You are on page 1of 30

Chapter 7

Particles and Particle Clouds

Small particles are often encountered in fan engineering. In air-


conditioning applications, the air drawn from the atmosphere can contain dust
and other contaminants, and droplets may be carried over from humidifiers or
dehumidifiers. Local exhaust systems are used to capture various particles.
High concentrations of particles are moved in pneumatic conveying systems.
Fans and fan systems that are exposed to particles may need special design
features. Protection against erosion and corrosion is discussed in various
application chapters. Drifting in ducts and increased energy requirements are
examined in the conveying chapter. Separation and retention are discussed in
the air-cleaning chapter.

Aerosols
Disperse systems in air have been given the general name aerosol, analo-
gous to the older term, hydrosol, which denotes a disperse system in water or
other liquid. An aerosol is defined as a stable suspension of ultramicroscopic
solid or liquid particles in air or gas. This denotes a two-phase system,
although the disperse phase alone (the particles) is often called the aerosol.
By common usage, aerosols have been given names (dust, fume, smoke, fog,
mist, etc.) that distinguish them as liquid or solid, as well as by their method
of formation and particle size. The size unit used for small particles is the
micrometer, abbreviated m. (The use of the term micron and its abbrevia-
tion is discouraged.) The unaided eye's limit of visibility is about 50 m,
the largest respirable particle is about 10 m, and the size of freshly formed
tobacco smoke is about 0.5 m.
Dust is formed by reducing earthy materials to small size. Processes such
as grinding, crushing, blasting, and drilling produce dust particles ranging
from the submicroscopic to the visible, their composition being the same as
that of the parent material. Common examples are mineral dusts, derived
from the disintegration of rock, and organic dusts, derived from wheat and
flour. Particle size is predominantly above 1 m for dust.
Fumes are formed by processes such as combustion, sublimation, and
condensation. Typical examples are the fumes of zinc oxide produced by zinc
vapor escaping the surface of molten metal. Its particle size is usually below
0.1 m at formation, but because these very small particles agglomerate
vigorously, their size increases rapidly.

#1999 Howden Buffalo, Inc.


7-2 FAN ENGINEERING

Smoke presupposes a certain degree of optical density. Generally, it is of


organic origin, such as the smoke from incomplete combustion of tobacco,
wood, oil, or coal. Particles of low vapor pressure that settle slowly under
gravity are also often called smokes. Smokes are characterized by a particle
size well below 0.5 m.
Fogs are formed by the condensation of water vapor upon suitable nuclei,
and mists by the atomization of liquids. Droplet size varies widely depending
upon formation conditions: persistent sea fogs are below 30 m, and sprays
range upward from 50 m.
When a solid or liquid is broken up into finely divided particles and
dispersed in air, two important changes take place: 1) the surface area is
greatly increased, and 2) the space occupied by the dispersed material is
expanded many times over the volume of the original mass. Thus, if 1 mL of
water is dispersed into droplets each 1 m3 in volume, 1012 particles will be
formed with a total surface area of 4.8 m2 compared with 0.00048 m2 for the
original 1 mL drop. Assuming a droplet concentration of a billion particles
per cubic meter of air, the 1 mL of original material will now be dispersed in
an air volume of 1000 m3. The enormous expansion in numbers and surface
area of the disperse phase of an aerosol has an important influence on chemi-
cal reactivity, including fire and explosion effects, cloud opacity, and particle
collection in air-cleaning equipment. The most important characteristics of an
aerosol are the size of the suspended particles and their concentration.

