CAR FILTERS

Chapter 1
General describtion of
filters
1. Definition
Filtration is a process of separating dispersed particles from a dispersing
fluid by means of porous media. The dispersing medium can be a gas (or
gas mixture) or a liquid. Particles can be solid or liquid for gas medium and
solid for liquid medium.
Upstream Downstream
Filter
Channel
wall
Dispersed
particles
Dispersing
fluid
Particles deposited
inside the filter
Filter
thickness
Face of the filter with filter cake of
deposited particles
2. Types of filtration
Concerning to filtration surrouding:
Air filtration / Liquid filtration
Concerning to size of filtered particles:
Macrofiltration for particle size dp: 10
-6
m < dp
Microfiltration 10
-7
< dp < 10
-6

Ultrafiltration 10
-8
< dp < 10
-7

Nanofiltration 10
-9
< dp < 10
-8

Reverse osmosis dp < 10
-9

Concerning to filtration mechanism:
Flat filtration / Depth filtration
2.1 Air / liquid filtration
Examples of air filtration:
respirators, air ventilation systems (air condition, air
cleaning etc), vacuum cleaners, industrial filters for
incineration, power plants, chemical processing, paint
boxes, car filters (cabin filters, engine filters, exhaust
filters)
Examples of liquid filtration:
drink water treatment, waste water treatment, chemical
processing, batteries, industrial filters (cutting
operations, cooling liquids, spunlace), car filters (oil
filters, fuel (petrol) filters)
2.2 Relative size of common filtered particles
2.3 Surface filtration
All particles which are bigger than pores are captured on the
flat filter surface. It is typical for example for fabric or spunbond
filters. Thus for these filters the pores distribution and permeability are
important properties. Surface filtration is common for liquid filtration.
Surface filters are described in subject High funcional textiles

Direction
of flow
Textile filter expressed as a set
of cylinders placed in parallel
Captured
particles
2.4 Depth filtration
Depth filter are able to capture particles that are too small to be sieved
out as in flat filtration. Particles, which can be a lot of smaller than the
distances between the fibers penetrate into the fiber structure. Filtered
particles are captured in terms of the filtration mechanisms. This type
of the filtration process is importatn for the most of filter applications.
Next chapters about filtration variables, properties and mechanisms
refer first of all to the deep filtration.
Direction
of flow
Textile filter expressed as a set
of cylinders placed in parallel
Captured
particles
3. FILTRATION THEORY
Filtration variables
Filter variables
Flowing medium
variables
Captured particles
variables
Filtration properties
Efficiency
Pressure drop
Lifetime
Resistivity against
environment
Others (permeability,
porosity...)
Filtration mechanisms
Diffusion deposition
Direct interception
Inertial deposition
Electrostatic
deposition
Sieve effect
Its simple to say what is filtration but difficult to describe relations between
filter properties and filtration variables which influence the filtration process
This is what
we can
change
This is what
we try know
This is what
we need
Filter efficiency
It is the ratio of particles captured by a filter over the total number
of particles found in the air upstream of the filter. Filter efficiency
can either be based on specific particle size ranges or based on the
total number of particles of all sizes.
3.1 Filtration properties I.
100 . 1
2
1
|
|
.
|

\
|
=
G
G
E
Efficiency can be defined by formula 1,
where G
1
is an amount of penetrated
particles (which havent been captured) and
G
2
is total amount of particles upstream formula 1.
Expression G
1/
G
2
is named Penetration of filter

Efficieny is changing during the filtration process (see chapter 6.3.4
Nonstationary filtration)
Pressure drop
Pressure drop indicates the restance to flow. It is defined as a difference between the pressure
of flowing media upstream and downstream of the filter. For expression of pressure drop is
necessary to assign air flow or air velocity (linear relation).
Ap = p
1
- p
2,

where p
1
is pressure drop upstream and p
2
downstream
of the filter. Pressure drop is changed during the filtration
proces (see chapter 6.3.4 Nonstationary filtration).

