NPTEL

Novel Separation Processes

Module :
3
Membrane based separation processes

Dr. Sirshendu De
Professor, Department of Chemical Engineering
Indian Institute of Technology, Kharagpur
e-mail: sde@che.iitkgp.ernet.in
Keywords:
Separation processes, membranes, electric field assisted separation, liquid
membrane, cloud point extraction, electrophoretic separation, supercritical fluid
extraction

Joint Initiative of IITs and IISc - Funded by MHRD

Page 1 of 120

NPTEL
Novel Separation Processes

Membrane Based Separation Processes

3.1 Definition of a membrane:
It is an interface that separates the two phases and restricts the transport of various
chemical species through it. Membrane can be homogeneous, heterogeneous, symmetric,
asymmetric, charged, neutral.

3.2 Membrane Casting:

Common polymeric membrane material is Cellulose Acetate. Various steps of a typical
casting process are:
1. Solvent (acetone) is added to polymer. A viscous solution is prepared.
2. It is put on a plate. Place another plate on top of it. The gap is of the order of
~ 0.25mm
3. Give a one directional motion to the top plate. A thin film is produced on the
bottom plate.
4. Take out bottom plate and place it in ice-water bath, which releases the film.
5. Heat treat the film in hot water.

Typical casting conditions are:

Casting solution

: CA 25%, Frmamide 30% and Acetone 45%.

Casting temperature : 250C

Evaporation time

: 1 minute

Joint Initiative of IITs and IISc - Funded by MHRD

Page 2 of 120

NPTEL
Novel Separation Processes

3.3 Types of barrier:

Homogeneous barrier:

Pore size is 2-10 A0

Diffusion is controlling mechanism.

Typical RO membrane

Microporous Barrier:

Pore size is 10-1000 A0

Both convection and diffusion are dominant mechanism.

Asymmetric Barrier:
In this case, a thin polymeric skin is cast on another polymeric support. It is shown
schematically in Fig. 3.1. The skin acts like a membrane. Since the thickness of the skin
is really small, the resistance offered is less and one can expect a higher flux compared to
homogeneous membranes.

Fig. 3.1: Schematic of an asymmetric membrane

Trans membrane permeate flux is inversely proportional to the membrane thickness.
If homogeneous barrier is the skin, micro porous barrier may be the support.

Joint Initiative of IITs and IISc - Funded by MHRD

Page 3 of 120

NPTEL
Novel Separation Processes

3.4 Types of motion of molecules through barrier:

1) Permeation:
a) Dissolution of permeating molecules in the membrane
b) Diffusion of dissolved molecules
c) Desorption of penetrant molecules to the downstream side.

2) Knudsen diffusion (d/ < 0.2):

Single gaseous molecules diffuse under rarefied conditions so that the mean free path
is longer than the pore diameter.

3) Convection (d/ > 20):

Viscous flow through the pores of ultrafiltration and microfiltration.

3.5 Categorization of such processes:

Reverse Osmosis:
Small solute particles to be separated.
Molecular weight < 100
Pore size: 2 10 A0
Pressure: > 25 atm.
Permeation is main transport mechanism
Example: Filtration of salt solution

Nanofiltration:
Particles to be separated with Molecular weight: 200 1000
Pore size: 5 20 A0
Pressure: 15 25 atm.

Joint Initiative of IITs and IISc - Funded by MHRD

Page 4 of 120

NPTEL
Novel Separation Processes

Particle retention of salts.

Example: Filtration of dyes, small molecular weight organics, etc.

Ultrafiltration:
Molecular weight of particles : 103 - 105
Pore size: 20 1000 A0
Pressure: 6 8 atm.
Transport Mechanism: Convection (main) + diffusion
Example: Filtration of protein, Red blood cells, polymers, etc.

Microfiltration:
Molecular weight > 1 lakh
Pore size: more than 1000 A0
Pressure: 2 4 atm.
Example: Filtration of clay solution, latex, paint, etc.

3.6 Useful definitions:

Some useful definitions are presented below.

(1) Osmotic Pressure:

Osmosis means craving for water.

h
Solution

H2O

Joint Initiative of IITs and IISc - Funded by MHRD

gh

Osmotic
pressure

Page 5 of 120

NPTEL
Novel Separation Processes

Fig.3.2: A schematic presentation of the osmosis process

Fig. 3.2 shows that a semipermeable barrier separates a solution having some solute in
one side and pure solvent (water in this case) at the other side. Due to presence of
difference in concentration of solvent in both sides, water from solution side permeate
through the membrane to the solution side. Therefore, the level of water in solvent side
goes down and that in the solution side goes up. At equilibrium, there exists a
difference in hydrostatic pressure in both the chambers, although the chemical potential
of the solvent is equal across the membrane. The developed hydrostatic head is called
osmotic pressure. If the solute concentration in the solution is more, the solution is
more thirsty of water and equilibrium hydrostatic pressure head is more.
When water is forced through a concentrated solution (by applying pressure), the
osmotic pressure resists. The energy required to overcome this resistance is called
osmotic energy.
As discussed above, osmotic pressure is a colligative property, directly proportional to
concentration.

1
Mw

and

and hence, =

RTC
Mw

(3.1)

The above equation is vant hoffs equation for dilute solutions.

It may be mentioned that any solution has its own osmotic pressure, but it can only be
realized in presence of a semipermeable barrier or membrane. For example, a glass of
saline water has its own osmotic pressure but it can only be realized, if we place it across
a semi-permeable barrier.

Joint Initiative of IITs and IISc - Funded by MHRD

Page 6 of 120

NPTEL
Novel Separation Processes

(2) Observed retention: (Selectivity of membrane)

This property indicates the extent of separation of a membrane can produce with
respect to the solute concentration in the feed. Thus, observed retention is defined as,
Ro = 1

Cp
Co

(3.2)

where, C p = Solute concentration in permeate

Co = Solute concentration in feed

If Ro 1.0, solute is completely retained by the membrane.

(3) Real retention:

Real retention is a constant that defines the partition of the solute concentration
across the membrane, i.e., between the the membrane-solution interface and the
permeate side. Since, this definition is not masked by any physical phenomenon like
concentration polarization, (defined later) etc., this definition indicates the true
separation efficiency of the solute by the membrane.
Rr = 1

Cp
Cm

(3.3)

Here, Cm = Solute concentration in membrane solution interface . It may be mentioned

here that since, membrane surface concentration of solute is always greater than the bulk
concentration, real retention is always greater than observed retention. For complete
solute retention, Rr = 1.0 .

(4) Molecular Weight Cut-Off (MWCO):

Joint Initiative of IITs and IISc - Funded by MHRD

Page 7 of 120

NPTEL
Novel Separation Processes

Molecular weight cut off is another concept to characterize a membrane. In this case,
generally neutral solutes of various molecular weights are considered to conduct
experiments. The operating conditions of these experiments are low transmembrane
pressure drop, high turbulence and low feed concentration. Experiments are conducted
using each of these solutes and the observed retention values at the steady state are
measured. The observed retention values are then plotted against the molecular weight of
the solutes in a semi-log plot. The typical solutes are glucose (molecular weight 180),
sucrose (molecular weight 342), various fractions of polyethylene glycol (molecular
weights: 200, 400, 600, 1000, 1500, 2000, 4000, 60000, 10000, 30000), dextran
(molecular weight 40000, 1,50,000), etc. The molecular weight at 90% solute retention
indicates roughly the molecular weight cut off of the membrane. The molecular weight
cut off curves are shown in Figs. 3.3 and 3.4

1.0
0.90

R0

Log MW

Fig.3.3: Typical molecular cut off curve of a membrane

Joint Initiative of IITs and IISc - Funded by MHRD

Page 8 of 120

NPTEL
Novel Separation Processes

Sharp cut off

R0

Diffused cut off

membrane

Log MW

Fig.3.4: Sharp and diffused molecular cut off curves

Molecular weight cut off curves may be a sharp cut off or a diffused cut off. If the
retention curve raises sharply to 90% level over a small span of molecular weight regime,
then the cut off curve is called a sharp cut off curve. If the retention curve rises over a
wide span of molecular weight region, it is a diffused cut off curve. One has to have an
accurate control over the operating conditions to achieve the sharp cut off membranes. In
fact, most of the commercial membranes are diffused type.

(5) Membrane Permeability (Lp):

This parameter shows how porous the membrane is. If Lp is more, then the
membrane is more porous. Mathematically, Lp is defined as, L p =

J0
, where, J0 is the
P

pure water flux and P is transmembrane pressure drop. This concept is elaborated in
detail in the subsequent sections.

Joint Initiative of IITs and IISc - Funded by MHRD

Page 9 of 120

NPTEL
Novel Separation Processes

(6) Estimation of retention & permeability:

Retention:
Observed retention (Ro): Estimate by direct experimental measurement
Real retention (Rr): One has to conduct batch experiments at high stirring speed, low
feed concentration and low operating pressure. In that case, it is assumed that there is
no formation of concentrated solute layer over the membrane surface and in absence of
polarized layer, observed retention is almost same as retention retention.

Permeability:
Membrane permeability is measured by distilled water runs. Experiments are conducted
using distilled water at various transmembrane pressure drops values. At various
pressure drops, the water flux is measured. A plot of permeate flux versus operating
pressure would be a straight line through the origin, as shown in Fig. 3.5. The slope of
this curve indicates the permeability (Lp) of the membrane. I may be noted here that the
permeability of a membrane is its pressure history only. In other words, permeability of
the membrane is independent of turbulence (stirring speed or cross flow velocity) in the

m3
m
or
.
flow channel. The unit of permeability is 2
m .Pa.s
Pa.s

Joint Initiative of IITs and IISc - Funded by MHRD

Page 10 of 120

NPTEL
Novel Separation Processes

J
3

(m /m2.s)
Slope=Lp

P
Fig.3.5: A typical flux versus pressure plot for ditilled water as feed

Membrane modules:
The practical equipment where the actual membrane based separation occurs is known
as membrane modules. The basic aim of development of these modules is to provide
maximum membrane area in relatively smaller volume, so that the permeate flux i.e.,
the productivity of the system is maximum. These membrane modules are of four types,
(i) plate and frame module, (ii) hollow fiber module, (iii) spiral wound and (iv) tubular
modules. Each of these modules is described below:

Plate and frame modules

The heart of plate-frame module is the support plate that is sandwiched between two
flat sheet membranes. The membranes are sealed to the plate, either gaskets with
locking devices, glue or directly bonded. The plate is internally porous and provides a
flow channel for the permeate which is collected from a tube on the side of the plate.
Ribs or grooves on the face of the plate provide a feed side flow channel. The feed
chanel can be a clear path with chanel heights from 0.3 to 0.75 mm. The higher channel

Joint Initiative of IITs and IISc - Funded by MHRD

Page 11 of 120

NPTEL
Novel Separation Processes

heights are necessary for high viscosity feeds; reduction in power consumption of 20 to
40% can be achieved by using a 0.6 mm chanel compared to a 0.3 mm chanel.
Alternatively, retentate separator screens (20 or 50 mesh polypropelyne) can be used.
Commercial plate-frame units are usually horizontal with the membrane plates mounted
vertically. They can be run with each plate in parallel plates in two or three series.
Laboratory are also available as preformed stacks up to 10 plates. A typical plate and
frame module is shown in Fig. 3.5a.

Fig.3.5a: A plate and frame module

Tubular modules
In such modules, the membrane is cast on the inside surface of a porous tube. Tubular
membranes operate in tangential, or cross-flow, design where process fluid is pumped
along the membrane surface in a sweeping type action. The feed solution is pumped
through the center of the tube at velocities as high as 6 m/s. These cross-flow
velocities minimize the formation of a concentration polarization layer on the

Joint Initiative of IITs and IISc - Funded by MHRD

Page 12 of 120

NPTEL
Novel Separation Processes

membrane surface, promoting high and stable flux and easy cleaning, especially when
the objective is to achieve high suspended solids in the MF, UF or NF concentrate.
Permeate is driven through the membrane to be directed out of the system or back
into the process depending on the application. There are many advantages in tubular
membrane configurations. Besides their rugged construction, they have a distinct
advantage of being able to process high suspended solids, and concentrate product
successfully and repeatedly to relatively high end point concentration levels without
plugging. A common objective of an end-of-pipe waste treatment UF system is to
reduce waste volume as much as possible to reduce concentrate hauling costs. For
juice clarification applications, tubular membrane systems produce the greatest
yields and the highest final suspended solids concentration levels. Tubular MF, UF
and NF systems do not require significant prefiltration. Some tubular products have
the ability to be mechanically cleaned with spongeballs. Spongeballs can be used in
process, and are also used to enhance chemical cleaning by reducing time and
cleaning chemicals. Tubular membranes are ideally suited to treatment of
metalworking oily waste, wastewater minimization and recovery from industrial
processes, juice clarification, treatment of pulp and paper industry waste, etc. Tubular
membranes typically have life upto 2 to 10 years. The following figure (Fig. 3.5 b)
shows some tubular membranes.

Joint Initiative of IITs and IISc - Funded by MHRD

Page 13 of 120

NPTEL
Novel Separation Processes

Fig.3.5b: Pictures of tubular modules

Hollow fiber module

In hollow fiber module, lots of hollow fibers (each fiber is a tubular module) are
kept in a large pipe. Geometry allows a high membrane surface area to be
contained in a compact module. This means large volumes can be filtered, while
utilizing minimal space, with low power consumption. Hollow fiber membranes
can be designed for circulation, dead end and single pass operation. Some of the
many hollow fiber membrane filtration applications include, potable water
treatment, juice clarification, wine filtration, dairy processing, etc. The advantages
of such modules include reduction in space requirement, lowering in labor cost,
lowering in chemical cost, delivery of high quality product water, etc. Hollow
fiber membranes offer the unique benefits of high membrane packing densities,
sanitary designs and, due to their structural integrity and construction,
can withstand permeate back pressure thus allowing flexibility in system design
and operation. Most KMS hollow fiber products are available in (i) 1" diameter
laboratory test cartridges ranging up to 10" diameter for commercial products,
(ii) Standard commercial cartridge lengths of 25", 43", 48", 60" and 72",
(iii) Nominal separation ranges from 0.2 micron down to 1,000 MWCO, (iv)
Joint Initiative of IITs and IISc - Funded by MHRD

Page 14 of 120

NPTEL
Novel Separation Processes

Fiber inside diameters from 0.02"(.5mm) up to 0.106"(2.7mm), (v) Various

materials of construction including polysulfone and polyacrylonitrile.
The following figure (Fig. 3.5c) shows some hollow fiber cartridges of 5, 8 and
10 diameter with endcaps.

Fig.3.5c: Pictures of hollow fiber modules with end cap

Benefits of hollow fiber membranes include (i) controlled flow hydraulics, (ii) tangential
flow along the membrane surface limits membrane fouling, (iii) membranes can be
backflushed to remove solids from the membrane inside surface, thus extending the time
between two chemical cleaning cycles, (iv) high membrane packing density resulting in
high throughput, (v) modular in structure so that future extension of the plant becomes
easier. The flow pattern in a typical hollow fiber module takes place as shown in Fig.
3.5d.

Joint Initiative of IITs and IISc - Funded by MHRD

Page 15 of 120

NPTEL
Novel Separation Processes

Fig.3.5d: Flow pattern in hollow fiber module

Spiral wound module

In spiral wound membrane, membrane is cast as a film onto flat sheet. Membranes are
sandwiched together with feed spacers (typical thickness 0.03 to 0.1 inch) and permeate
carrier. They are sealed at each edge and wound up around a perforated tube. The
module diameter ranges from 2.5 to 18 inch and length varies from 30 to 60 inch. The
typical cross section of the spiral wound module is shown below:

Joint Initiative of IITs and IISc - Funded by MHRD

Page 16 of 120

NPTEL
Novel Separation Processes

Fig.3.5e: Flow pattern in spiral wound module

The application of spiral wound module includes, seawater desalination, brackish water
treatment, potable water treatment, dairy processing, electrocoat paint recovery, protein
separation, whey protein concentration, etc.
Therefore, it can be identified that the modeling of plate and frame and spiral
wound module can be done by considering the flow through a rectangular channel. On
the other hand, that for a tubular and hollow fiber module are done by considering flow
through a tube.

Joint Initiative of IITs and IISc - Funded by MHRD

Page 17 of 120

NPTEL
Novel Separation Processes

3.7 Driving forces in membrane separation processes:

Separation in membrane is the result of differences in the transport rates of chemical
species through it.
Transport rate is determined by the driving force acting on individual components, their
mobility, concentration of solute in membrane phase, etc..
Mobility

: Depending on solute size and structure of membrane.

