed information than the alternative constant pressure ltration, for
the purposes of ltration analyses. Mass balancing tests, such
as measurement of the formed cake height, and its variability
over a number of different locations, can also be used to check
on the reliability of the information obtained.
The range over which the pressure may be varied is, of
course, limited by the vapourisation of the liquid. This is not a
signicant problem when ltering water based suspensions
at normal temperatures encountered in the laboratory, but
ashing is a known problem with vacuum ltration systems
when conducting constant pressure ltration tests using a
vacuum pump, and even vacuum ltration equipment. However, it is worth noting that the most noticeable variation of
specic resistance with pressure is found within the lower
pressure range; as pressure increases, the change in resistance
becomes less pronounced as would be expected of a power
law equation such as that shown in Eq. (1). Hence, the most
relevant region to investigate is the lower pressure one.
Acknowledgements
The author FMM is grateful to the Libyan Ministry of Higher
Education and Sirte University, Libya for sponsoring this work.
References
Chi, T., 2006. Introduction to Cake Filtration Analysis,
Experiments and Applications. Elsevier B.V., The Netherlands,
ISBN 0-444-521 56-9.
Couper, J.R., Fair, J.R., Penney, W.R., Walas, S.M., 2010. Chemical
Process Equipment: Selection and Design, 3rd ed. Elsevier Inc.,
Oxford, UK (p. 337, Chapter 11).
Di Giovanni, B.A., Mahdi, F.M., Starov, V.M., Holdich, R.G., 2012.
Particulate clusters and permeability in porous media. Chem.
Eng. Res. Des., Part (A) 90, 11681176.
Earle, R.L., Earle, M.D., 1983. Mechanical Separations Unit
Operations in Food Processing: An Introduction to the
Principles of Food Process Engineering. NZIFST (Inc.) (Chapter
10) http://www.nzifst.org.nz/unitoperations/index.htm
Greil, P., Gruber, U., Travitzky, N., Kulig, M., 1992. Pressure
Filtration of Silicon Nitride Suspensions with Constant
Filtration Rate. Materials Science and Engineering: A 151,
247254.
Holdich, R.G., 2002. Fundamentals of Particle Technology. Midland
Information Technology & Publishing, UK, ISBN 0954388100.
Iritani, E., 2003. Properties of Filter Cake in Cake Filtration and
Membrane Filtration, KONA (21), Japan.
Kotlyarov, G.G., 1976. Investigations of the relations between the
ltration constant of low-concentration suspensions and
ltration pressure and solid-phase concentration, chemistry
and materials science. Chem. Petrol. Eng. 12 (5), 416418.
Li, W., Kiser, C., Richard, Q., 2005. Development of a lter cake
permeability test methodology. In: American Filtration &
Separations Society Conferences, 19 September,
USA.
Mattsson, T., Sedin, M., Theliander, H., 2011. Zeta potential and
local ltration properties: constitutive relationships for TiO2
from experimental ltration measurements. Chem. Eng. Sci.
66, 45734581.
Nicholas, P., 1998. Cheremisinoff. In: Liquid Filtration, 2nd ed.
Butterworth-Heinemann, USA, ISBN 0-7506-7047-9.
Rushton, A., Ward, A.S., Holdich, R.G., 2000. SolidLiquid Filtration
and Separation Technology. VCH, Verlagsgesellschaft mbH,
Weinheim, Federal Republic of Germany.
Rushton, A., Matsis, V.M., 1994. Studies of Constant Rate
Filtration in Dead-End and Cross ow Modes, Filtration and
Separation. Elsevier Science Ltd, 31, p. 643.
Shpanov, N.V., 1973. Calculation of constant rate cake forming
ltration, chemistry and materials science. Invest. Calculat.
Thermophys. Process. 9 (12), 10771082.
Stickland, A.D., De Kretser, R.G., Scales, P.J., 2005. Nontraditional
constant pressure ltration. AIChE J. 51 (9.).
Svarovsky, L., 2000. Solid Liquid Separation, 4th ed. Oxford,
Butterworth-Heinemann, UK, ISBN 0 7506 4568 7.
Tarabara, V.V., Hovinga, R.M., Wiesner, M.R., 2002. Constant
transmembrane pressure vs. constant permeate ux: effect of
particle size on crossow membrane ltration. Environ. Eng.
Sci. 19 (6), 343355.
Tiller, F.M., Cooper, H.R., 1960. The role of porosity in ltration. IV:
constant pressures ltration. AIChE J. 6 (4), 595601.
Tiller, F.M., 1955. The role of porosity in ltration. Part 2:
analytical equation for constant rate ltration. Chem. Eng.
Prog. 54 (6), 282290.
Tiller, F.M., 1953. The role of porosity in ltration: numerical
methods for constant rate and constant pressure ltration
based on Kozenys law. Chem. Eng. Prog. 49 (9), 467479.
Virezub, A.I., Glazunov, V.B., Lobanova, N.N., Grenin, V.I.,
Shvartsman, I.G., 1977. Viscose ltration at constant speed
and pressure, chemistry and materials science. Fibre Chem. 9
(2), 181182.
Wakeman, R.J., Tarleton, E.S., 2005. Solid/Liquid Separation:
Scale-up of Industrial Equipment, 1st ed. Elsevier Ltd., Oxford,
UK, ISBN 1 8561 74204.
Please cite this article in press as: Mahdi, F.M., Holdich, R.G., Laboratory cake ltration testing using constant rate. Chem. Eng. Res. Des. (2012),
http://dx.doi.org/10.1016/j.cherd.2012.11.012