Filtration properties II.
Filter lifetime
Filter lifetime determines the time when the filter must be removed. It is defined as a time or
as an amount of the filtered particles, which are loaded into the filter until the filter is full.
According to EN 779 standard the filter lifetime is defined as a Dust holding capacity:
J = E
s
.m
p

where E
s
is mean filter efficiency and m
p
is the amount of the
particles loaded into the filter until the final pressure drop (250
or 400 Pa) was reached
Permeability
It is the ability of a material to allow the passage of a liquid or gas through porous material. It
is possible to find more defininitions, whic depend on the level of simplification:
1) According to EDANA 140.1 standard it is defined by formula:
where M
s
is permeability (l/dm
2
/min), Q is the flow
(l/min)and A is the filter surface. Permeability is tested with the
pressure drop 196 Pa (98,1 Pa for some standards)
2) According to the Darcys law the permeability is defined by formula:
where K is permeability (m/Pa/sec) and Ap is the pressure
drop (Pa).
3) According to the Darcy,s law is possible to define permeability as a permeability
coefficient defined by formula:
where k
1
is the permeability coefficient (m
2
), q is the
dynamic viskosity (Pa.sec), and h (m) is the thickness of the
filter.
Filtration properties III.
A
Q
M
S
=
p A
Q
K
A
=
.
p A
h Q
k
A
=
.
. .
1
q
4. According Hagen-Dupuit-DArcys model is permeability defined as:


where K
3
is permeability coefficient and C is form coefficient.
This model is suitable for higher flow of viscose liquid (such as water etc). When we
compare HDD model with DArcys law, the main difference is nonlinear relation between
flow and pressure drop.
Permeabilityof laminated textiles
For simple DArcys law it is possible to deduce relation between the permeability of one
layer and more layers. For most of the applications we can assume that the flow through
the laminated textile is the some as flow through one layer. Than the total pressure drop
and total permeability are defined:

and ,

where Ap
i
and K
1i
are pressure drop and peremability coefficients of each layers
2
3
.
. .
.
.
.
Q
A
h C
Q
A K
h
p
q
+ = A

A = A
i
i t
p p

=
i
i total
K K
1 1
1 1
Filtration properties IV.
Porosity and pore size
Porosity of porous medium is defined as a percentage of the porous material volume not
occupied by fibers.
Very important is size or size distribution of pores, which depends on the pore definition and
on the used test method.
Testing methods:
1. Image analysis of 2D microscopic wiew direct method
2. Sifting of defined particles through the textile
3. Penetration of liquid agent into the textile relation between pore size and surface
tension of liquid.
a) Wetting agent is pushed away from textile due to pressured gas Bubble point
method
b) Non-wetting agent is pushed into the textile Mercury porosimetry
For more informations see subject High functional textiles.
Filtration properties V.
Description of simple Bubble point method:
We assume circular pores. Wetting liquid (wetting angle = 0) try to go through the pores
due to wetting force F