Concentration : Chemical compatibility of solute & interface material.

3.8 Basic forces of transport:

Driving forces for transport:
Gradient of electro chemical potential of species is the driving force of transport of
species. This gradient may be caused by pressure difference, concentration, temperature
or electrical potential between two phases separated by the membrane.

(a) Passive Transport:

As described in Fig. 3.6, in such cases, the upstream chemical potential of a component
is more than that in the downstream.

A'
A' > A''

A''

Fig.3.6: Passive transport

Joint Initiative of IITs and IISc - Funded by MHRD

Page 18 of 120

NPTEL
Novel Separation Processes

(b) Facilitated transport:

A'

A' > A''

AB

A''

A + B AB

Fig.3.7: Facilitated transport

In this case also the chemical potential of a species in upstream is more than that
in the downstream. However, as shown in Fig. 3.7, components to be transported are
coupled with a carrier in the membrane phase. So, it is a special form of passive
transport and very selective and at the same time, the transport is facilitated by the
carrier component.

(c) Active transport:

A''
A

AB
A

A'

A + B AB

Fig.3.8: Active transport

Joint Initiative of IITs and IISc - Funded by MHRD

Page 19 of 120

NPTEL
Novel Separation Processes

As shown in Fig. 3.8, components are transported against driving force. Driving
force for transport is provided by the activation energy of chemical reaction in the
membrane phase (Living Cell). In this case, the upstream chemical potential is higher
than that of downstream potential.

3.9 Description of transport process by phenomenological equation

The transport processes are generally described by the phenomenological equations.
Such equation describes cause-effect relationship. It presents flux of anything is
proportional to the driving force. The proportionality constant is the transport property
specific to the system. Table 3.1 presents various phenomenological equations common
in chemical engineering applications.
Table 3.1: Examples of phenomenological equations
Phenomenon

Constant of
Flux

Driving force
property

relation

Ficks Law

Mass

Ohms Law

Electricity

Concentration difference

Electric potential
R

difference
Fouriers Law

Heat

Temperature difference

In membrane separation process, driving forces may be interdependent. Thus, a

concentration gradient across a membrane not only leads to flow of matter, can also cause
build up of pressure. An example is osmosis.

Joint Initiative of IITs and IISc - Funded by MHRD

Page 20 of 120

NPTEL
Novel Separation Processes

A pressure difference leads to volumetric flow as well concentration difference. An

example is reverse osmosis.
Temperature difference across the membrane leads to heat flow associated with flow of
matter. An example is Thermo diffusion/ Soret effect.
Concentration difference across the membrane causes mass flux as well as temperature
gradient. An example is Dufour effect.

3.10 Osmotic pressure: A thermodynamic point of view

Let us consider an experiment, where, a solvent and a solution are separated by a semipermeable membrane as shown in fig. 3.9.

B
Solution

Solvent

Fig. 3.9: Solution and solvent are separated by a semi-permeable barrier

Osmotic equilibrium means chemical potential of solvent at both chambers are

equal.
Let us use the notation, 2 is for solvent and 1 is for solute. P1 is pressure in
chamber A and P2 is pressure in chamber B.

At osmotic equilibrium:
2 ( at P1 ) = 2 ( at P2 )

(3.4)

The above equation can be written as,

2 = 20 + RT ln a2
Joint Initiative of IITs and IISc - Funded by MHRD

(3.5)

Page 21 of 120

NPTEL
Novel Separation Processes

Where, 20 is for pure solvent and a2 is solvent activity.

Following hypothetical experiment is conducted.

Hypothetical Experiment:
Both chambers have pure water and initially at pressure P1, 2 is same at both chambers.
Now small amount of solute is added in chamber B.
Then, 20 ( P1 ) becomes 2 ( P1 ) in B

2 ( P1 ) = 20 ( P1 ) + RT ln a2

(3.6)

In B, apply pressure and so that P1 becomes P2 and osmotic equilibrium is restored.

At osmotic equilibrium,

2 ( at P1 ) in 'B' = 2 ( at P2 ) in 'A'

(3.7)

The above equation can be written as,

20 ( P2 ) + RT ln a2 = 20 ( P1 )

(3.8)

Rearrangement of above equation leads to

20 ( P2 ) 20 ( P1 ) = RT ln a2

(3.9)

From the definition of partial molar volume, we get,

i0
Vi = Partial molar volume of i species =
. Thus, Eq.(3.9) is written as,
P T ,n
th

P2

( P2 ) ( P1 ) = V2 dP
0
2

0
2

(3.10)

P1

Assuming, V2 V2 ( P ) , Eq.(3.10) is expressed as,

V2 ( P2 P1 ) = RT ln a2

(3.11)

We, thus, obtain an expression of osmotic pressure,

Joint Initiative of IITs and IISc - Funded by MHRD

Page 22 of 120

NPTEL
Novel Separation Processes

V2 = RT ln a2

(3.12)

Here, a2 = Solvent activity

=

Solvent vapour pressure inpresence of solute

Pure solvent vapour pressure

PS
PS*

So,

RT PS*
ln
V2 PS

(3.13a)

The above equation is presented in the following more amenable forms:

Vant-Hoff equation

: = CRT for dilute solutions

(3.13b)

Virial Type equations : = B1C + B2C 2 + B3C 3 for concentrated solution.

(3.13c)

3.11 Transports in Reverse Osmosis (RO):

Transport equation for the solute and solvent through reverse osmosis membrane
is derived from irreversible thermodynamics.
The assumptions involved are,
(a)

Flux of ith species may be a function of every force that exists in the system.

(b)

Under conditions not too far from equilibrium the flux-force relation is linear.
nf

N i = Lij Fj

(3.14)

j =1

Here,

N i = Molar flux of i th species in a stationary frame of reference

Fj = jth driving force
Lij = Phenomenological coefficient

Joint Initiative of IITs and IISc - Funded by MHRD

Page 23 of 120

NPTEL
Novel Separation Processes

(c)

No other external/mechanical force is applied on any of the species

The notations are, 1 is for solute and 2 is for solvent. Thus, the fluxes for the solvent
and the solute are expressed as,
N 2 = L211 L222

(3.15)

N1 = L111 L122

(3.16)

Onsagers reciprocating relationship:

If fluxes & forces are properly selected, then, Lij = L ji

Coupling:
The addition to the flux of one species due to the existence of force acting on another
species is termed as coupling. For RO, it is assumed that no coupling takes place between
solute and solvent flow.
Therefore, L12 = L21 = 0 . Therefore, the solvent and solute fluxes are expressed as,
N 2 = L222 ;

(3.17)

N1 = L111

(3.18)

The gradient of chemical potential in absence of any electric potential gradient is

expressed as,


i = i P + i Ci
P T ,n
Ci P ,T

(3.19)

Therefore, the solvent flux is expressed as,

N 2 = L22 2 P +
C

2m
P T ,n

C2 m P ,T

Joint Initiative of IITs and IISc - Funded by MHRD

(3.20)

Page 24 of 120

NPTEL
Novel Separation Processes

Where, C2 m is molar concentration of solvent in membrane .

At the steady state, one dimensional transport across the membrane of thickness l is
shown in Fig. 3.10.

''
C2m

'
C2m

x=0

x=1

Fig. 3.10: Steady state, one dimensional transport across the membrane.

Integrating Eq.(3.20) across the membrane thickness, the following expression

is obtained.
P1
C2 m

2
2
=

N
dx
L
dC
dP

22
2m
0 2

P n,T
C2' m C2 m P ,T
P2

''

(3.21)

Where, P2 is higher, upstream pressure and P1 is lower, downstream pressure. At steady

state, N2 is constant. The definition of partial molar volume is,

Vi = Partial molar volume of i th species = i . Using this definition of
P T ,n

partial molar volume, the integration of above equation results,

P1
C2 m

2
2
+

N 2l = L22
dC
dP

2m

P n,T
C2' m C2 m P ,T
P2

''

So,

(3.22)

At osmotic equilibrium, N2 = 0 (from closure equation). Assuming, V2 f ( P ) over the

range of pressure, the following expression is obtained.
V2 ( P2 P1 ) =

C2'' m

2
dC2 m
2 m P ,T
C2' m

Joint Initiative of IITs and IISc - Funded by MHRD

Page 25 of 120

NPTEL
Novel Separation Processes

= V2

(3.23)

Combining Eqs.(3.22) and (3.23), the following equation of the solvent flux is obtained.
P2

N 2l = L22 V2 V2 dP
P1

= L22 V2 V2 P =

L22V2
[ P ]
l

(3.24)

The above equation is re-written as,

N 2 = ksol [ P ]

(3.25)

Where, ksol is a constant related to membrane permeability.

Thus, the phenomenological equation of the solvent flux through reverse osmosis
membrane is derived.

Solute flux through the membrane:

The assumptions involved are:
(a) Steady state
(b) One dimensional
(c) No coupling
Therefore, integration of Eq.(3.16) results into,
P1
C1m

1
1
=

N
dx
L
dC
dP

11
1m
0 1

P n,T
C1' m C1m P ,T
P2

''

= L11

For dilute solution,

C1''m

1
dC1m + V1 L11P
'
1
m
P ,T
C1 m

(3.26)

a1m C1m and the expression of chemical potential is,

= 0 + RT ln a1m
Joint Initiative of IITs and IISc - Funded by MHRD

Page 26 of 120

NPTEL
Novel Separation Processes

= 0 + RT ln C1m

(3.27)

The derivative of chemical potential with concentration becomes,

ln C1m

= RT
C1m
C1m T

(3.28)

Combination of Eqs.(3.26) to (3.28) results into following expression of solute flux,

through the membrane,
N1l = L11 RT ln

N1 =

C1' m
+ V1 L11P
C1''m

L11 RT C1' m V1 L11

ln '' +
P
l
C1m
l

In most of the cases,

(3.29)

V1 L11
P 0
l

Thus, expression of solute flux becomes,

N1 =

L11 RT C1' m
ln ''
l
C1m

For dilute solution, L11

(3.30)
D1m C1m ,avg . Thus, solute flux expression becomes,

1
N1 = L11
C1m
C1m

= D1m 1 C1mC1m
C1m

d ln C1m
dC1m
= D1m RT
C1m
dC1m
dx

= D1m RTC1m

d ln C1m
dx

Joint Initiative of IITs and IISc - Funded by MHRD

(3.31)

Page 27 of 120

NPTEL
Novel Separation Processes

The final expression of solute flux hence becomes,

N1l = D1m RT

C1''m

dC

1m

= D1m RT C1''m C1' m

C1' m

N1 =

D1m RT '
C1m C1''m
l

N1 = K S C1' m C1''m

(3.32)

Where, KS = solute permeability, which is depending on solute diffusivity in membrane

matrix. The model of solute flux through the membrane presented by Eq.(3.32) is
known as solution-diffusion model.
It may be noted that the solvent flux N2 presented above is molar flux with the unit

moles
. This can be converted to volumetric flux (J), m3/m2.s. Following equation
2
m .s
presents the conversion.
J =

N 2 M 2W 103

(3.33)

solution

The solute flux in Eq. (3.32), is therefore presented as,

JC1''m = K s C1' m C1''m

(3.34)

Combining the above equation with the definition of observed retention, the permeate
flux can be expressed in terms of retention.
1

1
J = B
1 R0

(3.35)

The solvent flux, in terms of volumetric flux and membrane permeability (Lp), can be
expressed from Eq.(3.25),
J = L p ( P )

Joint Initiative of IITs and IISc - Funded by MHRD

(3.36)

Page 28 of 120

NPTEL
Novel Separation Processes

Combining the above two equations, the observed retention can be expressed in terms of
operating pressure.
R0 =

Lp ( P )

(3.37)

L p ( P ) + B

For high values of P, R0 approaches to one. (R0 1.0). Physically, as P increases,

solvent flux will increase due to increase in driving force. So, solute becomes diluted and
permeate concentration becomes extremely small and observed retention tends to 1.

3.11.1 Solution-diffusion imperfection model

In case of solution diffusion imperfection model, the solution diffusion model is modified
to incorporate the convective effects, i.e., the effects due to pressure drop across the
membrane. Therefore, the solute and solvent flux through the membrane are written as,
N 2 = k1 ( P ) + k2 P

(3.38)

N1 = k3 C1' C1'' + k4 P

(3.39)

3.12 Concept of concentration polarization and membrane fouling

3.12.1 Concentration polarization
Accumulation of solute particles as shown in Fig. 3.11 over the membrane surface is
defined as concentration polarization. It has the following effects
(i)

Increase in osmotic pressure of the solution.

(ii)

Formation of gel over the membrane surface.

(iii)

Increases the viscosity of the solution.

(iv)

Solute enters into the pores and pores are blocked partially or completely.

Joint Initiative of IITs and IISc - Funded by MHRD

Page 29 of 120

NPTEL
Novel Separation Processes

First phenomena decrease in driving force. Second and third increases the
resistance against flux. Fourth decreases the membrane permeability. All these effects
lead to decrease in permeate flux.
C = C0

Boundary layer

C = Cm
Fig. 3.11: Schematic of concentration polarization

Concentration polarization cannot be avoided, it can only be minimized.

3.12.2 Fouling of membrane:

Fouling of membrane is of two types reversible and irreversible.

Reversible fouling:
It can be washed away by adopting an appropriate cleaning protocol, like membrane
washing. After cleaning, membrane permeability is restored. Concentration polarization
is reversible fouling.

Irreversible fouling:
In this case, membrane pores are blocked permanently and they cannot be removed, even
after proper washing. Permeability is lost permanently.

3.13 Film theory

It must be clear by now that in any membrane separation process, there exists a
concentration boundary layer between the bulk of the solution and the membrane surface,

Joint Initiative of IITs and IISc - Funded by MHRD

Page 30 of 120

NPTEL
Novel Separation Processes

that is termed as concentration polarization as discussed in the preceding section. Film

theory is used to quantify this concentration polarization or expresses the permeate flux
in terms of solute concentration at the membrane surface, bulk and the permeate stream.
At the membrane surface, three solute fluxes can appear. Convective flux towards the
membrane (J C); diffusive flux away from the membrane (-D

dC
) and convective flux
dy

away from the membrane (J Cp). A solute mass balance at the steady state results,

J (C C p ) + D

dC
=0
dy

(3.40)

The main assumption in film theory is consideration of constant thickness or film

of the mass transfer boundary layer. The above equation must satisfy the conditions at y=
0 (at the membrane surface), C= Cm. At the edge of mass transfer boundary layer, i.e., at
y= , C=C0. The solution of Eq.(3.40) is given as,

Cm C p

J
= exp
C0 C p
k

(3.41)

In the above equation, k is mass transfer coefficient and is expressed as k = D / .

Mass transfer coefficient

We can find out the mass transfer coefficient (k), from the appropriate Sherwood number
relation. Leveque relation is valid for the laminar flow through a conduit.
For tube,
1

kd
d 3

= 1.62 Re Sc
Sh =
D
l

for Re < 2100

Joint Initiative of IITs and IISc - Funded by MHRD

(3.42)

Page 31 of 120

NPTEL
Novel Separation Processes

Where, Re =

u0 d

; Sc =

For flow through rectangular conduit,

(3.43)

kd
d 3

Sh = e = 1.85 Re Sc e
D
l

Where, Re =

for Re < 2100

u0 d e

; Sc =

For turbulent flow, Dittus Boelter relation is used.

Sh = 0.023 ( Re )

0.8

( Sc )

0.33

(3.44)

In case of stirred cell,

Sh =

kr
0.55
0.33
= 0.285 ( Re ) ( Sc )
D

(3.45a)

In the above relation, Reynolds number is defined on stirrer speed ( ) in rad/s as,

r 2
Re =
. The above relation is valid when Re<32000. For Re>32,000, the following

relation is used.
Sh =

kr
0.8
0.33
= 0.0443 ( Re ) ( Sc )
D

(3.45b)

In the above equation, r is the radius of the stirred cell.

3.14 Cross flow filtration

It may be noted here that the concentration polarization cannot be avoided
altogether. It can only be minimized. The popular way to reduce concentration
polarization is imparting more turbulence in the flow channel or filtration unit. In a

Joint Initiative of IITs and IISc - Funded by MHRD

Page 32 of 120

NPTEL
Novel Separation Processes

stirred cell, this can easily be done by increasing stirring. For a continuous system, the
feed is allowed to flow tangentially over the membrane surface and the growth of
concentration boundary layer is arrested. Since the direction of feed flow is normal to the
direction of the permeate flow, the system is defined as a cross flow system.

3.15 Prediction of system performance

Prediction of membrane system performance includes prediction of permeate flux (J),
i.e., throughput of the process and the permeate quality (permeate concentration or
observed retention).

