EKC313 Separation Processes Semester 1, 2010

Lecture #1: Filtration

What is separation process?

In chemistry and chemical engineering, a separation process is used to transform a mixture of
substances into two or more distinct products. The separated products could differ in chemical
properties or some physical property, such as size, or crystal modification or other separation
into different components. [Wikipedia]

Separation process and chemical engineering

The most conventional separation process used in Chemical Industry is distillation process. In
fact distillation technique itself is much older than the Chem. Eng. discipline. Distillation was
introduced to medieval Europe through Latin translations of Arabic chemical treatises in the 12th
century [Wikipedia]. Nevertheless, distillation play a pivotal role in establishing Chemical
Engineering as an independent discipline itself, which, differentiate our job scope from
mechanical engineers and chemist. Strictly speaking, our unique job function started from the
time when new engineering skill sets were desperately needed to handle a newly founded
industry: Refinery of crude oil. In other word, mastering separation technique really defines
chemical engineer uniqueness during that time. Ironically, this saying is almost still true in this
new millennium.

Figure 1. Larger distilling apparatus for getting cone oil from Abies Alba (European Silver Fir)
[http://chestofbooks.com/health/aromatherapy/The-Volatile-Oils-Vol2/54-Cone-Oil-From-Abies-Alba.html]

Early civilization developed techniques to (1) extract metals from ores, perfumes from flowers,
dyes from plants, and potash from the ashes of burnt plants, (2) evaporate sea water to obtain
salt, (3) refine rock asphalt, and (4) distill liquors. [Seader and Henley]

1
EKC313 Separation Processes Semester 1, 2010
Lecture #1: Filtration

Filtration
In filtration, suspended solid particles in a fluid and liquid or gas are physically or mechanically
removed by using a porous medium that retains the particles as a separate phase or cake and
passes the clear filtrate. [Geankoplis]

The objectives for performing filtration usually fall into one of the following
categories:
1. clarification for liquor purification,
2. separation for solids recovery,
3. separation for both liquid and solids recovery, and/or
4. separation aimed at facilitating or improving other plant operations.

Clarification involves the removal of relatively small amounts of suspended solids from
suspension (typically below 0.15% concentration). A first approach to considering any
clarification option is to define the required degree of purification. That is, the maximum
allowable percentage of solids in the filtrate must be established. Compared with other filter
devices, clarifying filters are of lesser importance to pure chemical process work. They are
primarily employed in beverage manufacturing and water polishing operations, pharmaceutical
filtration, fuel/ lubricating oil clarification, electroplating solution conditioning, and dry-
cleaning solvent recovery.

In filtration for solids recovery, the concentration of solids suspension must be high enough to
allow the formation of a sufficiently thick cake for discharge in the form of a solid mass before
the rate of flow is materially reduced. However, solids concentration alone is not the only
criterion for adequate cake formation. For example, an 0.5% suspension of paper pulp may be
readily cake-forming whereas a 10% concentration of certain chemicals may require thickening
to produce a dischargeable cake. [Cheremisinoff]

2
EKC313 Separation Processes Semester 1, 2010
Lecture #1: Filtration

Industrial filtration equipments

For chemical industry, the quality and quantity of feed or slurry solution handled for daily
operation varying significantly due to different application purposes. Those streams may carry a
heavy load of solid particles or just trace amount of them. Due to the wide diversity of these
encountered filtration problems, various types of filtering unit/process have been developed to
cope with different situations.

Bed filters

Figure 2. Typical bed filter.

[http://kscst.iisc.ernet.in/rwh_files/images/Sand%20bed%20filter.jpg],
[http://www.cee.vt.edu/ewr/environmental/teach/wtprimer/rapid/filter3.jpg]

• Used to handle large amount of water with relatively small amount of solids
• Coarse pieces of gravel at the bottom while fine sands at the top layer (Why?)
• The processed water is drawn out at the bottom
• Normally used to remove precipitates which are not adhere strongly to the sand and can
be easily removed by backwashing.
• This kind of filter is very popular to treat municipal water. [Geankoplis]

3
EKC313 Separation Processes Semester 1, 2010
Lecture #1: Filtration
Plate-and-frame filter presses

Figure 3. Plate and frame filter presses

[http://www.stpats.com/images2007/PF30-2.jpg], [http://www.mine-engineer.com/mining/Flow_Illus.jpg]

• The plate are covered with a filter medium/cloth

• Each plate is squeezed tightly together by a screw or a hydraulic ram.
• Solids are deposited on the cloth-covered faces of the plates while liquid passes through
the cloth, down grooves or corrugations in the plate faces and out of the press.
• Operated under a pressure of 3 to 10 atm. [McCabe and Smith]

Figure 4. Plate and frame filter in one of the refinery plant. (Left) The technician is tightening up
the filter cloth. (Right) Fully automated continuous filtration process.