= t.D. . Against this force we can act by pressured gas (F

= . t . D
F
p
= p . A
pore

textile
Wetting agent
bubble
Filtration properties VI.
3.1.1 Change of filtration properties
Statinary and nonstacionary filtration
It is important that the filtration properties are changing during the filtration process. A
captured particle, since it occupies a finite space, becomes part of the filter structure,
able to contribute both to pressure drop and to filtration efficiency. When we neglect
this assumption the filtration process is named stationary. It is possible in the
beginning of the filtration process. When we assume that the deposited particle
influences filter properties the filtration process is named nonstationary [Pich,
1964]. Secondary proceses of nonstationary filtration are:
1. Filter clogging particles fill the filter structure
increase of pressure drop
increase of filter efficiency
2. Particle disengagement
decrease of filter efficiency
3. Capillary phenomena
flushing of drops
formation of fluid layers in placed where the fibers are spiced
condensation of water
4. Loss of electric charge
decrease of filter efficiency
5. Filter destruction
3.1.2 Test method of filtration properties:
Tested properties are efficiency, fractional efficiency, pressure drop, pressure drop vs. air flow,
filter lifetime etc... Properties are tested as initial or during filtration process. Methods are
differ in the particle substance (electrical properties, adhesion etc...), particle size (coarse/fine),
particle size range (monodisperse, polydisperse), particle concentration etc...
1) Synthetic dust
The dust is blend prepared from melted anorganic (and organic) particles. The most known is
ASHRAE dust that has the some parameters as the dust around Arizona roads [ASHRAE 52,2,
1999]. It is used for coarse filters (particles are coarse and polydisperse). It is possible to test
change of properties during the filtration process and filter lifetime. Dust is measured by
weighting method. This method is very popular and easy to use. However, it is open to
criticism because weight measurements give predominantly the weight of the largest particles
in the sample. Used standards are: EN 779 [EN 779, 200], ASHRAE 52,2 etc...
2) Athmospheric dust spot efficiency
In the Atmospheric Dust Spot Efficiency ambient outdoor atmospheric air is passed through
the unit being tested and samples are taken at the inlet and outlet of the unit to evaluate its
collection efficiency on the dust particles suspended in the atmosphere. This test is replaced
with DEHS aerosol method because athmosperic air composition is changing. Used standard
was older version of EN 779 [Gustavsson, 1999] .

3) Oil aerosols (DEHS, DOP, paraffin oil)
As the test matter is used aerosol from liquid oily substances. The most known are:
dioctylphtalate (DOP), diethylhexylsebacate (DEHS) and paraffin oil. Two types of oil
aerosol are known: Cold and hot. If the oil is dispersed and dryed in cold ambient conditions
(Laskin nozzle) then the size range of particles is wider (polydiperse aerosol). If the oil is
dispersed and dryed in hot ambient conditions then is possible to obtain monodisperse
particles (0,1-0,3 m). Particles are analyzed by laser particle counter or by spectrofotometric
method. It is possible to detect efficiency of selected particle size (except paraffin oil).
Particles are insenzitive to electrostatic field. Initial values of This method is used for fine
and high efficient filters HEPA (high efficiency particulate air filter) and ULPA (ultra low
penetration air filter) filters.
4) NaCl aerosol
Sodium Chloride aquelous solution is dispersed and dryed. These polydisperse particles have
mean size 0, 65 m and their penetration through the filter is analysed by spectrofotometer.
This method is suitable for quick test of high efficient filters (respirators especially). Used
standards are: BS 4400 [BS 4400, 1969], EN 143 [EN 143, 2000], etc...
5) Methylen blue test
The solution of methylen blue is dispersed and dryed. Particles are analysed by comparing of
the blue colour intensity upstream and downstream the filter. It is suitable to high efficient
filters. By reason of narow gauge usage is replaced by sodium chloride aerosol test.
Summary of test methods:
method Test standard
name
particle substance particle
diameter
(m)
particle
preparation
particle
detection
ANSI/AHAM

Arizona roads dust 0,5 - 3 aerosol
generator
aerodynamic
sorter
ASHRAE
EN
CAN
72% fine dust
23% molocco black
5% cotton linters
-
synthetic
dust
ISO
SAE
Testing dust 2 125
10 - 40
injector

weighting
method

athmospheric
dust
ASHRAE
CAN
Athmospheric dust Cca. 0,3 straight from
air
opacitometer
(light opacity)
0,3
0,2 0,3
evaporation,
condensation

ASTM
ASME/ANSI
IES
MIL-STD
UL
DOP test;
di-octylphtalate
0,3 2 Laskin
nozzle
0,1 0,3 evaporation,
condensation
EN DEHS aerosol
diethylhexylsebacate
0,2 3