3.15.1 Osmotic pressure model

For membrane separation systems, where, osmotic pressure plays important and limiting
role, osmotic pressure model is used to predict the system performance. Eq, (3.36)
represents the flux as a function of operating pressure and osmotic pressure difference
across the membrane. The osmotic pressure across the membrane is expressed as,
= m p

(3.46)

Using the osmotic pressure relation for concentrated solution Eq.(3.13c), Eq.(3.46) can be
expressed as,
= B1 ( Cm C p ) + B2 ( Cm2 C p2 ) + B3 ( Cm3 C p3 )

(3.47)

Since real retention is constant for a membrane solute system, using Eq.(3.3), permeate
concentration can be expressed in terms of membrane surface concentration as,
C p = Cm (1 Rr )

Joint Initiative of IITs and IISc - Funded by MHRD

(3.48)

Page 33 of 120

NPTEL
Novel Separation Processes

Combining Eqs.(3.47) and (3.48), the osmotic pressure difference can be expressed in
terms of Cm only.
2
3
= B1Cm Rr + B2Cm2 1 (1 Rr ) + B3Cm3 1 (1 Rr )

(3.49)

Eq.(3.41) can also be expressed in terms of Cm only as,

Cm Rr
J = k ln
C C (1 R )
m
r
0

(3.50)

Therefore, Eqs. (3.36) and (3.50) can be solved simultaneously with two unknowns J and
Cm. An iterative algorithm like Newton-Raphson technique may be used. Once Cm is
determined, the permeate concentration is calculated from Eq.(3.48).
It may be mentioned that osmotic pressure model can be used for reverse osmosis,
nanofiltration and some of the ultrafiltration processes for prediction of system
performance.

Some simplified cases for osmotic pressure = bC :

(i)

No concentration polarization,

Cm = C0

ek =1

(3.51)

Thus the permeate flux becomes from Eq.(3.36),

J = L p [ P bRr C0 ]

(ii)

(3.52)

Low polarization: J/k <<1

J
k

e = 1+

J
+ ..........
k

(3.53)

Combining Eqs.(3.53), (3.50) and (3.36), the following expression of

permeate flux is resulted.
Joint Initiative of IITs and IISc - Funded by MHRD

Page 34 of 120

NPTEL
Novel Separation Processes

J
1+

J = L p P bRr C0
J

Rr + (1 Rr ) 1 +

(3.54)

The above solution results into a quadratic in J and can be solved, explicitly.
(iii)

Low polarization and Cp=0 i.e. perfectly rejecting membrane.

J
J = L p P bC0 1 +
k

(3.55)

By simplifying this we get,

J=

Lp ( P bC0 )
bC L
1+ 0 p
k

Lp =

(3.56)

1
Rm

Here, Rm is membrane resistance. Thus, the above expression can be written

as,
J=

P bC0
bC
Rm + 0
k

(3.57)

Here, the first term in denominator represents the membrane resistance and the second
term in the denominator represents the mass transfer resistance and both are in series.

Another method for determination of real retention using film theory:

This is another experimental method and is known as velocity variation technique.
Recall the film theory equation as,
Cm C p

J
= exp
C0 C p
k

Joint Initiative of IITs and IISc - Funded by MHRD

(3.58)

Page 35 of 120

NPTEL
Novel Separation Processes

Using the definition of real retention, Rr = 1

Cp
Cm

Using the definition of observed retention, Ro = 1

one can write C p = Cm (1 Rr ) .

Cp
Co

, one can write, C p = C0 (1 R0 ) .

Thus, Cm and Cp in Eq.(3.58) can be replaced and that equation can be expressed as,
Cp
1 Rr
Cp
1 R0

Cp
=e

J
k

(3.59)

Cp

The above expression can be simplified as,

J

Rr =

R0 e k
vw
K

R0 e + (1 R0 )

(3.60)

The above equation can be rearranged as,

R
R J
ln 0 = ln r
1 Rr k
1 R0

(3.61)

Now, at various operating conditions, experiments are conducted. The corresponding

steady state values of the permeate flux and observed retention are noted. A plot of

J
R
would be a straight line, as shown in Fig. 3.12. From the intercept
ln 0 versus
k
1 R0
of the plot, the value of real retention, Rr is estimated.

Joint Initiative of IITs and IISc - Funded by MHRD

Page 36 of 120

NPTEL
Novel Separation Processes

Rr from intercept
R
ln 0
1 R0

J /k

J
k
Fig. 3.12: Schematic of velocity variation technique

3.15.2 Solution diffusion model for RO/NF:

It can be identified that the concept of real retention is nothing but a partition coefficient
between the solute concentration in the permeate and that at the membrane surface.
Hence for a more realistic situation, the solute flux through the membrane is considered
using solution diffusion model described earlier. The osmotic pressure is considered here
as linear with concentration as occurs generally in case of salt solution, = aC . The
starting equations are osmotic pressure equation Eq.(3.36) and the film theory equation,
Eq.(3.41). The osmotic pressure model can be written by inserting the expression of the
osmotic pressure difference,
J = J 0 1 ( Cm C p )

Where, =

(3.62)

a
, J 0 = L p P is the pure water flux. The above equation can be
P

equated with the film theory equation and the following equation is resulted.

Joint Initiative of IITs and IISc - Funded by MHRD

Page 37 of 120

NPTEL
Novel Separation Processes

J 0 1 ( Cm C p ) = k ln

Cm C p

(3.63)

C0 C p

From the solution-diffusion model, the solute flux is written as,

JC p = B ( Cm C p )

(3.64)

Combining Eqs. (3.62) and (3.64), the following equation is obtained.

Cm C p
J 0 1 ( Cm C p ) = B
Cp

(3.65)

The above equation can be simplified as,

1 Cm + C p =

Where, =

Cm C p
Cp

(3.66)

B
. From above equation, the membrane surface concentration is obtained
vw0

as,

1
Cm = C p 1 +

+ C p

(3.67)

Put Cm from above equation in Eq.(3.63) we get,

Cp
=0
k ln
Cp +
( C p + )( C0 C p )

vw0

(3.68)

Again, a trail and error solution for Cp is sought using some standard iterative technique.

3.15.3 Kedem-Katchalsky equation for RO/NF/UF:

Another variant of osmotic pressure model is Kedem-Katchalsky equation. In this
case, the imperfect retention of the solutes by the membrane is incorporated by a
reflection coefficient in the expression of permeate flux.
Joint Initiative of IITs and IISc - Funded by MHRD

Page 38 of 120

NPTEL
Novel Separation Processes

J = L p ( P

);

(3.69)

= Reflection coefficient . Incorporating the expression of osmotic pressure in the

above expression leads to the following expression of flux.
J = L p P a ( Cm C p )

(3.70)

From the film theory equation, the membrane surface concentration can be expressed as,
J

Cm = C p + ( C0 C p ) e k

(3.71)

Combining Eqs. (3.70) and (3.71), the following equation is resulted

J

J = L p P a ( C0 C p ) e k

(3.72)

The permeate concentration, Cp can be expressed in terms of Cm by either using

the definition of real retention or solution diffusion model, Eq.(3.64). After that,
Eq.(3.72) has to be solved by adopting some iterative scheme.

3.15.4 Modified solution diffusion model for RO/NF/UF:

In this case, the solute transport through the membrane is modified by
incorporating the convective transport of the solutes through the pores, in addition to the
diffusive transport. Thus, this model is more applicable for more porous membrane. The
solute flux is written as,
JC p = ( Cm C p ) + (1 ) JCavg
Where,

Cavg =

(3.73)

Cm C p
C
ln m
Cp

By combining the equations (3.69), (3.71) and (3.73) we get,

Joint Initiative of IITs and IISc - Funded by MHRD

Page 39 of 120

NPTEL
Novel Separation Processes
J

J
(1 ) ( C0 C p ) e k

C p = ( C0 C p ) e k +
J
C C p Jk

ln 1 + 0
e
Cp

(3.74)

We have 3 equations (3.69), (3.71) and (3.3) and 3 unknowns in Cm, Cp and J. We can
solve the above problem iteratively, to have a system prediction.

3.16 Limitation of film theory:

The description of existence of concentration profile across the mass transfer
boundary layer outside the membrane in the flow channel by film theory has certain
shortcomings. These are,

Constant thickness of mass transfer boundary layer underpredicts the permeate

flux. Mass transfer boundary layer is actually developing.

No effect of suction in Shrewood number is taken directly from Heat-mass

transfer analogy.

Physical properties are constant.

These effects are more prominent for ultrafiltration. As, for RO, permeate flux is

less and the effects of suction are less prominent.

Joint Initiative of IITs and IISc - Funded by MHRD

Page 40 of 120

NPTEL
Novel Separation Processes

3.17 Developing mass transfer boundary layer and relevant mass

transfer coefficient

Rectangular thin channel

The schematic of the flow geometry and flow pattern in a rectangular thin channel is
shown in Fig. 3.13.

x
Flow

Membrane

h/2

Fig. 3.13: Flow geometry in a rectangular channel

The following assumptions are made: (i) steady state, fully developed, laminar flow; (ii)
permeation rate is negligible compared to cross flow rate and hence the parabolic velocity
profile remains intact; (iii) there is no adsorption of solute on the membrane surface.
Under these assumptions, the solute balance equation in the thin concentration boundary
layer becomes,
u

c
c
2c
+v
=D 2
x
y
y

(3.75)

The above equation is coupled with the momentum balance equations in x and y
directions and the coupling points are the velocity components in x and y direction. This
complicates the calculation procedure. To overcome this, the velocity fields can be
decoupled without compromising the physical understanding of the system using the

Joint Initiative of IITs and IISc - Funded by MHRD

Page 41 of 120

NPTEL
Novel Separation Processes

assumption (i) and (ii). By virtue of this, it is safely assumed that the x-component
velocity remains a parabolic one, given by the following expression,
u( y) =

2
3 yh
u 0 1

2 h

(3.76)

Therefore, we are dealing with a problem where the hydrodynamic boundary layer is
completely developed and the mass transfer boundary layer is still developing. Assuming
the mass transfer boundary layer is very thin, the domain of applicability of the velocity
profile given by Eq.(3.76) becomes negligible compared to the physical dimension of the
channel. In other words, within thin concentration boundary layer where, Eq.(3.76) is
valid, the dimension y is much less than channel half height h. This further simplifies
2
Eq.(3.77) by opening up the square and neglecting the term y

h2

. Therefore, the x-

component velocity profile becomes,

u( y) =

3u 0 y
h

(3.77)

Within the thin concentration boundary layer, it is further assumed that the y-component
velocity remains constant and is,
v = J ( x)

(3.78)

Inserting these velocity profiles in Eq.(3.75), the governing equation of the solute
becomes,
3u0 y c
c
2c
J
=D 2
y
y
h x

(3.79)

The above equation is a parabolic partial differential equation. We require one condition
on x and two conditions on y. These are, at the channel inlet,

Joint Initiative of IITs and IISc - Funded by MHRD

Page 42 of 120

NPTEL
Novel Separation Processes

At x =0,

c=c0

(3.80)

At the edge of the concentration boundary layer, the concentration is almost the feed
concentration. Therefore,
At y=,

c=c0

(3.81)

The above boundary condition requires some discussion. Since, the thickness of the mass
transfer boundary is very small compared to the channel height, therefore, concentration
remains almost unchanged in the channel, beyond the boundary layer. Thus, any location
in the channel cross section can be assumed to be located at infinity. Hence, the boundary
condition, Eq.(3.81) can be replaced as,
At y = ,

c=c0

(3.82)

This enables Eq.(3.79) to be eligible for having a similarity solution. The other boundary
condition at the membrane surface is that at the steady state all fluxes towards the
membrane is zero.
JC + D

C
= JC p
y

(3.83)

Now, let us find out the similarity parameter for this system of equations. The similarity
parameter is obtained by evaluating the governing equation (Eq. 3.79) at the edge of the
mass transfer boundary layer, recognizing the fact that at this location the term
3u 0 y c
2c
=D 2
h x
y

c
=0.
y

(3.84)

Now, doing an order of magnitude analysis between x = 0 , y = 0 and any x,y, we

obtain,
3u 0 c
c
D 2
h x

Joint Initiative of IITs and IISc - Funded by MHRD

(3.85)

Page 43 of 120

NPTEL
Novel Separation Processes

From the above equation, the spatial variation of the concentration boundary layer is
obtained.
1

hDx 3

=
3
u
0

(3.86)

Thus, the similarity parameter is selected as

1

u 3 y
1 = = 0 1
hD 3
x
y

(3.87)

We have carried out the above analysis in the dimensional form to show that ultimately
the similarity parameter turns out to be non-dimensional. Now, the partial derivatives in
Eq.(3.79) can be expressed in terms of the similarity parameter.
1

c dc 1 u0 3 y 1 1 dc
=
=

x d1 3 hD 13 x 3 x d1
x

(3.88)

c u0 3 1 dc
=
y hD 13 d1
x

(3.89)

2 c u0 3 1 d 2 c
=
y 2 hD 32 d12
x

(3.90)

Substituting the above derivatives in Eq.(3.79), the following equation is obtained after
simplification.
1

2
3
2 J hx dc = d c

1
u0 D 2 d1 d12

(3.91)

Since mass transfer boundary layer varies directly as one third power with x, the
permeate flux inversely varies with x.

Joint Initiative of IITs and IISc - Funded by MHRD

Page 44 of 120

NPTEL
Novel Separation Processes

(3.92)

1
3

In other words, Jx 3 = constant. Therefore, the following equation can be written,

1

hx 3
J
= cons tan t = A1
2
u
D
0

(3.93)

Therefore, Eq.(3.91) now becomes,

d 2c
dc
= 12 A1
2
d1
d1

(3.94)

In the above equation, all the quantities except c is non-dimensional. Therefore, the
concentration is made non-dimensional using the bulk concentration c0 as,
d 2 c*
dc*
2

1
1
d12
d1

(3.95)

The above equation can be integrated once to result,

13

dc*
= K 3 exp
A11
d1
3

(3.96)

One more integration leads to the solution of the concentration field.

13

A11 d1 + K 4
c (1 ) = K 3 exp
3

0
*

(3.97)

The non-dimensional boundary conditions in terms of the similarity parameter become,

at 1 = 0 ,

dc*
+ A1 cm* Rr = 0
d1

(3.98)

at 1 = ,

c* = 1

(3.99)

Joint Initiative of IITs and IISc - Funded by MHRD

Page 45 of 120

NPTEL
Novel Separation Processes

From Eq.(3.97), it may be observed that the membrane surface concentration is evaluated
at 1 = 0 , i.e., cm* = K 4 . Applying the boundary conditions, Eqs.(3.98) and (3.99), the
constants of integration can be obtained,
K3 =

A1 Rr
1 A1 Rr I1

K 4 = cm* =

(3.100)

1
1 A1 Rr I1

(3.101)

and, I1 = exp 1 A1 d1
3

(3.102)

The non-dimensional form of the osmotic pressure relation (Eq. 3.36) in this case
becomes,

Pew x * = B1

( )

(3.103)

Where, B1 = Lp d e P / D and the expression of

in terms of dimensionless

concentration remains as Eq.(3.47). By defining the equivalent diameter of the thin

channel as d e = 4h , Eq.(3.93) can be made dimensionless
1

1
Jd
d 3 3

Pew = e = 4 3 A1 Re Sc e x*
D
L

(3.104)

The constant can be expressed now as,

A1 =

1
3

Pew
1
3

4 Re Sc e
L

1
3

x*

(3.105)

With the above formulation, it is possible to calculate the permeate flux profile along the
channel length, using the following algorithm:

Joint Initiative of IITs and IISc - Funded by MHRD

Page 46 of 120

NPTEL
Novel Separation Processes

At a particular x* location,
(i)

the value of c m* is guessed.

(ii)

Pew is calculated from Eq.(3.103)

(iii)

A1 is evaluated from Eq.(3.105)

(iv)

I is evaluated from Eq.(3.102)

(v)

c m* is evaluated from Eq.(3.101)

(vi)

Convergence between calculated c m* from Eq.(3.101) with guessed valued at

the step (i) is checked.

(vii)

If convergence is achieved in the previous step, calculation for the next axial
location is done until the end of the channel is reached (x*=1).