4
EKC313 Separation Processes Semester 1, 2010
Lecture #1: Filtration
Continuous rotary-drum filter

Figure 5. Rotary drum filter

[http://upload.wikimedia.org/wikipedia/commons/thumb/a/a2/Rotary_vacuum-drum_filter.svg/480px-
Rotary_vacuum-drum_filter.svg.png], [http://img.directindustry.com/images_di/photo-g/rotary-drum-vacuum-filter-
384475.jpg]

• In most cases, operated at 0.1 to 2 rotations/min in an agitated slurry trough.

• A filter medium covers the face of the drum and is partly submerged in the liquid
• Vacuum and air are alternately applied through central duct as the drum rotates.
• The filter cake is removed by scraping it off with horizontal knife. [McCabe and Smith]

Principles and Basic Theory of Filtration

Figure 6. Pictorial representation of cake filtration. (a) True case scenario, and (b) Schematic
showing how mathematical analysis can be done on (a).

5
EKC313 Separation Processes Semester 1, 2010
Lecture #1: Filtration
Due to the combining effects of hydrodynamic and physicochemical factors, the study of cake
structure and resistance is extremely complex, and any mathematical description based on
theoretical considerations is at best only descriptive.
[Liquid Filtration: http://www.knovel.com/knovel2/Toc.jsp?BookID=427&VerticalID=0] (This link contains a free
liquid filtration textbook in pdf format)

The flow of filtrate through the packed bed of cake can be described by an equation similar to
Poiseuille’s (Pwah-sui-ya) Law (Assumptions: flow is laminar, viscous, incompressible and the flow
is through a constant circular cross-section with length >> diameter). Poiseuille’s equation for
laminar flow in a straight tube:

32 µνL
− ∆p = (1.1)
D2
Where ∆p is pressure drop in N/m2, ν is open-tube velocity in m/s, D is diameter in m and L is
length in m, µ is viscosity in kg/m*s. For laminar flow in a packed bed of particles, the Eq. 1.1
can be rewrite according to Carman-Kozeny relation as:

k1 µν (1 − ε ) 2 S 02 L
− ∆pc = (1.2)
ε3
Where k1 is a constant and equals to 4.17 for random particles of definite size and shape, ν is
linear velocity based on filter area in m/s, ε is void fraction or porosity of cake, L is thickness of
cake in m, S0 is specific surface area of particle m2 of particle area per m3 volume of solid
particle, and ∆pc = pa – pb is pressure drop in cake in N/m2.
The linear velocity is based on the empty cross-sectional area and is defined as:

dV / dt
ν= (1.3)
A
Where V is total volume of filtrate (m3) collected up to time t s and A is filter area in m2. The
cake thickness L may be related to the volume of filtrate V by a material balance. If cs is kg-
solid/m3 of filtrate, a material balance gives

6
EKC313 Separation Processes Semester 1, 2010
Lecture #1: Filtration
LA(1 − ε ) ρ p = cs (V + εLA) (1.4)

Substituting Eq. 1.2 and 1.3 into 1.4 to eliminated L, we obtain the final equation as:

dV − ∆pc − ∆pc
= = (1.5)
Adt k1 (1 − ε ) S 0 µcsV
2
µc V
α s
ρ pε 3
A A

Where α is the specific cake resistance in m/kg defined as:

k1 (1 − ε ) S 02
α= (1.6)
ρ pε 3
For the filter-medium resistance, we can write, by analogy with Eq. 1.5

dV − ∆p f
= (1.7)
Adt µRm
Where Rm is the resistance of the filter medium to filtrate flow in m-1 and ∆p f = pb – pc is the

pressure drop across the filter medium. When Rm is treated as an empirical constant, it includes
the resistance to flow of the piping leads to and from the filter and the filter medium resistance.