Laskin
nozzle
optical particle
counter,
spectrofotometer

oil aerosol
EN
BS
Paraffin oil;
CP27 DAB7
0,40,26 evaporation,
condensation
photometer of
the light
diffusion
aerosol NaCl BS
EUROVENT
EN
NF
NaCl particles 0,02-2
median
0,6
dispersion,
drying
spectrofotometer
Methylene
Blue test
BS

Methylen blue
particles
- dispersion of
water
solution
blue spot size

3.2 Filtration variables
Filtration variables are divided onto three groups:
1. Variables of filter material
2. Variables of filtered particles
3. Variables of filtration process
3.2.1 Variables of filter material:
Filtration area
Filter thickness
Density and surface density of filter
Uniformity of fibrous material
Parameters of filter material
surface interactions between the filter material and filtered particles
electrical properties
mechanical characteristics (tenacity, elongation...)
resistance against surrounding factors (heat, solvents...)
Parameters of fibers
fiber diameter, fiber fineness
shape of fiber cross-section
fiber surface preparations
Mechanical characteristics
Filter structure
filter density gradient
fiber orientation
3.2.2 Variables of filtered particles
Particle size
Distribution of particle size
Concentration of particles
Shape and surface of particles
Particle density
Electrical properties
3.2.3 Variable of filtration process
Face velocity (speed of filtered particles in front of filter)
Viscosity of the flow
Temperature, pressure, humidity
3.3 Filtration mechanisms
Air (gas filtration) Liquid filtration
Type of filtration Surface
Depth more common
Surface more common
Depth

Mechanisms direct interception
inertial impaction
diffusional deposition
capture by electrostatic
forces
sieve effect
direct interception
inertial impaction
sieve effect

3.3.1 Filtration mechanisms of depth filtration
R
fiber
charge on the
fiber surface
diffusional
deposition
inertial
impaction
direct
interception
capture by
electrostatic
forces
streamlines (air
moving trajectory)
Total filtration efficiency

Ec is total efficiency, E
r
is efficiency of direct interception mechanism represented by
parameter N
r
, E
i
is efficiency of inertial impaction represented by Stokes number Stk, E
d
is
efficiency of diffusional deposition mechanism represented by Peclet number Pe and E
e
is
efficiency of electrostatic mechanism represented by the parameter Nq.
( ) ( ) ( ) ( ) { } Nq E Pe E Stk E N E E E
e d i r r c c
, , , =
Mechanisms:
direct interception
inertial impaction
diffusional
deposition
capture by
electrostatic forces
.
Direct interception
Direct interception occurs when airborne particles behave in an entirely passive way with
respect to the airflow. Airborne particles follow the streamline, which in steady state are
independent of the air velocity. Particle will be captured when it is close to the fiber. This
mechanism is independent of air velocity, air viscosity and density. Particle must be small,
because inertial effects and external forces are neglected. This type of mechanism is common
for simple respirators made from fibers of about 20 m, which operate in filration velocity
about 0,04 m/sec. Furthermore interception acts along with other filtration processes.
Parameter of direct
interception:
N
r
= dp/df
(d
p
is particle diameter, d
f
is
fiber diameter)
d
f

fiber
streamlines (air
moving trajectory)
d
p

Relation between parameter N
r
and efficiency of direct interceptiom mechanism:
E
R
~ N
r
2
;
the simpliest relation is: E
R
=N
R
2
/,, more exactly:
where ,=-0,5.ln(c)-0,75 is hydrodynamic factor and m = 2/(3.(1-c))
( )
m
R
R
R
N
N
E
+
=
1 .
2
,
Inertial impaction
Any convergence, divergence or curvature of streamlines involves acceleration of the air, and
under such conditions a particle may not be able to follow the airflow. What particle does
depends upon its mass (inertia) and upon the Stokes drag exerted by the air. Stokes drag is
defined as a force which acts on the moving sferical object inside of viscous liquid: F =
3.t.q.d
p
.v (where F is the force, d
p
is the particle diameter, q is the dynamic viscosity and v is
the face velocity of the airflow).
fiber
inertial
impaction
streamlines (air
moving trajectory)
Intensity of the point particle inertia is determined by Stokes number:


where d
p
is particle diameter, is particle density, U is air face velocity, q is air viscosity
and d
f
is fiber diameter.
f
p
d
U d
St
. . 18
. .
2
q