Length averaged permeate flux and the mass transfer coefficient

The length averaged permeate flux is obtained by integrating Eq.(3.83)
1

1
3

3
d 3
d 3

Pew = Pew ( x* )dx* = 4 A1 Re Sc e x* dx* = 2.38 A1 Re Sc e

L 0
L

0
1

(3.106)

The above equation is rearranged to express the constant A1 in terms of length averaged
permeate flux.
A1 = 0.421

(3.107)

Where, the suction parameter is expressed,

1 =

Pew

(3.108)

de 3

Re Sc L

The definite integral I1 is expressed in terms of the suction parameter as,

Joint Initiative of IITs and IISc - Funded by MHRD

Page 47 of 120

NPTEL
Novel Separation Processes

I1 = exp 1 0.4211 d1
3

(3.109)

The mass transfer coefficient is defined by

C

y y =0
k=
Cm C0

(3.110)

Eq.(3.110). In terms of non-dimensional concentration and similarity variable, the

following expression is obtained.
1

*
u 3 dC
k C 1 = D 0

hxD d1 1 =0

*
m

(3.111)

Substituting the values of c m* and dc * / d1 at 1 = 0 from Eqs.(3.96) and (3.97), the

expression of mass transfer coefficient becomes,

k =

K 3 u0 D

K 4 1 hx
2

1
3

(3.112)

Combining Eqs.(3.100), (3.101) and (3.112) and after simplification, the following
equation is obtained.
1 u D2
k ( x) = 0
I 1 hx

(3.113)

Expressing the mass transfer coefficient in terms of Sherwood number ( kd e / D ) as a

function of dimensionless channel length, one can write,
1

43
Sh( x* ) =
I1

de 3 * 3

Re Sc L ( x )

(3.114)

The average Sherwood number over length is thus obtained.

Joint Initiative of IITs and IISc - Funded by MHRD

Page 48 of 120

NPTEL
Novel Separation Processes
1

de 3
2.381
Sh = Sh( x* )dx* =
Re Sc
I1
L
0

(3.115)

Case 1: For no wall suction: Pew=0

The integral given by Eq.(3.109) becomes,

3
I1 = exp 1 d1 = 1.288
3
0

(3.116)

The corresponding equation for Sherwood number becomes,

1

d 3

Sh = 1.85 Re Sc e
L

(3.117)

The above equation is same as the Leveque solution for flow through a thin channel.
Case 2: For wall suction: Pew 0

For a typical reverse osmosis and ultrafiltration system, the suction parameter 1
vary from low value to upto about 10. The variation of 1/I1 with 1 is shown in Fig. 3.**,
and the results are fitted in the form of a third degree polynomial and the final expression
of Sherwood number becomes,
1

d 3

Sh = 1.85 Re Sc e 1 + 0.321 + 0.0212 8.05 104 13

L

(3.118)

Tubular Module
The detailed derivation of Sherwood number for such a case is available in literature. The
corresponding equation for Sherwood number becomes,
Sh = 1.62( Re Scd e / L)1/3 1.0 + 0.372 + 0.0322 1.05 104 23

Joint Initiative of IITs and IISc - Funded by MHRD

(3.119)

Page 49 of 120

NPTEL
Novel Separation Processes

Where, 2 =

Pew
d

Re Sc
L

1
3

3.18 Faster Procedure/algorithm for calculation of length averaged

permeate flux, membrane surface concentration, etc., in case of
cross flow filtration

Rectangular cross flow geometry

The convective-diffusive boundary condition (known as Robin mixed boundary
condition), presented by Eq.(3.83) holds good at the membrane surface (y=0), over the
element of length x, between x and x+x (i.e., at every x-location on the membrane
surface). Now, if the element x is considered as the membrane length itself, then all the
dependent can be represented as the length averaged values, as shown below.
C
J ( Cm C p ) = D

y y =0

(3.120)

In the above equation, vw, cm, cp, etc., all are length averaged values. This equation can be
combined with Eq.(3.110), definition of mass transfer coefficient as,
J ( Cm C p ) = k L ( Cm C0 )

(3.121)

In the above equation, all the quantities are length averaged. The non-dimensional form
of the above equation is,
Pew =

Where, Pew =

ShL
1
1 *
Rr Cm

(3.122)

Jd e
k L de
and Sh L =
.
D
D

Joint Initiative of IITs and IISc - Funded by MHRD

Page 50 of 120

NPTEL
Novel Separation Processes

The non-dimensional form of the Darcys law or osmotic pressure relationship (Eq. 3.36)
over the membrane length becomes,

Pew = B1 1

(3.123)

The expression of length averaged Sherwood number is derived for different geometries,
flow regimes and rheologies, which are essentially functions of length averaged permeate
flux ( Pew ). For example, Eq.(3.118) for Newtonian fluid, laminar flow. The appropriate
expression of average Sherwood number may be inserted in Eq.(3.122). Eqs. (3.122) and
(3.123) are then solved by simultaneously using trial and error. Therefore, it is noted that
using similarity solution method, the governing partial differential equation along with its
boundary conditions and osmotic pressure relationship finally boil down to two nonlinear algebraic equations to be solved simultaneously to predict the system performance.
It may be emphasized that the permeate flux, membrane surface concentration, etc.,
obtained by this method are all length averaged quantities and profiles of these are not
possible to be obtained by this method. Therefore, similarity solution method is a
powerful tool to estimate quickly the behavior of the state variables of the systems.

3.19 Unstirred Batch systems:

In case of the filtration in an unstirred batch cell, we assume that the permeate volume is
much less than the feed volume. Therefore, the feed volume remains almost constant
leading to the unchanged concentration of the feed. The species mass balance equation
within the concentration boundary layer is,

Joint Initiative of IITs and IISc - Funded by MHRD

Page 51 of 120

NPTEL
Novel Separation Processes

C
C
2C
+v
=D 2
t
y
y

(3.124)

In the above equation, it may be assumed that within thin concentration boundary layer,
v = J

(3.125)

Thus, the species balance equation becomes,

C
C
2C
J
=D 2
t
y
y

(3.126)

The above equation is a parabolic partial differential equation. The pertinent initial
condition is
at t=0, C=C0

(3.127)

At the membrane surface, assuming no adsorption of solutes, total flux towards the
membrane surface becomes zero. This results in the following boundary condition at the
membrane surface,
at y=0, J (Cm C p ) + D

C
=0
y

(3.128)

We require one more boundary condition on y as Eq.(3.3) is second order with respect to
y. The other boundary condition will be,
at y=, C=C0

(3.129)

Since, the thickness of concentration boundary layer is in the order of micron and the
solute concentration is at c0 from the edge of the boundary layer to the bulk of the feed
solution, Eq.(3.129) can be written as,
at y=, C=C0

Joint Initiative of IITs and IISc - Funded by MHRD

(3.130)

Page 52 of 120

NPTEL
Novel Separation Processes

Eq.(3.124) can be made non-dimensional, using the following non-dimensional variables,

c * = c / c 0 and y * = y / R , where, R is the cell radius. If these two non-dimensional
quantities are inserted in Eq.(3.124), the following equation is emerged.
R 2 C *
C * 2C *
Pew * = *2
D t
y
y

(3.131)

In the above equation, Pew is non-dimensional permeate flux ( JR / D ) which is a function

of time. It may be noted that using the above analysis, the non-dimensional form of time
is emerged as = tD / R 2 . Therefore, the solute balance equation in non-dimensional
form becomes,
C *
C * 2C *
Pew * = *2

y
y

(3.132)

The non-dimensional forms of the initial and boundary conditions become,

at = 0 ,

C* = 1

(3.133)

at y*=0,

C *
+ Pew Cm* Rr = 0
*
y

(3.134)

at y*=,

C* = 1

(3.135)

Any governing equation is valid inside as well as on the boundary of the control volume.
Thus evaluating Eq.(3.132) at the edge of the boundary layer and recognizing the fact that
at the edge of boundary layer

C *
= 0 , the following equation is evolved:
y*

C * 2C *
= *2

(3.136)

Now, doing an order of magnitude analysis between = 0, y * = 0 and any , y * , we

obtain,

Joint Initiative of IITs and IISc - Funded by MHRD

Page 53 of 120

NPTEL
Novel Separation Processes

C *

C *

(3.137)

From the above equation, the time variation of the concentration boundary layer is
obtained.

(3.138)

Thus, the similarity parameter is obtained as

(3.139)

Now, the partial derivatives in Eq.(3.9) can be expressed in terms of the similarity
parameter,
C * dC *
1 y* dC *
=
=

d
2 d

(3.140)

C * dC *
1 dC *
=
=
y*
d y *
d

(3.141)

2C *
1 dC * 1 dC * 1 d 2C *
=
=

y*2 y * d d y* d 2

(3.142)

Substituting these partial derivatives in Eq.(3.132), the following second order ordinary
differential equation is obtained,
*
d 2C *

dC

+
J

d 2
2
d

(3.143)

The permeate flux is inversely proportional to the thickness of the concentration

boundary layer,
J

(3.144)

Combining Eqs.(3.138) and (3.144), it can be written as,

Joint Initiative of IITs and IISc - Funded by MHRD

Page 54 of 120

NPTEL
Novel Separation Processes

J = constant=A

(3.145)

Thus, the governing equation becomes,

*
d 2C *

dC
= + A
d 2
2
d

(3.146)

The boundary conditions in terms of the similarity parameters now become,

at = 0 ,

dC *
+ ACm* Rr = 0
d

(3.147)

where, Rr is the real retention defined by Eq. (3.3) and

at = ,

C* = 1

(3.148)

Integrating Eq.(3.146) once, the following expression is obtained,

2

dC *
= K1 exp A
d
4

(3.149)

One more integration of the above equation results the profile of the concentration,

A d + K 2
c * ( ) = K 1 exp
0
4

(3.150)

Use of boundary condition given by Eq.(3.147) results

K 1 + ARr K 2 = 0

(3.151)

Use of boundary condition given by Eq.(3.148) results

K1 I1 + K 2 = 1

(3.152)

where, the parameter I1 is,

A
I 1 = exp
0
4

(3.153)

Solving Eqs.(3.151) and (3.152), the integration constants are obtained,

Joint Initiative of IITs and IISc - Funded by MHRD

Page 55 of 120

NPTEL
Novel Separation Processes

K1 =

ARr
1 ARr I

and K 2 = Cm* =

(3.154)

1
1 ARr I

(3.155)

It may be noted that K2 is identical with the membrane surface concentration.

The above formulation must be coupled with the osmotic pressure model, assuming there
is no solute adsorption. The non-dimensional form of Darcys law (Eq.3.36) becomes,

Pew ( ) = B1

(3.156)

where, B = L p RP / D and the expression of

in terms of dimensionless

concentration becomes,
= b1c * + b2 c m*2 + b3 c m*3

(3.157)

Where, b1 = a1c0 Rr ; b2 = a 2 c 02 1 (1 Rr ) ; b3 = a3 c 03 1 (1 Rr3 ) .

2

The solution algorithm is as follows:

At a particular operating time ,
(i)

a value of Cm* is guessed.

(ii)

Pew is calculated from Eq.(3.156).

(iii)

A is calculated from Eq.(3.145).

(iv)

Definite integral I is evaluated from Eq.(3.153) (the upper limit of this

integral can be taken safely as 10).

(v)

Cm* is calculated from Eq.(3.155).

Joint Initiative of IITs and IISc - Funded by MHRD

Page 56 of 120

NPTEL
Novel Separation Processes

(vi)

The values of Cm* from steps (i) and (v) are compared and if they are
sufficiently close, the iteration is stopped and convergence is achieved at
that time point.

(vii)

The whole calculation is proceeded for next time point.

It may be noted that the convergence is faster if the guess value of membrane
surface concentration at + d is taken as the converged value at .

Mass transfer coefficient

Mass transfer coefficient is a vital design parameter for the design of membrane
modules. The definition of mass transfer coefficient is,
C
k (Cm C0 ) = D

y y =0

(3.158)

In terms of similarity parameter, the non-dimensional form of the above equation

becomes,
C *
1 dC *
K
*

1
y

kR

y* = 0
=0

Sh =
=
=
=
*
*
D
K
1

( Cm 1)
( Cm 1)
2

(3.159)

By substituting the expressions of K1 and K2 from Eqs.(3.154) and (3.155) in the above
equation, the following equation is obtained,
Sh =

1
I

Joint Initiative of IITs and IISc - Funded by MHRD

(3.160)

Page 57 of 120

NPTEL
Novel Separation Processes

3.20 Gel layer controlled filtration:

A gel can be formed over the membrane surface due to two ways,
(i)

Membrane surface concentration increases with pressure or along the length

or a channel and it can exceed the solubility limit of the solutes at that
temperature. Thus, the solid particles settle over the membrane surface as a
gel.

(ii)

Some solutes form a thick viscous layer of constant concentration over the
membrane surface from the very beginning of the filtration. For example,
PVA (poly vinyl alcohol), pectin etc. It is assumed that solute concentration in
the gel layer is constant. Theory for the gel layer controlling filtration is
presented below for viscous cases.

(i) Steady state:

The steady state permeate flux is obtained from film theory as,
J = k ln

Cg
C0

(3.161)

(ii) Most likely gel polarized filtration:

In most of the realistic cases, gel layer grows with time and then attain a steady
state, as shown in Fig. 3.14.

Joint Initiative of IITs and IISc - Funded by MHRD

Page 58 of 120

NPTEL
Novel Separation Processes

C = C0

y=0

C = Cg

y =
y=+L

Permeate flux, J
Fig. 3.14: Formation of gel layer and mass transfer boundary layer over the membrane

There exists a concentration boundary layer beyond gel layer. Where the
concentration varies from C0 to Cg. Within the gel layer, Cg remains constant. Solute
balances at various domains of flow regime are presented below:
For 0 < y <
Solute balance = g

dL
dC
= JC D
dt
dy

(3.162)

The boundary conditions on concentration should be,

and

At y=0,

C=C0

(3.163)

at y=,

C=Cg

(3.164)

The above equation is a non-homogeneous ordinary differential equation. Solution of

above equation is given as,
J
Cg C0 exp
dL
k
g
=J
dt
J
1 exp
k

(3.165)

Please note that the Eq. (3.165) at steady state is reduced to film theory equation.

Joint Initiative of IITs and IISc - Funded by MHRD

Page 59 of 120

NPTEL
Novel Separation Processes

The phenomenological equation of flux can be written as,

J=

Where,

P
( Rm + Rg )

Rm = Membrane resistance;

(3.166)
Rg = Gel layer resistance

Gel layer resistance can be written as,

Rg = (1 g ) g L

(3.167)

Where, = specific gel layer resistance; g = gel porosity; g = gel layer density and L =
gel layer thickness.
can be obtained by from Kozney-carman equation as,

180 (1 g )

(3.168)

g3 g d p2

Eq. (3.165) to (3.168) can be solved simultaneously and numerically to obtain L and J as
a function of time.

Estimation of various parameters in the above model:

Estimation of specific cake layer resistance ():
This can be estimated from the experiments conducted in unstirred batch cell. The
filtration flux for such a system is given as,
J=

Where,

1 dV
A dt

(3.169)

A = filtration area
V = cumulative volume

By combining the Eq. (3.166) and (3.169), we get

Joint Initiative of IITs and IISc - Funded by MHRD

Page 60 of 120

NPTEL
Novel Separation Processes

1 dV
P
=
A dt ( Rm + Rg )
=

Noting,

P
R
Rm 1 + g
Rm

J w0 = pure water flux =

(3.170)

P
Rm

Eq. (3.170) can be written as,

1 dV
=
A dt

J w0
R
1+ g
Rm

(3.171)

A solute mass balance in gel layer leads to,

LA (1 g ) g = C0V

(3.172)

Combining Eq. (3.166) and (3.172) we get,

Rg =

C0V

(3.173)

Inserting Eq. (3.173) in Eq. (3.170) one gets,

1 dV
=
A dt

J w0

(3.174)

C0
1+
V
ARm

An integration of above equation between the limits 0 to t results in,

C0 2
0
V +
V = AJ wt
2 ARm

(3.175)

A rearrangement of Eq. (3.175) results,

C

t
1
=
+ 2 00 V
0
V AJ w 2 A J w Rm

Joint Initiative of IITs and IISc - Funded by MHRD

(3.176)

Page 61 of 120

NPTEL
Novel Separation Processes

A plot of t/V vs V would result in a straight line. The slope can give an estimate
as other parameters of the slope are known.
Since, an unstirred batch experiment can be conducted at various transmembrane
pressure drops, can be obtained as a function of P as,

= 0 ( P )

(3.177)

If n = 0,

gel layer is incompressible

If n < 1,

gel layer in compressible

Estimation of g, dp:
If dp is particle molecular diameter then volume of one molecule is
V0 =

d 3p

(3.178)

Volume of 1 mole = N A

d 3p ,

Number of moles in 1 m3 solution = 103

Where,

NA = Avagadro number
Cg
Mw

Cg = gel layer concentration in kg/m3

Mw = molecular weight of solutes

Thus, volume of gel in 1 m3 solution is = 103

Cg
Mw

NA

d 3p

The gel layer porosity is thus given as,

g = 1 103

Cg
Mw

NA

d 3p

(3.179)

Combining Eq. (3.168) and (3.177),

Joint Initiative of IITs and IISc - Funded by MHRD

Page 62 of 120

NPTEL
Novel Separation Processes

1 g
180
3 2
d

g
p
g

0 ( P ) =

(3.180)

Eq. (3.180) has only one variable dp on the right hand side as g can be expressed in terms
of dp by Eq. (3.179).
Eq. (3.180) can be solved using Newton-Raphson technique to obtain dp and g at various
values of P.
It may be mentioned that above method is for malleable particles like polymeric and
polysaccharide solutes. For proteins and colloids, hard sphere diameter is considered as
dp.