Particle Size
Clouds in which all the particles are the same size are termed monodis-
perse, and a single number is adequate to define their size. Aside from
airborne pollens and spores, monodisperse aerosols are rare in nature and
difficult to prepare in the laboratory. Almost every aerosol contains a range
of sizes around a central value (mode, mean, or median). These aerosols are
called polydisperse, and whereas a single number is adequate to describe a
central point of the distribution (for example, the median), this number fails to
indicate whether the sizes of the other particles are closely clustered around
the central size or are widely spread. This difference is illustrated in Figure 7.
1, which shows two particle-size distribution curves having the same median
value (that is, half the particles are larger and half smaller) but a markedly
different size range. Using familiar statistical terminology, the size range of
each of these bell-shaped, normal-probability distribution curves can be
quantified by its standard deviation.
If a plot of size versus frequency, similar to Figure 7.1, were prepared for
a foundry-shakeout dust cloud, it would have the skewed shape (Figure 7.2)
characteristic of most aerosol clouds. Figure 7.2 shows the mean (weighted
average) and the median (half larger, half smaller) of the size distribution,
neither of which is the same as the mode (greatest-frequency class interval)
plotted in the histogram.

#1999 Howden Buffalo, Inc.


CHAPTER 7 PARTICLES AND PARTICLE CLOUDS 7-3

Mathematical treatment of the skewed curve is difficult, but it can be


made manageable by plotting the logarithm of the size instead of the size.
When this is done, the curve is transformed into a bell-shaped one, as seen in
Figure 7.3. Customarily, particle size is plotted as a cumulative distribution
curve on logarithmic-probability (or log-normal) graph paper. It gives a
straight line when size follows the likely log-normal distribution. This
transformation is shown in Figure 7.4, where the 50% value is the median size
and the ratio of the median size to the 15.87% size is the geometric standard
deviation of the size distribution. On a cumulative (or summation) size plot,
the geometric standard deviation is also equal to the 84.13% size divided by
the 50% size. When sizes are closely clustered around the central value, the
geometric standard deviation will be a small number, reducing to 1 at the limit
for a monodisperse cloud. Industrial-dust clouds commonly have a geometric
standard deviation from 2.5 to 3.5 and contain particles that range in size over
3 to 4 orders of magnitude. For example, foundry-shakeout dust clouds
contain particles of from 0. 1 to 100 m.

Figure 7.1 Histogram of a Normal Probability Size Distribution

Adapted from the data of L. Silverman, C. E. Billings, and M. W. First, Particle Size Analysis
in Industrial Hygiene, Academic Press, New York, 1971, p. 236.

#1999 Howden Buffalo, Inc.


7-4 FAN ENGINEERING

The data from which the size curves of Figures 7.2, 7.3, and 7.4 were
derived were obtained by examining a representative sample of collected dust
under the microscope and measuring the diameter of particles, one by one.
This results in an analysis of particle size by number (or count), and the 50%
point is the count-median size. Similar data are derived from widely used,
automatic optical (and laser) single-particle counters and sizers. These
devices continuously conduct part of the aerosol through a small, illuminated
sensing volume in the interior of the device and direct a scattered light signal
onto a sensitive photomultiplier tube from each particle as it passes through.
The intensity of the scattered light signal is proportional to a well-defined
function of particle size. However, the characteristics of dust collectors and
many other devices, such as crushing and grinding machinery, are rated on a
mass rather than a count basis, so it is necessary to transform size-by-count
curves to size-by-mass curves to properly relate particle size to mass effi-
ciency.
Whenever the cumulative particle-size distribution by count plots as a
straight line on logarithmic probability graph paper, as in Figure 7.4, the
transformation to a mass basis can be easily made by using the following
equation:

log M m = log M c + 6.9078 log 2 g , (7.1)

Figure 7.2 Histogram of a Skewed Particle-Size Distribution

Adapted from the data of L. Silverman, C. E. Billings, and M. W. First, Particle Size Analysis
in Industrial Hygiene, Academic Press, New York, 1971, p. 237.

#1999 Howden Buffalo, Inc.


CHAPTER 7 PARTICLES AND PARTICLE CLOUDS 7-5

where M m and M c are the 50% size by mass and by count, respectively, and
g is the geometric standard deviation. The entire size-distribution curve by
mass can be plotted as a straight line passing through the mass median size
M m at 50% and parallel to the count curve. This is because, for the same
aerosol, the slopes of both the mass and count curves will be identical (that is,
the geometric standard deviations will be the same).
When the size-by-count distribution fails to give a straight line on
logarithmic probability graph paper, it is necessary to divide the curve into
small size intervals and construct the mass-distribution curve by simple
iterative calculations using the center point of the interval (the average size)
and the total number within the interval to calculate total particle mass for the
frac- .