Since the cake and the filter medium are positioned in series, hence
dV − ∆p
= (1.8)
Adt  αc V 
µ  Rm + s 
 A 

Where ∆p = ∆p f + ∆pc . Eq 1.8 can be further modified to accommodate the practical ability

aspect of filtration as follow:

dV − ∆p
= (1.9)
µα
Adt c s
(V + Ve )
A
Where Ve is a volume of filtrate necessary to build up a fictitious filter cake whose resistance is
equal to Rm.

7
EKC313 Separation Processes Semester 1, 2010
Lecture #1: Filtration
The volume of filtrate V can also be related to W, the kg of accumulated dry cake solids as
follow:

ρc x
W = c sV = V (1.10)
(1 − mc x )
Where cx is mass fraction of solids in the slurry, m is mass ratio of wet cake to dry cake, and ρ is
density of filtrate in kg/m3.

 Why we even bother to rearrange those equations as such from Eq. 1.8 to 1.10??

For constant-pressure filtration done in batch process, the basis equations are (rearranged from
Eq. 1.8)

dt µαc µ
= 2 s V+ Rm = K pV + B (1.11a)
dV A (−∆p) A(−∆p)
Where Kp is in [s/m6] and B in [s/m3]

µαcs
Kp = and (1.11b)
A (−∆p )
2

µ
B= Rm (1.11c)
A(−∆p )
For constant pressure, constant α, and incompressible cake, V and t are the only variables in Eq.
1.11a:

t V
Kp
∫ dt = ∫ ( K pV + B)dV 
0 0
t=
2
V 2 + BV (1.12)

t Kp
Dividing Eq. 1.12 by V = V +B (1.13)
V 2
Where V is the total volume of filtrate in m3 collected to t secs.

8
EKC313 Separation Processes Semester 1, 2010
Lecture #1: Filtration
Example 1.1 Evaluation of Filtration Constant for Constant Pressure Filtration [Geankoplis]
Data for the laboratory filtration of CaCO3 slurry in water at 298.2 K are reported as follows at
constant pressure of 338 kN/m2. The filter area of the plate-and-frame press was A = 0.0439 m2
and the slurry concentration was 23.47 kg/m3. Calculate the constant α and Rm from the
experimental data given.

t (s) V (m3) V (m3) (t/V) (s/m3)

4.4 0.000498 0.000000
9.5 0.001000 Firstly the 0.000498 8835.341365
data are 0.001000 9500
16.3 0.001501 calculated as
24.6 0.002000 0.001501 10859.42705
t/V
34.7 0.002498 0.002000 12300
46.1 0.003002 0.002498 13891.11289
59.0 0.003506 0.003002 15356.42905
73.5 0.004004 0.003506 16828.29435
89.4 0.004502 0.004004 18346.65335
107.3 0.005009 0.004502 19857.84096
0.005009 21421.44141

The data are then plotted as t/V versus V in following figure:

9
EKC313 Separation Processes Semester 1, 2010
Lecture #1: Filtration
From the figure, the intercept is determined as B = 6414 s/m3 and the slope as Kp/2 = 3 x 106
s/m6 and hence Kp = 6 x 106 s/m6.

At 298.2 K the viscosity of water is roughly 8.9 ×10 −4 Pa*s = 8.9 ×10 −4 kg/m*s.
From Eq. 1.11b

µαcs (8.9 ×10 −4 )α (23.47)

Kp =  6 ×10 6 =
A 2 (−∆p ) (0.0439) 2 (338 ×10 3 )

 α = 1.863 ×1011 m/kg

From Eq. 1.11c

µ (8.9 × 10 −4 ) Rm
B= Rm  6414 =
A(−∆p ) (0.0439)(338 × 10 3 )

 Rm = 10.63 ×1010 m-1

10
EKC313 Separation Processes Semester 1, 2010
Lecture #1: Filtration

Range of void function:

11
EKC313 Separation Processes Semester 1, 2010
Lecture #1: Filtration
Example 1.2 Time required to perform a filtration [Geankoplis]
The same slurry used in example 1.1 is to be filtered in a plate-and-frame press having 20 frames
and 0.873 m2 area per frame. The same pressure will be used in constant-pressure filtration.
Assuming the same filter cake properties and filter cloth, calculate the time to recover 3.37m3
filtrate.