Estimation of gel layer concentration (Cg):

Steady state experiments are conducted at various degrees of turbulence or mass
transfer coefficient, at various feed concentration. A plot of steady state concentration vs
ln C0 is done, as shown in Fig. 3.15.

k increasing

ln C0

ln Cg

Fig. 3.15: Estimation of gel concentration

All the curves are extrapolated and they cut at ln Cg on the axis.

Joint Initiative of IITs and IISc - Funded by MHRD

Page 63 of 120

NPTEL
Novel Separation Processes

Thus, experimentally gel layer concentration is determined.

3.21 Resistance in series model

It is a black box approach. Various resistances are presented in series and flux is
proposed as,
J=

P
( Rm + Rg + Rbl + Rirr )

(3.181)

Where, Rm is membrane resistance, Rg is gel layer resistance, Rbl is boundary layer

resistance and Rirr is irreversible membrane resistance. These resistances are related to the
operating conditions.

3.22 Modeling of Membrane Modules

In this section, design equations of various modules under various flow regimes have
been developed. A one-dimensional analysis is mainly considered. The development of
the equations for membrane module is carried out in step by step manner, with increasing
order of complexity. First, the case is considered when the permeate flux is constant.
Next, the permeate flux is considered as proportional to transmembrane pressure drop
and negligible osmotic pressure build up across the membrane. Finally, the inclusion of
osmotic pressure is considered.

Joint Initiative of IITs and IISc - Funded by MHRD

Page 64 of 120

NPTEL
Novel Separation Processes

Simplified models (Newtonian fluid and laminar flow)

3.22.1: Permeate flux is constant
Spiral wound module
The flow through a spiral wound module can be reasonably approximated as the flow
through a rectangular channel when the module is opened up. In this case, we consider
the permeate flux is constant. This situation corresponds to the case, where, very dilute
solution is filtered. The assumptions involved are: (i) the flow is steady; (ii) no
concentration build up over the membrane surface, leading to almost constant permeate
flux; (iii) permeation rate is negligible compared to bulk cross flow; (iv) the flow regime
is laminar. The flow geometry is presented in Fig.3.16.

Qout

Qin

x=0

vw

x=L

Fig. 3.16: Flow through a reactangular channel

For slit flow with impermeable wall, the volumetric flow rate at any axial x-location can
be calculated by solving equations of motion and is expressed as,
Q=

2h 3 w dP

3 dx

Joint Initiative of IITs and IISc - Funded by MHRD

(3.182)

Page 65 of 120

NPTEL
Novel Separation Processes

For impermeable channel, the volumetric flow rate is constant and therefore pressure is a
linear function of axial location. At this point the definition of pressure needs to be
clarified. P in Eq.( 3.182) is absolute value of pressure at any axial location. The term
dP
represents the axial pressure drop. It may be noted here that the pressure in the
dx
permeate side is near atmospheric pressure. Therefore, transmembrane pressure
difference at any axial location is defined as,
P( x) = P ( x) Patm

(3.183)

Differentiating Eq.( 3.183) with respect to x, one obtains,

dP dP
=
dx
dx

(3.184)

Therefore, variation of pressure inside the channel along the axial direction is equivalent
to variation of transmembrane pressure drop along same direction.
The volumetric flow rate (Q) is not constant for membrane channel. In case of membrane
channel, permeation occurs through the walls of the channel. Therefore, the volumetric
flow rate is a decreasing function of x due to loss of the materials. Now, consider, a
differential element of the membrane channel. The overall material balance over the
differential element leads to the governing equation of change of volumetric flow rate at
the steady,
Rate of material in Rate of material out = Rate of accumulation

Q x Q x +x p J (2wdx) = 0

(3.185)

The factor 2 appears in Eq.( 3.185), because of there is suction across the membrane wall
from both sides of the channel. Assuming, the density of the feed and permeate are same,
the governing equation of the flow rate in the channel is obtained.

Joint Initiative of IITs and IISc - Funded by MHRD

Page 66 of 120

NPTEL
Novel Separation Processes

dQ
= 2 wJ
dx

(3.186)

Differentiating Eq.(3.182) with respect to x, the following equation is obtained,

2 h 3 w d 2 P dQ

=
3 dx 2 dx

(3.187)

Combining Eqs.(3.186) and (3.187), the explicit expression for profile of transmembrane
pressure drop is obtained.
d 2 P 3
= 3 J
dx 2
h

(3.188)

Since, the above equation is a second order ordinary differential equation, two boundary
conditions are required to solve the same. It may be remembered that only the inlet
conditions are specified for this problem. Therefore, we have to formulate two distinct
boundary conditions from the inlet conditions (at x=0). The first one is that the
transmembrane pressure drop is known at the inlet.
P = Pi

(3.189)

The second inlet information that is known to us is the flow rate. This is directly obtained
from Eq.(3.182),
3
dP
= 3 Qi
dx
2h w

(3.190)

It is interesting to note that this problem is a typical example where both the boundaries
are specified at the same location and one is dirichlet (value of dependent variable
specified) and the other one is neumann (derivative of the dependent variable is
specified). Thus, this problem is truly an initial value problem. Eq.(3.188) is now solved
using the boundary conditions given by Eqs.(3.189) and (3.190) and the solution is given
below,
Joint Initiative of IITs and IISc - Funded by MHRD

Page 67 of 120

NPTEL
Novel Separation Processes

Pi P ( x) =

2 Jwx
3
Qi x 1

3
2hw
2Qi

(3.191)

The total pressure drop in the full channel is given as,

Pi P ( L) =

2 JwL
3
Qi L 1

3
2hw
2Qi

(3.192)

= ( P ) vw = 0 1
2

In the above equation, f is the fractional permeate recovery which is the ratio of total
permeate flow rate to total inlet flow rate. Physically, f indicates the fraction of the net
material entering into the system is recovered in the permeate stream. Higher be the value
of f, better is the throughput of the system.

Tubular module
The above analysis is carried out for tubular module as described below.
For tubular flow with impermeable wall, the volumetric flow rate at any axial x-location
can be calculated by solving equations of motion and is expressed as,

R 4 dP
Q=

8 dx

(3.193)

Overall material balance over a differential element located between x and x+x, at the
steady state results,

( Q x Q x +x ) p J ( 2 Rx ) = 0

(3.194)

The above equation is simplified to,

dQ
= 2 RJ
dx

Joint Initiative of IITs and IISc - Funded by MHRD

(3.195)

Page 68 of 120

NPTEL
Novel Separation Processes

Differentiating Eq.(3.193) with respect to x, the following expression is obtained,

dQ R 4
=
dx
8

d 2 P

2
dx

(3.196)

Equating Eqs. (3.195) and (3.196), the governing equation of trnsamembrane pressure
drop is obtained,
d 2 P 16
= 3 J
dx 2
R

(3.197)

Eqs.(3.189) and (3.190) present the boundary conditions for the above equation. The
solution is as follows,
Pi P ( x) =

JRx
8
Qi x 1

4
R
Qi

(3.198)

The axial pressure drop over the module length L is given as,
Pi PL =

8
f

Qi L1

4
R
2

(3.199)

Where, the fractional recovery of feed f is expressed as,

f =

2RLv w
Qi

Joint Initiative of IITs and IISc - Funded by MHRD

(3.200)

Page 69 of 120

NPTEL
Novel Separation Processes

3.22.2: Permeate flux is proportional to trasnsmembrane pressure

difference, completely retentive membrane and osmotic pressure is
negligible
Spiral wound module
In this case, the concentration polarization is assumed to be negligible. This indicates, the
osmotic pressure generated in the feed channel is small compared to the transmembrane
pressure difference and the feed channel has uniform solute concentration at the entry
level. Therefore, the permeate flux at any x-location of the module is proportional to the
transmembrane pressure difference at that location.
J = LP P

(3.201)

As shown earlier, the governing equation of transmembrane pressure drop is obtained by

Eq.(3.188):
d 2 P 3LP
=
P
dx 2
h3

(3.202)

Using the boundary conditions presented by Eqs. (3.189) and (3.190), the profile of
pressure drop is,
3 Qi
P ( x)
sinh(x)
= cosh(x)
2 h 3 w Pi
Pi

(3.203)

The profile of axial pressure drop becomes,

Pi P ( x) = Pi [1 cosh (x )] +

Where, the parameter is given as, =

3 Qi
sinh (x )
2 h 3 w

3L p
h3

Joint Initiative of IITs and IISc - Funded by MHRD

(3.204)

Page 70 of 120

NPTEL
Novel Separation Processes

The axial pressure drop over the full length of the module is obtained by substituting x
by L.
Pi P ( L) = Pi [1 cosh (L )] +

3 Qi
sinh (L )
2 h 3 w

(3.205)

The fractional recovery of the feed over the complete module is given as,
L

f =

Qp
Qi

2wLp P( x)dx
0

Qi

2wLp Pi

3Qi
{cosh(L) 1}
sinh(L) 3
Qi
2h wPi

(3.206)

The profile of permeate flux is obtained from Eq.(3.201) as,

3 Q
J ( x) = LP Pi cosh( x ) 3 i sinh ( x )
2h wPi

(3.207)

The profile of cross flow velocity is obtained by substituting flow rate Q = 2 xhu in
Eq.(3.186).
du
J
=
dx
h

(3.208)

Expression of permeate flux is then substituted in Eq.(3.208) from (3.207) and integration
is carried out from inlet to any location in the module,
L p Pi x 3Qi

u du = h 0 2h 3 wPi sinh (x ) cosh(x ) dx

i
u

(3.209)

Carrying out the above integration, the profile of cross flow velocity along the module
length is obtained.

Joint Initiative of IITs and IISc - Funded by MHRD

Page 71 of 120

NPTEL
Novel Separation Processes

Lp Pi
u( x)
= 1
ui
hui

3Qi
{cosh(x) 1}
sinh(x) 3
2h wPi

(3.210)

The cross flow velocity at the channel outlet is obtained by substituting x by L in the
above equation. Since, the membrane is completely retentive membrane, the species
balance over the differential element results,
d ( uC 2 wh ) = 0

(3.211)

The channel dimensions remain constant, the above expression reduces to the following
equation,
u ( x)C ( x) = ui Ci

(3.212)

Combining Eq.( 3.212) and (3.210), the profile of bulk concentration is obtained as,
u
C ( x)
= i =
Ci
u ( x)

1
L P

3 Q
1 p i sinh( x) 3 i {cosh( x) 1}
h ui
2h wPi

(3.213)

The bulk concentration at the module outlet is obtained by putting L in lace of x in the
above equation.

Tubular module
In case of tubular module, the cross sectional averaged velocity is obtained from
Eq.(3.193),
u=

R 2 dP

8 dx

(3.214)

The above equation is valid assuming that the permeation velocity is so less, it does not
affect magnitude of cross flow velocity, keeping its parabolic profile intact. Eq.(3.214)
presents the governing equation of transmembrane pressure drop,
Joint Initiative of IITs and IISc - Funded by MHRD

Page 72 of 120

NPTEL
Novel Separation Processes

8
dP
= 2u
dx
R

(3.215)

Writing down the material balance over a differential element, the governing equation of
cross flow velocity is obtained at steady state.

u ( u + du ) R 2 p J 2 Rdx = 0

(3.216)

Assuming the feed and permeate density do not vary much, the governing equation of
cross flow is becomes,
du
2J
=
dx
R

(3.217)

Since, the osmotic pressure is negligible compared to transmembrane pressure drop,

utilizing Eq.(3.201), the following equation is resulted.

2 L p P
du
=
dx
R

(3.218)

Differentiating Eq.(3.215) with respect to x and combining Eq.(3.217), the governing

equation of pressure drop is obtained
d 2 P 16L p
=
P
dx 2
R3

(3.219)

Eqs(3.189) and (3.190) are the boundary condition of the above equation. Therefore, the
profile of trans membrane pressure drop becomes,
P( x) = Pi cosh(mx) sinh(mx)
where, m =

16L p
R

and =

(3.220)

8 u i
. The profile of axial pressure drop along the module
mR 2

becomes,

Joint Initiative of IITs and IISc - Funded by MHRD

Page 73 of 120

NPTEL
Novel Separation Processes

Pi P( x) = Pi [1 cosh(mx)] + sinh(mx)

(3.221)

the total axial pressure drop across the module becomes,

Pi P( L) = Pi [1 cosh(mL)] + sinh(mL)

(3.222)

The permeate flux is obtained from Eq.(3.201),

J ( x) = L p Pi cosh ( mx ) sinh ( mx )

(3.223)

Substituting profile of tarns membrane pressure in Eq.(3.218) and integrating

between x=0 and any location in the module, the profile of cross section averaged
velocity is obtained,
2L p
u
= 1
[Pi sinh(mx) + {1 cosh (mx )}]
mR
ui

(3.224)

As formulated in Eq.(3.212), the bulk concentration of the solute at any x-location is

given as,
u
C ( x)
= i =
Ci
u ( x)

1
1

2 Lp

Pi sinh( x) {cosh( x) 1}
mR

(3.225)

3.22.3 Osmotic pressure is not negligible and partially retentive

membrane
In this section, we have developed the one dimensional design equations for membrane
modules. The model is fairly detailed. The osmotic pressure and effects of concentration
polarization are included. The membrane is considered as partially retentive. The effects
of suction on the mass transfer coefficient are also included. The section is divided into
several subsections, each elaborates calculation procedure for various types of fluids,
flow regime and geometry of flow.

Joint Initiative of IITs and IISc - Funded by MHRD

Page 74 of 120

NPTEL
Novel Separation Processes

Newtonian fluids Laminar flow

Spiral wound module:
In this case, the governing equation of transmembrane pressure drop is obtained from
Eq.(3.182) as
dP
3Q
= 3
dx
2h w

(3.226)

By substituting the flow rate as Q = 2 whu in above equation, the governing equation of
pressure drop is obtained.
3 u
dP
= 2
dx
h

(3.227)

The governing equation for cross flow velocity is subsequently obtained from Eq.( 3.208)
which is directly derived from Eq.( 3.186). Osmotic pressure across the membrane is not
neglected in this case. The permeate flux through the membrane is expressed by Darcys
law for the porous medium,
J = Lp ( P )

(3.228)

Using Eqs. (3.3), (3.47) and (3.228), the permeate flux can be expressed as a
function of membrane surface concentration,

J = L p P B1cm Rr B2 cm 2 1 (1 Rr )

) + B c (1 (1 R ) )}
3

3 m

(3.229)

A solute mass balance in the differential element results,

J .C p
d (uC )
=
dx
h

(3.230)

Using Eq. (3.208), Eq. (3.230) is simplified to,

Joint Initiative of IITs and IISc - Funded by MHRD

Page 75 of 120

NPTEL
Novel Separation Processes

u.

dC J
= .(C C p )
dx h

(3.231)

Where, Cp is the permeate concentration.

Thus, three ordinary differential equations are arrived at for three independent variables
transmembrane pressure drop, cross flow velocity and bulk concentration.
The membrane surface concentration, c m is related to the bulk concentration c by the
definition of mass transfer coefficient at any x ,
k ( Cm C ) = D (

C
) y =0
y

(3.232)

At the steady state, summation all the fluxes towards the membrane surface is zero. This
leads to the following condition,
J ( Cm C p ) = D (

C
) y =0
y

(3.233)

Combining Eqs. (3.3), (3.232) and (3.233), the following expression is obtained,
k ( Cm C ) = JCm Rr

(3.234)

Combining Eqs.(3.229) and (3.234), the following algebraic equation for membrane
surface concentration with the bulk concentration is obtained,
k ( Cm C )
= P ( A1Cm + A2Cm 2 + A3Cm3 )
Cm Rr Lp

(3.235)

Where, A1=B1Rr ; A2=B2[1-(1-Rr)2]; A3=B3[1-(1-Rr)3] . The mass transfer

coefficient can be expressed as function of x and u as,
1 uD 2

k ( x ) =
I hx

1 3

(3.236)

Where, I can be calculated by evaluating the following definite integral

Joint Initiative of IITs and IISc - Funded by MHRD

Page 76 of 120

NPTEL
Novel Separation Processes

0.42 d
I = exp
3

(3.237)

In the above equation, is the suction parameter and is defined as,

dL
=J e 2
uD

(3.238)

Where, J is the length averaged permeate flux represented as,

1
J=
L

J ( x ) dx

(3.239)

For impervious conduit (no suction), the value of is zero and the value of I becomes,

I = 1.2879 .