Figure 7.3 Histogram of a Log-Normal Size Distribution

Adapted from the data of L. Silverman, C. E. Billings, and M. W. First, Particle Size Analysis
in Industrial Hygiene, Academic Press, New York, 1971, p. 238.

1999 Howden Buffalo, Inc.


7-6 FAN ENGINEERING

tion. The mass-median diameter of a polydisperse aerosol is always larger


and usually many times larger, than the corresponding count-median diameter
because of the dominating influence of the larger particles in a size-by-mass
distribution. For example, for a hypothetical cloud made up of two spherical
particles, one 1 m and one 10 m in diameter, 50% of the particles, by count
or number, will be 1 m and 50% 10 m. But, by mass, the 10 m particle
.

Count median Mc = 4.7m


Mass median Mm = 5.2m

Figure 7.4
Cumulative Particle-Size Distribution Plotted
on Logarithmic-Probability Graph Paper

Adapted from the data of L. Silverman, C. E. Billings, and M. W. First, Particle Size Analysis
in Industrial Hygiene, Academic Press, New York, 1971, p. 240.

#1999 Howden Buffalo, Inc.


CHAPTER 7 PARTICLES AND PARTICLE CLOUDS 7-7

counts for 1000 units, whereas the 1-m particle counts for only 1 mass unit,
so the 1-mm particle now accounts for only 0.1% of the total mass (that is,
100 1 / (1000 + 1) ). A similar analysis shows that the surface-median di-
ameter, important in light-scattering and cloud-opacity calculations, will be
between the count- and mass-median diameters.

Dynamic Behavior of Aerosol Particles in Still Air


Small particles follow the speed and direction of the suspending air unless
acted upon by a force that either does not affect the suspending medium or
affects it less. The simplest situation is one where the air is still and an
external force acts on the suspended particles but not on the air. Just such a
situation exists for a small, spherical particle suspended in still air and experi-
encing the gravitational sedimentation force Fg . In any consistent units,

mg
Fg = , (7.2)
gc

where m is the particle mass and g is the acceleration of gravity. Almost


immediately after it starts to settle, air resistance, or drag, will balance the
gravitational force, and the particle will then settle at a constant speed known
as the terminal or free-fall settling velocity. The nature of the resisting force
was investigated by Stokes in 1851, and the equation describing the terminal
settling velocity of small particles in still air bears his name.
The drag force FD on a settling particle is

CD A aV 2
FD = (7.3)
2 gc

where A is the projected area of the particle in the direction of motion and a
is the air density. Drag coefficients CD for small particles are related to the
particle Reynolds number, for which the characteristic dimension is the
particle diameter rather than the duct diameter, and the characteristic velocity
V is the velocity of the particle relative to the velocity of the air rather than
the velocity of the suspending air. Therefore, it is possible for the particle
Reynolds number to be a very small value, such as less than one, at the same
time that the suspending air has a Reynolds number higher than 100000. The
distinction between the particle Reynolds number and the suspending-air
Reynolds number is very important for understanding the dynamic behavior
of small particles. The coefficient of drag CD as a function of Reynolds
number Re is shown in Figure 7.5 for several particle shapes. This chart is
correct for smooth particles but only approximate for particles with rough
surfaces. Of the three zones marked on the figure, the one called the Newton
.

#1999 Howden Buffalo, Inc.


7-8 FAN ENGINEERING

zone, for which the Reynolds-number range is 500 to 200000, is unimportant


in aerosol technology. This is because particles with Reynolds numbers in
this range usually settle so rapidly that most attention must be given to
preventing premature settling or drifting in ducts. The fact that the drag
coefficients for each of the various particle shapes are very nearly constant in
this zone is noted and utilized in the chapter on conveying.
The Stokes zone, which covers a range of Reynolds numbers from 0.0001
to 2.0, is the only one that applies to the motion of aerosol particles. The drag
coefficients for particles in this zone vary inversely with the Reynolds num-
ber. For spherical particles,

24 24
CD = = (7.4)
Re d pV a

where d p is the particle diameter and is the air viscosity.