=
Efficiency of inertial impaction
E
i
depends on the intensity of the
point particle inertia. If inertia is
negligible then E
i
will be zero, if
the inertia is infinite then E
i
will
be 100 %.
Relation between the Stokes number and efficiency of inertial impaction:
For low Stokes number efficiency is lead by direct interception:
E
ir
=E
R
+(2.)
-2
.J.St,
where E
R
is efficiency of direct interception, is hydrodynamic factor dependent on packing
fraction c and J is constant dependent on c and parameter of direct interception Nr.

For high Stokes number efficiency of inertial impaction is defined:
E
I
=1-(/St),
where is constant dependent on flow field.
Diffussional deposition
The trajektories of individual small particles
do not coincide with the streamlines of the
fluid because of Brownian motion. With
decreasing particle size the intensity of
Brownian motion increases and, as a
consequence, so does the intensity of
diffusion deposition [Pich J,1964]. However
the air flow effects on the particles motion
too. Thus the real motion of small particles
depends on Brownian motion and air flow.
Brownian motion is determined by diffusion coefficient D defined by the Einstein equation:


where k
B
is Boltzmann constant, K is Kelvin temperature, q is air viscosity, d
p
is particle
diameter and Cn is the Cunningham correction, which involve aerodynamic slip flow of
particles:

where is mean free path of molecules (at NTP it is 0,065 m) and A, B, Q are constants
(A=1,246; B=0,87; Q=0,42) [Brown RC, 1993].
diffusional
deposition
streamlines (air
moving trajectory)
fiber
p
B
d
T k Cn
D
. . . 3
. .
q t
=
|
|
.
|

\
|
+ + =

. 2
.
. .
. 2
1
p
d B
p
e Q A
d
Cn
Coefficient of diffusional deposition:
Capture of particles by a diffusional deposition will depend on the relation between the
diffusional motion and the convective motion of the air past the fiber. Dimensionless
coefficient of diffusional deposition N
D
is defined:


where d
f
is fiber diameter, U is air flow velocity and P
e
is named Peclet number.
Diffusional capture efficiency:
According to Fokker-Planck equation was aproximated relation between the N
D
(or 1/P
e
) and
diffusional capture efficiency
E
D
= 2,9 .
-1/3
. Pe
-2/3

where is hydrodynamic factor ( = -0,5. ln(c)-0,75 by Kuwabara) [Brown RC, 1993].
Previous equation was verified by experiments with model filters with the some and
observed functional dependance was the some with little different numerical coefficient:
E
D
= 2,7

. Pe
-2/3

When we calculate with the slip flow (see chapter 9) the resulting capture efficiency is bigger.
U d
D
Pe
N
f
D
.
1
= =
Electrostatic forces:
Both the particles and the fibers in the filter
may carry electric charges. Deposition of
particles on the fibers may take place
because of the forces acting between
charges or induced forces. [Pich J, 1964].
The capture of oppositely charged particles
is given by coulomb forces. The capture of
neutral particles comes about by the action
of polarisation forces. We can define three
cases of interaction between particle and
fiber. Used equations were derived from
Coulombs law.
fiber
charge on the
fiber surface
capture by
electrostatic
forces
streamlines (air
moving trajectory)
1. Charged particle, charged fiber
where q is the particle charge, Q is fiber charge per unit lenght
of fiber and x is the distance between fiber and particle.
2. Charged fiber, neutral particles
where D
1
is the dielectric constant of the particle and d
p
is
particle diameter.
3. Charged particles, neutral fiber
where D
2
is dielectric constant of the fiber and d
f
is fiber
diameter.
x
q Q
F
. . 2
1
=
3
3
1
1
2
2
.
2
1
. . 4
x
d
D
D
Q F
p
+