Tubular module:
The governing equation of transmembrane pressure drop in terms of tube
diameter is obtained from Eq.(3.215)
32 u
dP
=
dx
d2

(3.240)

The governing equation of cross flow velocity is obtained from Eq.(3.217),

du
4J
=
dx
d

(3.241)

and that for the solute concentration is obtained by carrying out a solute mass balance in
the differential element,
u

dC 4 J
=
(C C p )
dx
d

(3.242)

The expression of permeate flux remains same as Eq. (3.229). The equation
relating the membrane surface concentration with the bulk concentration through mass

Joint Initiative of IITs and IISc - Funded by MHRD

Page 77 of 120

NPTEL
Novel Separation Processes

transfer coefficient remains as Eq.(3.234). The mass transfer coefficient for the tubular
module is derived as a function of x and u as,
1 u D2

k =
I 1 d x

(3.243)

Where, I 1 can be calculated by evaluating the following definite integral,

8 3 2
d
I 1 = exp
9
3

(3.244)

is defined as in Eq.(3.238) replacing de by d. J is the length averaged permeate

flux represented as in Eq.(3.239). For impervious conduit (without suction, =0), the
value of the integral I1 is 0.926.

Newtonian fluids turbulent flow

Spiral wound module:
In case of turbulent flow regime, the governing equations of cross flow velocity and
solute concentration remain same as Eqs.(3.208) and (3.231). The governing equation of
transmembrane pressure drop is obtained by carrying out an overall energy balance over
the differential element. The energy balance equation in the differential element becomes,
P ( P + dP ) +

u2
2

( u + du )
2

Am J 2 2
4 f dx u 2
+
( u vw ) = 2d
Q 2
e

(3.245)

where, Am is the area of the permeation. In the differential element, Am=2 w dx. It
may be noted here that P in Eq.(3.245) is the gauge pressure and since the permeate is
collected at near atmospheric pressure, P in Eq.( 3.245) is equivalent to P , the
transmembrane pressure drop. Inserting the definition of Am in Eq.( 3.245), the following
equation is obtained,

Joint Initiative of IITs and IISc - Funded by MHRD

Page 78 of 120

NPTEL
Novel Separation Processes

wvw
wJ
4 fu 2
2
2
dx
d P ( u du ) +
u dx
J dx =
Q
Q
2d e

(3.246)

The equivalent diameter is defined as d e 4h for a thin channel. Using the definition of
equivalent diameter, Eq.( 3.246) can be simplified to the following equation,

fu 2
d P
du w J
wJ
= u
+
u2
J2
dx
dx Q
Q
2h
Replacing the flow rate, Q=2xhu and

(3.247)

du
from Eq.(3.208), the governing
dx

equation of the transmembrane pressure drop across the module length is obtained as,

d P
J3
=

fu 2
3
u
J

dx
2h
u

(3.248)

The fanning friction factor, f for the turbulent flow in a smooth pipe is given by Blasius
formula,
f =

0.079
Re 0.25

(3.249)

In Eq.( 3.249), Reynolds number is defined based on the equivalent diameter. The
expression of permeate flux remains same as Eq.( 3.229). The equation relating the
membrane surface concentration with the bulk concentration through mass transfer
coefficient remains as Eq.( 3.232). The mass transfer coefficient for the turbulent flow
can be expressed as function of x and u as,
1

0.236 D Re1.75 Sc 3
k ( x) =

2
I
xd e

(3.250)

Where, the integral I is calculated as,

2.82 d
I = exp
0
3

Joint Initiative of IITs and IISc - Funded by MHRD

(3.251)

Page 79 of 120

NPTEL
Novel Separation Processes

In the above equation, is the suction parameter and is defined as,

vw

Re Sc
D

2
de L
1.75

(3.252)

1
3

Where, J is the length averaged permeate flux defined as Eq.(3.239). For impervious
conduit, the value of is zero and the value of I becomes 1.29.

Flow through a tubular module:

For flow through a tubular module of diameter d, Eqs.(3.241) and (3.242)
present the governing equations of cross flow and solute concentration. An energy
balance over the differential element like the spiral wound module, the following
equation for the transmembrane pressure drop is obtained.

v w3
d P 2
=
fu 2
3u v w
dx
d
u

(3.253)

The mass transfer coefficient for the tubular module is derived by Minnikanti et al. and
Eqs. (3.250) to (3.252) remain in the same form; only de is replaced by the tube diameter
d.

3.23 Dialysis
In dialysis system the transportation of solute is takes place due to concentration
gradient. The schematic of a batch and continuous dilayzer are shown in Figs. 3.17 a and
3.17 b.

Joint Initiative of IITs and IISc - Funded by MHRD

Page 80 of 120

NPTEL
Novel Separation Processes

Feed chamber

Dialysate
chamber

Membrane

Fig. 3.17 (a): Schematic of a batch dialyzer

In a batch system there is a level difference after some time as both solute and
solvent move together.

Continuous System:
Feed out

Dialysate in

Feed in

Dialysate out

Fig. 3.17 (b): Schematic of a continuous dialyzer

Transport mechanism across the membrane:

Diffusion is the main mechanism.
N i = Mass flux of i th species
= Dim

dCim
Cim
= Dim
dx
L

Here, Dim = Diffusivity through the membrane ;

Joint Initiative of IITs and IISc - Funded by MHRD

(3.254)
L = membrane thickness

Page 81 of 120

NPTEL
Novel Separation Processes

Ni =

Dim F
Cim CimD
L

(3.255)

It is assumed that,
CimF
CimD
=
m
=
i
CiF
CiD

(3.256)

Where, CiF , CiD = bulk concentration of Feed and Dialysate and

mi = distribution coefficient . Thus, the solute flux across the membrane becomes
Ni =

Dim
mi CiF CiD
L

(3.257)

Here, mi = 0.5 for urea in water and regenerated cellulose and Dim
For phenol,

1011 m 2 /s .

mi = 0.01 (for a polyethylene film)

= 17.5 (for an Ethyl cellulose film)

Determination of Dim:
Diffusivity of solute through the membrane phase can be determined as,
Dim =

DiB

(3.258)

Fdrag

Where, DiB = bulk diffusivity ; = membrane porosity ; = tortuosity

and Fdrag = drag coefficient on the particle of radius ri in a pore of radius rs
r
= 1 2.1 i
rs

ri
ri
+ 2.09 0.95

rs
rs

(3.259)

Various resistive films of liquids across the membrane is shown in Fig. 3.18 (a) and 3.18
(b).

Joint Initiative of IITs and IISc - Funded by MHRD

Page 82 of 120

NPTEL
Novel Separation Processes

Rf

Rd

Fig. 3.18 (a): Schematic of various films across the membrane

R f = liquid film in feed side

Rd = liquid film in dialysate side
CiF = bulk concentration of feed at steady state
CiF = bulk interface concentration of membrane solution

CiF
CiF

CiD

CiD

Fig. 3.18 (b): Concentration gradient across the membrane

The solute flux across the membrane in the various flow domains are presented below:
Ni =

CiF CiF
Rf

(Across liquid film in feed)

Dim
( CiF CiD ) (Across membrane)
L

Joint Initiative of IITs and IISc - Funded by MHRD

Page 83 of 120

NPTEL
Novel Separation Processes

CiD CiD
Rd

(Across dialysate side)

(3.260)

Thus, the overall solute flux becomes,

Ni =

CiF CiD
R0

(3.261)

Where, overall resistance is given as,

R0 = R f +

L
+ Rd
Dim

(3.262)

In terms of overall mass transfer coefficient, the overall mass transfer coefficient is
expressed as,
1
koverall

koverall =

1
L
1
+
+
k f Dim kd
1
L
1
1
+
+
k f Dim kd

(3.263)

The above equation is valid for every location, along the length of the dialyzer.

Joint Initiative of IITs and IISc - Funded by MHRD

Page 84 of 120

NPTEL
Novel Separation Processes

Calculation of net mass flow rate across the membrane in a dialysate unit:
The concentration gradient in a counter current dialyzer is presented in Fig. 3.19.

CiF
COD

COF

dm

CiD

dA

Fig. 3.19: Concentration gradient across a counter current dialyzer

dm = Mass flow rate across dA

= k0 CdA

(3.264)

VF = volumetric flow rate of feed

and VD = volumetric flow rate of dialysate .

balance over differential element becomes,

dm = VF dCF = VD dCD

(3.265)

Rearrangement of above equations leads to the following,

dCF =

dm
;
VF

dCD =

dm
VD

(3.266)

C = CF CD

(3.267)

d ( C ) = dCF dCD

(3.268)

1
1
d ( C ) = +
dm
V
V
D
F

(3.269)

From Eq. (3.264), the following equation is resulted.

Joint Initiative of IITs and IISc - Funded by MHRD

Page 85 of 120

Mass

NPTEL
Novel Separation Processes

1
1
d ( C ) = +
( k0 dA ) C
VF VD

(3.270)

Integration of above equation is carried out between the inlet and exit of the channel,
A
1
d C
1
=

+
i C VF VD 0 k0 dA
0

(3.271)

The final result becomes,

ln

A
1
C |0
1
= +
k0 dA
C |i
VF VD 0

(3.272)

Integration of Eq.(3.269) results,

m
1
1
d

C
=

+
i ( ) VF VD 0 dm
0

1
1
C |0 C |i = +
m
VF VD

(3.273)

Combining Eqs. (3.272) and (3.273), the following is the expression of mass transported
across the membrane per unit time.
( C )0 ( C )i
m ( Kg/s ) = k0 A
( C )0
ln
( C )i

(3.274)

Thus, the mass transported can be expressed in terms of overall mass transfer coefficient
and Log mean temperature difference.
m = k0 A ( C ) LMTD
k0 =

1
1
1
L
+
+
k f Dim kd

Joint Initiative of IITs and IISc - Funded by MHRD

(3.275)
(3.276)

Page 86 of 120

NPTEL
Novel Separation Processes

Batch dialysis analysis:

A typical batch dialyzer is shown in Fig. 3.20.

VF

VD

CiD(t)

Fig. 3.20: A typical batch dialyzer

VF , VD = volumes of feed and dialysate chambers (fixed)

Assume the both chambers are well stirred, so that, the concentration is uniform
throughout the chambers.
Mass balance in dialysate and feed side are written as,
A D
d
( CiDVD ) = m im ( CiF CiD )
dt
L

(3.277)

A D
d
( CiFVF ) = m im ( CiF CiD )
dt
L

(3.278)

The initial conditions are:

CiF = CiF0 and CiD = 0 .

at t=0,

Applying Laplace transform to equations (3.277) and (3.278), the following equations
are resulted.
SCiD =

SCiF CiF0 =
Let us take,

Am Dim
CiF CiD
LVD

Am Dim
CiF CiD
LVF

(3.279)

(3.280)

Am Dim
= K and the above equations become
L

Joint Initiative of IITs and IISc - Funded by MHRD

Page 87 of 120

NPTEL
Novel Separation Processes

SCiD =

K
CiF CiD
VD

SCiF CiF0 =

K
CiF CiD
VF

(3.281)

(3.282)

The above two equations can be solved as,

K 0
CiF
VD
CiD =
1
1
S 2 + SK +

VF VD

(3.283)

1
1
Considering, a = K +
, the above equation becomes,
VF VD
K
CiD =
VD

CiF0 1
1

a S S +a

(3.284)

Taking inverse laplace, the solution becomes,

K CiF0
1 e at )
CiD ( t ) =
(
VD a

VF
=C
VF + VD
0
iF

K + t

VF VD

1
e

(3.285)

Now, with measured values of CiD at different time, a plot of CiD with time is shown in
Fig. 3.21.

CiD

Time (t)

Joint Initiative of IITs and IISc - Funded by MHRD

Page 88 of 120

NPTEL
Novel Separation Processes

Fig. 3.21: Typical variation of CiD with time in a batch dialyzer

A rearrangement of Eq.(3.285) becomes,

C
ln 1 iD
R1

= KR2t

(3.286)

C
Thus, a plot of time vs ln 1 iD , is a straight line as shown in Fig. 3.22.
R1

C
ln 1 iD
R1

Time (t)
C
Fig. 3.22: Typical variation of time vs ln 1 iD
R1

in a batch dialyzer

We get a straight line with a negative slope of KR2, where, K =

and R2 =

Am Dim
L

D
1
1
+
. Therefore, im can be estimated.
L
VF VD

Joint Initiative of IITs and IISc - Funded by MHRD

Page 89 of 120

NPTEL
Novel Separation Processes

Continuous Dialysis:
A continuous counter current dialysis system is shown in Fig. 3.23.

Fe, CiFe

Fi, CiFi

Di, CiDi

DO, CiDe

Fig. 3.23: Typical schematic of a counter current dialyzer

The assumptions involved in the analysis are listed below:

1) Constant feed flow rates
2) Dialyser performance depends on the value of film and membrane resistance
3) Film thickness depend upon the geometry and velocity profiles in channel

Design Equations:
Tranmembrane molar flow rate (mole/sec):
mi = k0 Am ( Ci ) LMTD

(3.287)

Molar Flux can also be presented as:

(
= L (C

mi = VF CiFi CiFe
D

iDe

)
)

CiDi

(3.288)

Using above equations, one can design membrane area for a given separation.
Dialyzer efficiency (), can be defined as:

Joint Initiative of IITs and IISc - Funded by MHRD

Page 90 of 120

NPTEL
Novel Separation Processes

Actual amount of solute depletion in feed

Maximum amount separated porsion in a dialyzer

VF CiFi CiF0

(3.289)

VF CiFi LD CiDi

Enhancement of separation in dialysis by secondary chemical reaction:

In a dialyzer chamber, select a dialyzer such that diffused components react and
product ions cannot diffuse back and therefore, a maximum concentration gradient is
always maintained. For example, Aniline is to be removed, using dialysate as H2SO4.
Aniline + H 2SO 4
(1)

Anilinium ion

(2)

(3)

Component 2 is in excess such that reaction is always complete.

Mass balance:
Solute balance in the feed chamber becomes,
VF

D
dC1F
= Am im ( C1F C1D )
dt
L

(3.290)

The above equation is rearranged as,

dC1F Am Dim
=
( C1D C1F )
dt
LVF

(3.291)

Solute balance in the dialysis chamber becomes,

dC
dC1D Am Dim
=
( C1F C1D ) 3D
dt
LVD
dt
Considering

(3.292)

Am Dim
= K , the above equations become,
L

dC1F K
=
( C1D C1F )
dt
VF

Joint Initiative of IITs and IISc - Funded by MHRD

(3.293)

Page 91 of 120

NPTEL
Novel Separation Processes

dC3 D dC1D K
+
=
( C1F C1D )
dt
dt
VD

(3.294)

C1F = C10F

Initial conditions are: at t=0,

C1D = 0
C3 D = 0
At any timet, C3D is in equilibrium. So that,
K eq =

C3 D
C1D C2 D

(3.295)

Here, C2D is constant because it is in excess.

Take laplace transform to equations (3.293) and (3.294), the following equations are
obtained.

(3.296)

(3.297)

K
C1D C1F = SC1F C10F
VF
K
C1F C1D = SC1D C10F + SC3 D C30D
VD
From above two equations, the following equation is resulted.
K
V
K

C1D = + S 1 + D (1 + C2 D ) S C1D C10F

VF

VF
K

(3.298)

By simplifying the above equation finally we get it as,

C1D =
Where, A =

C10F
AS + BS 2

VD
(1 + C2 D ) ;
VF

(3.299)
B=

VD
(1 + C2 D )
K

Laplace inverse can be taken of Eq.(3.299) to obtain the variation of C1D as a function of
time.
C1D = f (t )

Joint Initiative of IITs and IISc - Funded by MHRD

(3.300)

Page 92 of 120

NPTEL
Novel Separation Processes

Detailed two dimensional analysis of Dialysis:

Dialysate side
Blood
in
laminar
flow

Concentrati
on profile

x
y
x=0
impermeable
entrance

membrane

Dialysate side

Fig. 3.24: A typical dialysis channel.