In the previous example the area A = 0.0439 m2, Kp/2 = 3 x 106 s/m6, and B = 6414 s/m3. Since
the α and Rm will be the same as before, Kp can be corrected. From Eq. 1.11b Kp is proportional
to 1/A2 and hence the new Kp with A = 0.873 m2 is:

K p = 6 × 10 6 (0.0439 / 17.46) 2 = 37.93 s/m6

The new B is proportional to 1/A from Eq. 1.11c:

0.0439
B = 6400 = 16.10 s/m3
17.46

Substituting these new values into Eq. 1.12

Kp 37.92
t= V 2 + BV = (3.37) 2 + (16.10)(3.37) = 269.7 s
2 2

12
EKC313 Separation Processes Semester 1, 2010
Lecture #1: Filtration
Equations for washing of filter cakes and total cycle time
Filtration for both solids and liquid recovery differs from filtration for solids recovery alone in
the cake building, washing and drying stages. If the filtrate is valuable liquor, maximum washing
is necessary to prevent its loss; but if it is valueless, excess wash liquor can be applied without
regard to quality [Cheremisinoff].

The washing of a cake after the filtration cycle is necessary by displacement of filtrate and by
diffusion. The amount of washing liquid should be sufficient to give the desired washing effect.
To calculate washing rates, it is assumed that the conditions during washing are the same as
those that existed at the end of the filtration. It is assumed that the cake structure is not affected
when wash liquid replaces the slurry in liquid in the cake.

For constant pressure filtration, using the same pressure in washing as in filtering, the final
filtering rate is the reciprocal of Eq. 1.11a

 dV  1
  = (1.14)
 dt  f K pV f + B
where (dV/dt)f = rate of washing in m3/s and Vf is the total volume of filtrate for the entire period
at the end of filtration in m3.

For plate-and-frame filter presses, the wash

liquid travels through a cake twice as thick and
an area only half as large as in filtering, so the
predicted washing rate is one-fourth of the final
filtration rate:

 dV  1 1
  = (1.15)
 dt  f 4 K pV f + B

[http://www.andritz.com/ep-filter_cake_washing.jpg]

13
EKC313 Separation Processes Semester 1, 2010
Lecture #1: Filtration
Another reason for dewatering the filter cake:
A question posted by Mansperger on ChemE forum at [http://www.finishing.com/146/91.shtml]

We're having problems with our sludge filter-press cake. It is not dewatering properly. Instead of
a nice solid, we're getting slime. I'd like some information about a floc or filtering aid which
would help resolve this problem.

TIA, Debra Mansperger (- Chandler, AZ)

Figure 7. Good filter cake

[http://www.goodquarry.com/images/productiontechnology/PPT_photo25_large.jpg]

Filter cloths also come in many different styles of fabric weave. Mesh opening or pore size is
also an important consideration when selecting the proper filter cloths for a particular
application. The mesh opening or pore size is determined by the number of fibers, size of the
fiber, and the type of weave. These factors in turn govern flow rates, particle retention, and the
strength of the fabric itself. As shown below, these two fabrics, having the same type of open
area, will have similar flow rates, but the fabric on the left will retain finer particles.

Figure 8. Filter cloth with different scale of permeability.

http://www.water.siemens.com/SiteCollectionImages/products/sludge_biosolids_processing/filter_press/faq1_R.jpg

14
EKC313 Separation Processes Semester 1, 2010
Lecture #1: Filtration
Example 1.3 Rate of washing and total filter-cycle time [Geankoplis]
At the end of the filtration cycle in previous examples, a total filtrate volume of 3.37 m3 is
collected in a total time of 269.7 s. The cake is to be washed by through-washing in the plate-
and-frame press using a volume of wash water equal to 10% of the filtrate volume. Calculate the
time of washing and the total filter-cycle time if cleaning the filter takes 20 mins.