=
( )
1
1
.
. 4
2
2
2
2
3
+

=
D
D
d x
q
F
f
Coefficient of electrostatic mechanism, efficiency of electrostatic mechanism
We can interpret this parameter as a ratio of electrostatic forces to drag forces. From this
parameter were derived equations for efficiency [Pich J, 1964].










B is mechanical mobility of the particle, U
0
is the velocity far form the fiber, d
f
is fiber
diameter, d
p
is particle diameter and q is viscosity

Coefficient of electrostatic
mechanism
Efficiency of electrostatic
mechanism
Charged fiber and
charged particle
Charged fiber and
neutral particle
Carged paricle and
neutral fiber
0 0
. . . . . 3
. . 4
.
. . . 4
U d d
q Q
U d
B q Q
N
f p f
Qq
q t
= =
q t . .
.
.
2
1
.
. 3
4
0
3
2 2
1
1
0
U d
Q d
D
D
N
f
p
Q
+

=
1
1
.
. . . . 3
2
2
0
2
2
0
+

=
D
D
U d d
q
N
f p
q
t
0
. . . . 3
. . 4
.
U d d
q Q
N E
f p
Qq Qq
q
t = =
( ) ( )
2
1
0
2
1
.
Re ln 2
2
q Qq
N E

=
3
1
0
3
1
0
.
2
3
Q Q
N E
|
.
|

\
|
= t
3.3.2 Filtration variables vs.capture efficiency of filtration
mechanisms
Efficiency of
each filtration
mechanisms

Relations how some filtration variables increase or decrease or not affect the
efficiency of each filtration mechanisms

filter
density

fiber
diameter

particle
diameter

particle
mass

face
velocity

viscosity
of air

relative
charge

direct
interception

-

+

|

-

-

-

-

inertial
impaction

|

+

|?

|

|

+

-

diffusional
deposition

|

+

+

-

+

+

-

electrostatic
deposition

-

+

+|

-

+

+

|

3
2
1
|
|
|
|
|
.
|

\
|
+

dp
d
e
p
d
p
Efficiency of
each filtration
mechanisms

Numeric relations between the filter variables and capture efficiency of each
mechanisms

filter
density
c

fiber
diameter
d
f


particle
diameter
d
p


particle
mass


face
velocity
U

viscosity
of air
q

relative
charge
q, Q

direct
interception

-


1/d
f
2


d
p
2


-

-

-

-

inertial
impaction

1/(ln c)
2


1/d
f
or
1 k.d
f


d
p
2
or
1-1/d
p
2


or
1-k/

U or
1-k/U

1/q

-

diffusional
deposition

1/(ln
c)
1/3


1/d
f
2/3


-

1/U
2/3


1/q
2/3


-

electrostatic
deposition

-

1/d
f


1/d
p
or
d
p
2/3
or
1/d
p
1/2


-

1/U or
1/U
1/3
or
1/U
1/2


1/q

q.Q or
Q
2/3
or
q

3.3.3 Filtration mechanism of flat filtration Sieve effect
E
s
= 1 for d
p
> d
pore;
; E
s
= 0 for d
p
< d
pore
,

where E
s
is efficiency of sieve effect and d
pore
is pore diameter.
Relation between fiber and pore diameter according to Neckar [Neckar B.,
2003]:
()

where q is fiber shape factor (zero for cylindrical fibers), c is packing factor, a
and k are constats related to filter structure (usually a is ).
For cylindrical fibers with hexagonal structure is k = 2
-1/2
.

. d
pore f
d =
a
f pore
c
c
q
k
d d
|
.
|

\
|

+
=
1
.
1
.