A typical dialysis channel with the coordinate system is shown in Fig. 3.24. The origin at
the center-line of channel. We consider a Newtonian-laminar flow in a rectangular
channel with the velocity profile,
3
y2
u ( y ) = u0 1 2
2 h

(3.301)

y is at mid plane and h is channel half height

Q = 2u0WH = flow rate .
The various assumptions involved are channel is very wide compared to channel height H
and flow and mass transport is two dimensional
At steady state, for H y H , the solute mass balance becomes,
u

C
2C
=D 2
x
y

At x = 0,

C = C0

(3.302)
(3.303)

Symmetry at mid plane provides another boundary condition:

C
=0
y

at y = 0

Joint Initiative of IITs and IISc - Funded by MHRD

(3.304)

Page 93 of 120

NPTEL
Novel Separation Processes

at y = h;

molar flux = PC

Where, P is membrane permeability.

(i)

Flux is positive if it is from blood to dialysate side.

(ii)

P includes a partition coefficient for species between blood and membrane

phase as well as diffusivity of toxic species through the membrane.

(iii)

C is concentration of species in feed side and species on dialysate side is

extremely dilute.

Therefore, the third boundary condition is

D

C
= PC
y

at y = h

(3.305)

Here both terms in the above equation are positive. This is a standard eigen value
problem or Sturm-Louiville problem.
We define non-dimensional forms as:
C =

C
;
C0

y =

y
;
h

x* = x

Ph
D
= Membrane Biot number
; P* =
2
D
u0 h

If we substitute the above non-dimensional numbers in the above equation we get

*
3
2C *
*2 C
1 y
= *2
2
x*
y

(3.306)

Boundary conditions are:

C* = 1

at x* = 0,

y*

(3.307a)

C *
=0
y*

at

y* = 0,

x*

(3.307b)

C *
+ P*C * = 0
y*

at

y* = 1,

x*

(3.307c)

Analytical solution of the above equation is:

Joint Initiative of IITs and IISc - Funded by MHRD

Page 94 of 120

NPTEL
Novel Separation Processes

n
2

C * = Am exp m2 x* anm y*
3
n =0
m =1

(3.308)

It is a series solution where, eigen values = m .

Rate of removal of pollutants is given as,
M = 2u0 hW ( C0 Ccm )

(3.309)

(in Kg/s)

Here, Ccm = cup-mixing concentration and is given as,

3 y2
0 2 u0 1 h2 C ( x, y ) dy
=
H
3 y2
0 2 u0 1 h2 dy
H

Ccm

(3.310)

The non-dimensional cup-mixing concentration becomes,

*
Ccm
=

Ccm
C0

2 2 x*
anm
= 3 Am exp m

3 n =0 ( n + 1)( n + 3)

m =1

(3.310)

Here eigen values (m) are roots of polynomial of the form,

1
2 5

0 = P* + P* m2 + + P* m4
3 12
20 45
Here, a0 m = 1;

a4 m =

a1m = 0;

m2 ( 2 + m2 )
24

Am =

a2 m =

a5 m = 0;

m2
2

a6 m =

(3.311)
a3m = 0

m4

(14 + ) ;
720
2
m

a7 m = 0

anm

( n + 1)( n + 3)
n=0

p=n

a pm a n p m

( n + 1)(( n +) 3)
n =0 p =0

Joint Initiative of IITs and IISc - Funded by MHRD

Page 95 of 120

NPTEL
Novel Separation Processes

Simplifications:
Exponential terms decay rapidly and so, use first eigen value,
P*
=
2 5 *
+ P
3 12
2
1

(3.312)

and consider first term of series. A typical plot of cupmixing concentration with axial
distance according to Eq.(3.310) is shown in Fig. 3.25.

*
log Ccm

P*= 1
P* 0

P*= 0.1

2
P *x*

Px
=
u0h

Fig. 3.25: Variation of cupmixing concentration with axial distance

Simplistic approach:
Take the flow as plug flow. This can be realized at higher flow rate of blood in the feed
chamber. Therefore, the assumptions are:
1. Plug flow
2. dialysate is dilute
3. steady state
The solute mass balance in the feed chamber is given as,

Joint Initiative of IITs and IISc - Funded by MHRD

Page 96 of 120

NPTEL
Novel Separation Processes

u0

Take, C * =

C
2C
=D 2
x
y

(3.313)

C
y
; y* = , x*=x/L, the above equation is made non-dimensional.
C0
h
A

C * 2C *
= *2
x*
y

(3.314)

u0 h 2 1
d
Where, A =
= ReSc e .
DL 4
L
The nondimensional boundary conditions become,
At y* = 0;

C *
=0
y*

At y = h;

C
+ PC = 0
y

(3.315b)

At y* = 1;

C *
+ PemC * = 0
*
y

(3.315c)

Where,
At x* = 0;

Pem =

(3.315a)

Ph
D

C* = 1

(3.315d)

( ) ( )

Using a separation of variable type solution, C * = X x* Y y*

A dX 1 d 2Y
=
= n2
*
*2
X dx Y dy

Therefore, the X varying part is

(3.316)

n2 *
dX
= dx
X
A

(3.317)

The solution is

n2 x*
X = c1 exp

(3.318)

The y-varying part is ,

d 2Y
+ n2Y = 0
*2
dy

(3.319)

Joint Initiative of IITs and IISc - Funded by MHRD

Page 97 of 120

NPTEL
Novel Separation Processes

Y = c3 sin n y * + c4 cos n y*

The solution is

(3.320)

Boundary conditions are

At y* = 0;

dY
=0
dy*

(3.321a)

At y* = 1;

dY
+ PemY = 0
dy*

(3.321b)

So after applying boundary condition we get

Y = c4 cos n y*

(3.322)

And for n we get,

n tan n = Pem

(3.323)

The above equation is called Transcendal equation.

The final solution is

2 x*
C * ( x* , y * ) = cn cos ( n y * ) exp n
A
n =1

At x* = 0;

Initial condition is

(3.324)

C* = 1

1 = cn cos ( n y* )

(3.325)

n =0

The constant cn can be evaluated using the orthogonal properties of cosine

functions.
1

cos ( y ) dy
*

cn =

0
1

cos ( y ) dy
2

(3.326)
*

Numerator = cos n y dy =
0

sin n y*

sin n

Joint Initiative of IITs and IISc - Funded by MHRD

Page 98 of 120

NPTEL
Novel Separation Processes
1

Denominator =

1
1 + cos 2n y* dy*
2 0

1 sin 2n
1 +

2
2n

But from the transcendal equation we finally get

Denominator =

cn =

So,

1 Pem2 + Pem + n2

2 Pem2 + n2

2sin n

Pem2 + n2
Pem2 + Pem + n2

(3.327)

2 x*
C * x* , y * = cn cos n y * exp n
A
n =1

Thus, the cupmixing concentration becomes,

n2 x*
*
*
exp
=
Ccm
x
c
( ) n A cos ( n )
n =1

Ccm =

V ( y ) C ( x, y ) dA
V ( y ) dA
x

u C ( x, y ) dy.W
0

u0 .Wh
h

Ccm =

1
C ( x, y ) dy
h 0
1

*
Ccm
= C * x* , y* dy*
0

n2 x* 1
*
= cn exp
cos n y
A 0

By simplifying the above equation we get,

*
cm

(x ) = c
*

sin n

n2 x*
exp

Joint Initiative of IITs and IISc - Funded by MHRD

(3.328)

Page 99 of 120

NPTEL
Novel Separation Processes

( )

M x* = Rate of removal of pollutant

= 2u0 hW ( C0 Ccm )

2 x*
sin n
exp n
= 2u0 hWC0 1 cn
n
A

(3.329)

Mass Transfer Coefficient (MTC):

k ( Ccm + Cm ) = D

*
k Ccm
+ Cm* =

Sh =

C
y

(3.330)
y* = h

D C *
h y*

y* = h

C *
y*

kh
y* = h
=
*
D Ccm + Cm*
n2 x*
C *
*

sin
exp
=

c
y
(
)
(
)

n n
n
y*
A

C *
y*

2 x*
= cn n sin n exp n
A

y* =1

For n=1 (first eigen value):

C *
y*

*
cm

2 x*
= c11 sin 1 exp 1
A

y* =1

= c1

sin 1

12 x*
exp

2 x*
Cm* x* = c1 exp 1 cos n
A

( )

2 x*
sin n
*
= c1 exp 1 cos n
Cm* Ccm

n
A

Joint Initiative of IITs and IISc - Funded by MHRD

Page 100 of 120

NPTEL
Novel Separation Processes

Sh =

Sh =

2 x*
c1 1 sin 1 exp 1
A

2 x*
c1 exp 1
A

sin n
cos n
n

1 sin 1
cos n

(3.331)

sin n

Solved Problems
1.

The performance of an osmotic pressure controlled ultrafiltration is investigated. The

pure distilled water flux is found to be 4.14x10-6 m3/m2.s at 276 kPa and 8.28x10-6
m3/m2.s at 552 kPa pressure. Under high stirring speed and at 138 kPa pressure and for
0.5 kg/m3 solute concentration, the permeate concentration is obtained as 0.04 kg/m3. The
2

osmotic pressure of the solute is given as, = 37.5 10 c + 10c , where, is in Pa and c
is in kg/m3.
Steady state cross flow UF of the solute is now conducted in a rectangular cell with
length 1 m, equivalent diameter 1 mm, feed concentration 10 kg/m3, transmembrane
pressure drop 345 kPa, cross flow velocity 0.5 m/s and solute diffusivity 10-11 m2/s. Find
out the permeate flux and permeate concentration. How much membrane area is required
to produce 500 L/hr of filtrate under the same operating conditions with this solute?
Assume validity of the film theory and the solute concentration in the permeate and the
membrane surface can be related through real retention. Assume viscosity and density are
those of water. Use: Sh=1.85(ReScde/D)1/3
Solution:
1) Pure water flux data:
Membrane permeability (Lp) is determined,

Joint Initiative of IITs and IISc - Funded by MHRD

Page 101 of 120

NPTEL
Novel Separation Processes

J w = Lp P
Lp =
Lp =

Jw
P

1 4.14 106 8.28 106

+
= 0.015 109
2 276 103
552 103

L p = 1.5 1011
Rr

( at high stirring speed )

R0

Rr = 1

m
pa.s

0.04
= 0.92
0.5

= 37.5 102 C + 10C 2 = B1C + B2C 2

L = 1 m; d e = 103 m; C0 = 10 kg / m3 ; P = 3.45 105 pa.
u0 = 0.5 m / s; D = 1011 m 2 / s

Mass transfer coefficient:

1/ 3

kd
d

Sh = e = 1.85 Re Sc e
D
L

1/ 3

D u d d
k = 1.85 0 e . . e
de D L
1/ 3

D u d 2
= 1.85 0 e
d e DL
1/ 3

u D2
k = 1.85 0
de L

1/ 3

0.5 1022
= 1.85

3
10 1
J = K ln
K ln

Cm C p
C0 C p

= 6.82 107 m / s

= Lp ( P )

Cm Rr
= L p ( P )
C0 Cm (1 Rr )

Joint Initiative of IITs and IISc - Funded by MHRD

Page 102 of 120

NPTEL
Novel Separation Processes

= B1 ( Cm C p ) + B2 ( Cm2 C p2 )
2
= B1Cm Rr + B2Cm2 1 (1 Rr )

6.82 107 ln
ln

0.92Cm
= 1.5 1011 3.45 105 3450Cm 9.94Cm2
10 0.08Cm

0.92Cm
= 7.6 1 0.01Cm 2.88 105 Cm2
10 0.08Cm
Cm

11

20

50

60

LHS 0.104 0.784 2.036 2.36

RHS

Cm

6.7

3.25

2.25

58 kg / m3

C p = Permeateconcentration = Cm (1 Rr )

= 0.08 58 = 4.64 kg / m3
J = K ln

Cm Rr
58 0.92
= 6.82 107 ln
C0 C p
10 4.64

= 1.57 106 m3 / m 2 .s
Productivity = 500 L/hr
=

500 106 103 3

m / s = 1.39 104 m3 / s
3600

Area required =

2.

1.39 104
= 88.5 m 2
6
1.57 10

Consider separation of 10kg/m3 concentration of a protein solution using ultrafiltration.

Filtration is gel layer controlled, with gel concentration 300 kg/m3. Filtration takes place
in a thin channel with equivalent diameter 2 mm and width 4 cm. The cross flow velocity

Joint Initiative of IITs and IISc - Funded by MHRD

Page 103 of 120

NPTEL
Novel Separation Processes

is 0.5 m/s and protein diffusivity is 2 x 10-11 m2/s. If the filtrate rate is 100 L/day, find the
length required of the membrane module.
Use the following correlations to estimate mass transfer coefficient:
1

d 3

Sh = 1.86 Re Sc e for laminar flow

L

Solution:
C0 = 10 kg / m3 ; Cg = 300 kg / m3 ; d e = 2 103 m; W = 4 cm

u0 = 0.5 m / s; D = 2 1011 m 2 / s
Filtrationrate = 100 L / day =

100 103 106 m3

24 3600
s

= 1.157 106 m3 / s
J = K ln

Cg
C0

8.59 107 300

ln
10
L1/ 3

2.92 106
J=
L1/ 3

A = W .L = 0.04 L m 2
J . A = Filtration rate

2.92 106
0.04 L = 1.157 106
L1/ 3
2

L3 = 9.9058
L = 31 m

3.

Consider, a batch gel layer controlling cross flow ultrafiltration. The permeate is not
recycled back. The feed is getting concentrated and the feed volume is reduced as time of
operation progresses. Assume, film theory is valid in the cross flow ultrafiltration unit.

Joint Initiative of IITs and IISc - Funded by MHRD

Page 104 of 120

NPTEL
Novel Separation Processes

The mass transfer coefficient in the ultrafiltration unit is 2 x 10-5 m/s, initial feed
concentration is 1 kg/m3. Effective filtration area of the membrane is 0.2 m2. Gel layer
concentration of the solute is 500 kg/m3. Initial feed volume is 5 liters. Assume that in the
permeate, there is no solute present and always the retentate concentration is much less
than gel concentration (cb<<cg). Find the time required concentrating the feed
concentration 10 times and what is the volume remaining in the feed chamber?

Solution:
C0 = 1 kg / m3 ; Cg = 500 kg / m3 ; K = 2 105 m / s;

A = 2 m2

V0 = initial feed volume = 5 L.

J = K ln

Cg

(A1)

Cb

Overall mass balance,

d
( V ) = JA p
dt
By assuming

dV
= JA
dt

(A2)

Overall species balance,

d
( CV ) = 0
dt
C0V0 = CbV

(A3)

From equation (A2),

Cg
d C0V0

= KA ln
dt Cb
Cb
Cg

1 dCb
C0V0 2
= KA ln
Cb
Cb dt

Joint Initiative of IITs and IISc - Funded by MHRD

Page 105 of 120

NPTEL
Novel Separation Processes

dCb KA 2 Cg
=
Cb ln
dt C0V0
Cb
dCb KA 2 Cb
=
Cb ln
dt C0V0
Cg

KA 2 Cg Cb
=

Cb ln 1
Cg
C0V0

Cg Cb
KA 2
=

Cb ( )
Cg
C0V0
KA Cb
=
1
C0V0 Cg

2
Cb

KA 2 Cb3
=

Cb
Cg
C0V0
dCb KA 2 Cb3
=

Cb
dt C0V0
Cg
dCb
C
Cb2 1 b
C
g

Cb
10 C0

C0

KA
=
dt
C0V0

Cg

Cb
Cg

KA
Cb2 dCb =
t
C0V0

1
1
KA

=
t
C0 10C0 C0V0
V
t = 0.9 0
KA
= 0.9

5 103 106
= 1125 sec
2 105 0.2

18.75 min

C0V0 = CbV

Joint Initiative of IITs and IISc - Funded by MHRD

Page 106 of 120

NPTEL
Novel Separation Processes

V = 5

4:

1
= 0.5 L
10

A macromolecular solution of diffusivity D=6x10-11 m2/s is flowing through a rectangular

channel of length 1m, width 5cm, half height 1mm at a flow rate of 40L/h and
transmembrane pressure drop 500kPa. Permeability of the membrane is 2.5x10-11 N.s/m3.
The feed concentration of the membrane is 10 kg/m3 and real retention Rr=1.0.
Membrane is placed at the bottom only. =1.5x104C, is in Pa and C in kg/m3. Using
the method in section 3.18, solve the problem.