For this filter, Eq. 1.15 holds. Substituting Kp= 37.93 s/m6, and B = 16.10 s/m3, and
Vf = 3.37 m3, the washing rate is as follow:

 dV  1 1
  = = 1.737 ×10 −3 m 3 / s
 dt f 4 (37 . 93)(3 . 37 ) + 16 . 10

The time of washing is then as follow for 10% of filtrate volume 0.10(3.37 m3) = 0.337 m3

0.337
t= = 194.0 s
1.737 ×10 −3

The total filtration cycle is

269.7 194.0
+ + 20 = 27.73 min
60 60

15
EKC313 Separation Processes Semester 1, 2010
Lecture #1: Filtration
Continuous filtration
In a filter that is continuous in operation, such as a rotary-drum vacuum type, the feed, filtrate
and cake move at steady, continuous rates. In a rotary drum the pressure drop is held constant for
the filtration. The cake formation involves a continual change in conditions. In continuous
filtration, the resistance of the filter medium is generally negligible compared with the cake
resistance, hence, B ≈ 0 . Integrating Eq. 1.11a with B = 0

t V
∫ dt = K ∫
0
p
0
VdV (1.16)

V2
 t = Kp (1.17)
2
where t is the time required for formation of the cake. In a rotary-drum filter, the filter time t is
less than the total cycle time tc by

t = ft c (1.18)
where f is the fraction of the cycle used for cake formation. In the rotary drum, f is the fraction
submergence of the drum surface in the slurry. Substituting Eq. 1.11b and Eq. 1.18 into Eq. 1.17
and rearranging,

1/ 2
V  2 f (−∆p) 
Flow rate = =  (1.19)
Atc  t c µαcs 

When short cycle times are used in continuous filtration and/or the filter medium resistance is
relatively large, the filter resistance term B must be included, and Eq. 1.11a becomes

V2
t = ft c = K p + BV (1.20)
2
1/ 2
 R 2  2c α (−∆p ) f 
− Rm / t c +  m 2  +  s 
V  tc   µt c 
Hence, Flow rate = = (1.21)
At c αc s

All the mathematical derivations so far are taken from [Geankopolis].

16
EKC313 Separation Processes Semester 1, 2010
Lecture #1: Filtration
Example 1.4 Filtration in a Continuous Rotary-Drum Filter
A rotary-vacuum-drum filter having a 33% submergence of the drum in the slurry is to be used
to filter a CaCO3 slurry as given in Example 1.1 using a pressure drop of 67.0 kPa. The solids
concentration in the slurry is cx = 0.191 kg-solid/kg-slurry and the filter cake is such that the kg-
wet-cake/kg-dry-cake, m = 2.0. The density and viscosity of the filtrate can be assumed as those
of water at 298.2 K. Calculate the filter area needed to filter 0.778 kg-slurry/s. The filter cake
time is 250 s. The specific cake resistance can be represented by α = (4.37 ×10 9 )(−∆p ) 0.3 , where
− ∆p is in Pa and α in m/kg.

For water ρ = 996.9 kg/m3, µ = 0.8937 x 10-3 Pa*s

ρc x (996.9)(0.191)
cs = = = 308.1 kg-solid/m3-filtrate
1 − mc x 1 − (2.0)(0.191)

Solving for α
α = (4.37 ×10 9 )(−∆p ) 0.3 = (4.37 × 10 9 )(67 × 10 3 ) 0.3 = 1.225 × 1011 m/kg. To calculate the flow rate
of the filtrate,
 
 
V cx  kg − slurry   kg − solid   1 
= 0.778 =  0.778  0.191 
tc c s  s  
kg − slurry  308.1 kg − solid 

 m − filtrate 
3

= 4.823 ×10 −4 m3-filtrate/s

Substituting into Eq. 1.19

1/ 2
V 4.823 × 10 −4  2(0.33)(67 × 10 3 
= = −3 
 250(0.8937 × 10 )(1.225 × 10 )(308.1) 
11
Atc A

Hence, A = 6.60 m2

17
EKC313 Separation Processes Semester 1, 2010
Lecture #1: Filtration
References
[Seader and Henley]
J.D. Seader & E.J. Henley John, 'Separation Process Principles' ,2nd Ed. John Wiley & Sons, 2006

[Geankoplis]
C.J. Geankoplis, 'Transport Processes and Unit Operations', 4th Ed. Prentice-Hall International, 2003.

[McCabe and Smith]

Mc Cabe W.L., J.C.Smith, P.Harriott, ‘Unit Operations in Chemical Engineering’,7th Ed. McGraw
Hill, 2005.

[Cheremisinoff]
Cheremisinoff N.P. ‘Liquid Filtration’ 1st Ed. Butterworth-Heinemann, 1998.

Good online-resources for ChemE Separation Techniques

http://people.clarkson.edu/~wwilcox/Design/refsepn.htm

18