Solution:
Area of cross section of channel
= 2 wh
= 2*0.05*103 m 2 = 104 m 2
Q = 40 L / h = 1.11105 m3 / s
uo =

1.11 105
m / s = 0.11m / s
104

Equivalent diameter, de = 4h = 4 103 m

Re Sc

de uo d e d e
=
= 2.93 104
D L
L

Osmotic pressure difference,

= 1.5 104 (cm c p )

= 1.5 104 cm Rr
= 1.5 104 cm co

Joint Initiative of IITs and IISc - Funded by MHRD

Page 107 of 120

NPTEL
Novel Separation Processes

= 1.5 105 cm*

1.5 105 *
=
cm = 0.3cm*
p
5 105

From equation (3.123), Pew = B1 1

P
B1 =

Lp Pd e
D

2.5 1011 5 105 4 103

= 833.33
6 1011

Therefore, Pew = 833.33 (1 0.3cm* )

From equation (3.108), 1 =

1 =

833.33

( 2.93 10 )

4 1/3

Pew
1/3

de

Re Sc L

(1 0.3c )
*
m

Rearranging above equation, we get an expression of cm* with respect to suction

parameter,
cm* = 3.33 0.1231

(A1)

Rearrangement of Eq(3.108) and (3.122), the following equation is obtained

1 =

Pew
1/3

de

Re Sc L

1
= 1.85 1 + 0.321 + 0.0212 8.05 104 13 1 * (A2)
cm

Eq(3.122) and Eq(3.123) have to be solved simultaneously.

It may be noted that, cm* >1.0
So, from the Eq(A1),

1 <18.94

Joint Initiative of IITs and IISc - Funded by MHRD

Page 108 of 120

NPTEL
Novel Separation Processes

Combining Eq(A1) and Eq(A2) presents only non-linear equation in 1 ,

1 = 1.85 1 + 0.321 + 0.0212 8.05 104 13 1

1
(A3)
3.33 0.1231

Eq(A1), gives an idea of the guess value of 1 ;

2.14 2.21 2.24

RHS of Eq. (A3) 2.14 2.21 2.24 2.25

Therefore, 1 =2.24
So, Pew =69.05 and w = 69.05 6 104

1
= 1.035 106 m3 / m 2 .s
3
4 10

From Eq.(A1), cm* =3.05

So, cm = 30.5kg/m3
Mass transfer coefficient:
S hL = 103.1
=

k L de
D

Thus, k L = 1.545 106 m / s

5.

A mixture of two solutes, A and B is subjected to osmotic pressure controlled laminar

cross flow nanofiltration in a rectangular channel at steady state. Solute A is completely
and B is partially retained by the membrane. The feed mixture contains 10 ppm of A and
B each. The average feed velocity in the channel is 0.3 m/s and channel geometry is
30cmx5cmx0.15cm. The membrane permeability is 2x10-11 Ns/m3. Real retention of the

Joint Initiative of IITs and IISc - Funded by MHRD

Page 109 of 120

NPTEL
Novel Separation Processes

membrane for solute B is 0.8 and the operating pressure is 300 kPa. Use stagnant film
theory. Calculate the permeate flux and permeate quality. A=6000 cA and . B=6000 cB.
Osmotic pressure is in Pa and c is in kg/m3. DA=2x10-10m2/s and DB=4x10-10m2/s .
Solution viscosity and density are taken to be same as those for water.
Solution:
C A0 = 10 103 = 102 kg / m3 = CB 0
u0 = 0.3m / s;

4h = 4

de

Lp = 2 1011 N .s / m3

L = 30 cm = 0.3 m;

RrB = 0.8;

RrA = 1.0;

P = 300 kPa

DA = 2 1010 m 2 / s;

A = 6000C A ;
J = k A ln

0.5
102 m = 3 103 m
2

DB = 4 1010 m 2 / s

B = 6000CB

CmA C pA
C A0 C pA

= k B ln

CmB C pB
CB 0 C pB

But C pA = 0
So, J = k A ln

CmB C pB
CmA
= k B ln
C A0
CB 0 C pB
1/3

u D2
k = 1.85 0
de L

1/3

0.3 4 1020
k A = 1.85

3
3 10 0.3
= 4.387 106 m / s

1/3

0.3 16 1020
k B = 1.85

3
3 10 0.3
= 6.96 106 m / s
4.387 106 ln

CmA
0.8CmB
= 6.96 106 ln 2
2
10
10 0.2CmB

Joint Initiative of IITs and IISc - Funded by MHRD

Page 110 of 120

NPTEL
Novel Separation Processes

80CmB
ln (100CmA ) = 1.586 ln

1 20CmB

(B1)

J = L p ( P )
J = L p [ P 6000CmA 6000 RrB CmB ]

= 2 1011 3 105 6000CmA 4800CmB

= 6 106 [1 0.02CmA 0.016CmB ]

J = k A ln

CmA
C A0

4.387 106 ln

CmA
= 6 106 [1 0.02CmA 0.016CmB ]
102

ln (100CmA ) = 1.37 [1 0.02CmA 0.016CmB ]

(B2)

From the equation (B1),

ln (100CmA )
80CmB
= exp

1 20CmB
1.586
ln (100CmA )
1 20CmB
= exp

80CmB
1.586

ln (100CmA )
1
1
= + exp

80CmB 4
1.586

80CmB =

CmB =

1
ln (100CmA )
0.25 + exp

1.586

0.0125
ln (100CmA )
0.25 + exp

1.586

(B3)

From equation (B2),

CmA = 0.01exp 1.37 (1 0.02CmA 0.016CmB )

Joint Initiative of IITs and IISc - Funded by MHRD

(B4)

Page 111 of 120

NPTEL
Novel Separation Processes

CmA,guess

0.02

0.03

0.04

CmB from B3

0.014

0.0167

0.187

CmA from B4

0.039

0.039

0.039

CmA = 0.039 kg / m3
CmB = 0.0187 kg / m3

In permeate = 0.0187 0.08 = 1.496 ppm

J = 4.387 106 ln

0.039
0.01

= 5.97 106 m3 / m 2 .s

6.

Water is flowing through a rectangular channel of half height 1 mm and width 8 cm. The
inlet flow rate is 40 L/h. Inlet transmembrane pressure of 500 kPa results in a constant
permeate flux of 2x10-5 m3/m2.s. The length of the channel is 2m.
(a) Find the axial pressure drop across the module.
(b) What is the transmembrane pressure drop at the module exit?
(c) Find out the fractional recovery of the feed.
(d) What is the flow rate at the channel exit?
(e) For a fractional recovery of 0.92 what is the length of the module is required? What is
the axial pressure drop and flow rate at the module exit?

Solution:
h=10-3 m; w=0.08 m; Pi=5x105 N/m2; L=2m; vw=2x10-5 m3/m2.s
Qi=40 l/h = 1.11x10-5 m3/s.
(a) From Eq.(3.191),

Joint Initiative of IITs and IISc - Funded by MHRD

Page 112 of 120

NPTEL
Novel Separation Processes

Paxial =

2v wx
3
Qi x1 w
3
2h w
2Qi

3 10 3 1.11 10 5 2 2 10 5 0.08 2
1

3
2
1.11 10 5
10 3 0.08

= 416.25(1 0.29)
= 296 Pa

(b) PL= transmembrane pressure drop at the outlet

= Pi-296 = 499.7 kPa

(c) Fractional feed recovery in the permeate

f =

2v w wL 2 2 10 5 8 10 2 2
= 0.576
=
Qi
1.11 10 5

(d) Flow rate at the channel exit

Differentiating Eq.(3.191) with respect to x the following expression is obtained

d P 3 Qi 2vw wx
=
1

dx
2 h3 w
2Qi

Combining the above equation with respect to Eq.(3.190), the expression of cross flow
rate is obtained.
2v wx
Q = Qi 1 w
2Qi

Substituting the values of various parameters in the above equation, at the channel exit,
i.e., at x=L,

Joint Initiative of IITs and IISc - Funded by MHRD

Page 113 of 120

NPTEL
Novel Separation Processes

2 2 10 5 0.08 2
Q = 1.11 10 (1
)
1.11 10 5
= 5 10 5 m 3 / s = 28.5 L / h
5

(e) f = 0.92 =

2v w wL
Qi

For the same flow rate, channel width and suction rate, length required,
L = 0.92

Paxial =

1.11 10 5
= 3.19 3.2 m
2 2 10 5 0.08

2v wx
3
Qi x1 w
3
2h w
2Qi

By putting the numbers in the above equation, axial pressure drop becomes
Paxial = 360 Pa
Similarly, the new flow rate becomes, 21.6 L/h

7.

In the above problem, if the half channel height of 0.5 mm. find the (a) axial pressure
drop, (b) transmembrane pressure drop at the module outlet, (c) fractional recovery of the
feed, and (d) flow rate at the channel outlet.

Solution:
(a) In this case, axial pressure drop will be changed.
P axial =

3
10 3

1.11 10 5 2(1 0.29 )

2 (0.5 10 3 )3 0.08

= 2364.3 Pa

(b) The pressure drop at the channel outlet:

PL =5x105-2364.3=4.97x103 Pa=497 kPa

Joint Initiative of IITs and IISc - Funded by MHRD

Page 114 of 120

NPTEL
Novel Separation Processes

(c) since, in the expression of fractional feed recovery, channel height does not appear,
f remains as 0.576
(d) Similarly, the flow rate at the channel outlet remains same as 28.5 L/h.

Note: The pressure drop across the module increases as the channel height decreases due

to increase in the flow rate.

8.

A protein solution of concentration 0.2kg/m3 is concentrated by ultrafiltration. The

operating pressure at the channel inlet is 5x105 Pa, width 0.08m and channel half length is
1mm. The membrane permeability is 10-11 N.s/m2. The feed flow rate at the channel
entrance is 40 L/h. The length of the channel is 2m. Find out (a) Pressure drop across the
channel (b) Fractional recovery of feed (c) Velocity and (d) concentration at channel
outlet.

Solution:
Pi= 5x105 Pa ; w= 0.08m ; h= 10-3m ; Lp= 10-11 N.s/m2 ; Q= 40 L/h = 1.11x10-5 m3/s ;
L= 2m
uo =

Q
1.11 105
=
= 0.07 m / s
2hw 2 103 0.08
3 Lp
3 103 1011
=
= 5.48 103
3
9
h
10
3 Qi
3
103 1.11105
=

= 0.076
2 h3 wPi 2 109 0.08 5.48 103 5 105

L = 0.011;sinh( L) = 0.011;cosh( L) = 1.0

(a) Pressure drop

Joint Initiative of IITs and IISc - Funded by MHRD

Page 115 of 120

NPTEL
Novel Separation Processes

P( L)
= cosh( L) sinh( L) = 1 0.076 0.011 = 0.999
Pi
P( L) =499.50kPa
So, P |axis = 5 105 499500 = 500 Pa
(b)Fractional feed recovery
f =

2 wLp P
Qi

[sinh( L) (cosh( L) 1)]

2 0.08 1011 5 105

=
[0.011 0] = 0.147
1.11105 5.85 103
(c)Velocity at the outlet
L p Pi
u ( L)
= 1
[sinh( L) (cosh( L) 1)]
uo
huo
= 1

1011 5 105
[0.011 0] = 0.86
103 5.85 103 0.07

So, velocity at the outlet is 0.06m/s

(d) Concentration at the outlet

u
1
C ( L)
= o =
= 1.16
Co
u ( L) 0.86
So, the concentration at the outlet is 0.232 kg/m3

9.

In a gel layer controlling ultrafiltration, the specific gel layer resistance is given as,

= 0 (P )0.3 , where, is in m/kg, 0 = 2.56 1016 and P is in Pa. Gel layer

thickness is given as L = L0 (P ) , where, L0=5.74x10-11 and P is in Pa. Gel layer
0.8

Joint Initiative of IITs and IISc - Funded by MHRD

Page 116 of 120

NPTEL
Novel Separation Processes

density is 1050 kg/m3 and gel layer porosity is 0.7. Membrane permeability is 5x10-11
m/Pa.s.
Find the value of limiting operating pressure and what are the values of flux and gel layer
resistance at this pressure?
Solution:
J=

P
( Rm + Rg )

Rg = (1 g ) g L
= 0 ( P )

0.3

(1 ) L ( P )
g

0.8

= 0 (1 g ) g L0 ( P )

1.1

= 4.36 108 ( P )

1.1

Rg = Rg 0 ( P ) ,
1.1

J=

Rg 0 = 4.36 108

where

Rm + Rg 0 ( P )

1.1

0.1
P.Rg 0 1.1( P )
dJ
1

=0
=

2
1.1
1.1
d P Rm + Rg 0 ( P )
Rm + Rg 0 ( P )

) (

Rm + Rg 0 ( P ) = 1.1Rg 0 ( P )
1.1

0.1

Rm = 0.1Rg 0 ( P )

1.1

( P )

1.1

10 Rm
Rg 0
1

10 Rm 1.1 10 2 1015
P =
=
8
R
4.63 10
g0

0.91

= 8.876 106 pa
= 8876 kPa

Joint Initiative of IITs and IISc - Funded by MHRD

Page 117 of 120

NPTEL
Novel Separation Processes

PLim = 8876 kPa

J=

8876 103
2 1015 + 4.63 108 88761.1

= 4.41106 m / s
Rg = 2 1016 m 1

10.

In a steady state, counter current dialyzer of rectangular cross section, urea is removed by
pure water as dialysate. Inlet feed concentration is 1500 mg/L. Feed and dialysate flow
rates are 18 L/h and 90 L/h, respectively. Durea in the membrane is 10-12 m2/s and in the
bulk is 10-10 m2/s. membrane thickness is 1 micron. Feed and dialysate chambers are
identical in shape. Width of each channel is 5 mm and height 5 mm. In the feed side, the
urea concentration has to be reduced from 1500 (inlet) to 300 mg/L (outlet).
Find the membrane area required. Neglect mass transfer resistances on both sides.
CiFi = 1500 mg / l ;

CiFe = 300 mg / l ;

VF = 18 l / h;

Dim = 1012 m 2 / s;

Di = 1010 m 2 / s;

L = 106 m;

h = height = 5 mm;

VD = 90 l / h

W = 5 mm;

CiDi = 0

m = VF CiFi CiFe = VD CiD0 CiDi

18 (1500 300 ) = 90 CiD0 0

18 1200 = 90CiD0
CiD0 =
(i)

18 1200
= 240 mg / l
90

Neglecting mass transfer resistance on both sides

k0 = overall mass transfer coefficient
Dim 1012
=
= 6 = 106 m / s
10
L

Joint Initiative of IITs and IISc - Funded by MHRD

Page 118 of 120

NPTEL
Novel Separation Processes

( C )i ( C )0
( C )i
ln
( C )0

( C ) LMTD =
( C )in = CiF

CiD0 = 1500 240 = 1260 mg / l

( C )out = CiF

( C ) LMTD =

CiDi = 300 0 = 300 mg / l

1260 300
= 668.95 mg / l
1260
ln
300

mg
l
m = VF CiFi CiFe = 18 1200
l
h

= 18 1200

mg
= 21.6 g / h
h

m = k0 Am ( C ) LMTD

= 106 Am 668.5

m 2 mg
.m .
s
l

= 106 668.5 Am 10-3 3600

g m3
h 103 m3

= 106 668.5 3600 Am g / h

= 2.41Am g / h

21.6 = 2.41Am
Am = 8.96 m 2

Joint Initiative of IITs and IISc - Funded by MHRD

Page 119 of 120

NPTEL
Novel Separation Processes

References:
1. M C Porter, Handbook of Industrial Membrane Technology, Crest Publishing House,
2005.
2. S. De and P. K. Bhattacharya, "Prediction of mass transfer coefficient with suction in the
application of reverse osmosis and ultrafiltration", Journal of Membrane Science, 128,
119-131, 1997.

3. V. S. Minnikanti, S. DasGupta and S. De, "Prediction of mass transfer coefficient with

suction for turbulent flow in cross flow ultrafiltration", Journal of Membrane Science,
157, 227-239, 1999.

4. R. Ranjan, S. DasGupta and S. De, Performance prediction of membrane modules

incorporating the effects of suction in the mass transfer coefficient under laminar flow
conditions, Separation & Purification Technology, 45, 109-118, 2005.

5. C. Prabhavathy and S. De, Treatment of fatliquoring effluent from a tannery using

membrane separation process: experimental and modeling, J Hazardous Materials, 176
(2010) 434-443

6. C. Prabhavathy and S. De, Estimation of transport parameters during ultrafiltration of

pickling effluent, Separation Science and Technology, 45, (2010) 1-10

Joint Initiative of IITs and IISc - Funded by MHRD

Page 120 of 120