Because the drag force FD equals the gravitational force Fg when the
settling particle reaches its terminal velocity (that is, Fg = FD ),

CD A aV 2
mg = . (7.5)
2

Figure 7.5 Drag Coefficients for Particles

Adapted from the data of R. H. Perry, C. H. Chilton, and S. D. Kirkpatrick: Chemical


Engineers' Handbook, McGraw-Hill Book Company, Inc., New York, 1963, pp. 5-60.

#1999 Howden Buffalo, Inc.


CHAPTER 7 PARTICLES AND PARTICLE CLOUDS 7-9

Expanding these expressions for spherical particles where


3 8
m = p a d 6 and A = d 4 and solving for V redesignated Vg , the
3 2

terminal settling velocity, a familiar form of Stokes' law results:

Vg =
3
d 2 p a g 8 . (7.6)
18

For aerosols, the buoyant effect of the air is negligible, so the expression
3 8
p a is usually simplified to p .
The part of Equation 7.6 that contains only the particle and air character-
istics is called the Stokes number and is often termed the relaxation time
because it has the units of time:

d 2 p
= . (7.7)
18

For example, the relaxation time of a 0.1-m water droplet is 8.7 10 8 sec-
onds. For 1.0- and 10-m particles, it is 3.6 106 and 31. 104 seconds, re-
spectively.
In a gravitational field, the local acceleration g can be assumed equal to
32.2 ft/s2. The viscosity of air at ambient temperature is 1.225 10 5 lbm/ft-s.
The terminal settling velocity Vg of a 25-m spherical particle having a den-
sity of 150 lbm/ft3 according to Equation 7.6 is

2
25
25400 12 (150 )
Vg =
(32.2 ) = 0.147 ft/s.
(
(18 ) 1.225 105 )
Because of viscosity, a shear stress is produced in the suspending air by
the velocity gradient. This gradient exists between the layer of air molecules
adhering to the surface of a settling particle (and settling with it) and those air
molecules at rest in the still air surrounding the settling particle. Air viscosity
increases at higher temperature but is largely independent of pressure. When
particles are substantially below 1 m, their size approaches the length of the
mean free path of the air molecule in which they are suspended, and, as they
settle in what, for them, increasingly becomes the equivalent of a noncontinu-
ous medium, they tend to slip between the suspending air molecules instead of
dragging them downward. This results in a lower drag-coefficient value and
more rapid settling than predicted by Stokes' law and requires a correction
known variously as the Stokes-Cunningham and the Cunningham-Millikan
molecular-slip correction factor CC , which has the values shown in Figure 7.6
for ambient air at 1 atm pressure. For particles greater than 1 m, the

1999 Howden Buffalo, Inc.


7-10 FAN ENGINEERING

correction factor does not differ significantly from unity, but for smaller
particles, it assumes great importance. For 0.1-m spheres, the slip correction
equals 3, and spheres of this size settle at a rate three times the value predicted
by the uncorrected Stokes' equation. For 0.01-m spheres, the slip correction
equals 23.
Particles that are not spherical settle at slower rates than predicted by
Stokes' equation because of the effect that a larger projected area A in the
direction of movement has on particle drag in Equation 7.3. (That is, a sphere
has the least projected area of any geometric shape containing equal volume.)
Spheroidal quartz particles, for example, settle at a rate that is only 65% that
of glass spheres of equal mass. Mica platelets and asbestos fibers are even
less like spheres, and they deviate even more from the settling velocity
calculated from Stokes' equation. Figure 7.5 gives drag coefficients for disks
and cylinders, but for less regular shapes, empirically derived particle-shape
factors are usually applied to Stokes' equation to correct for the effect of
increased drag. Similar difficulties arise in applying Stokes' equation when
the particles are not homogeneous and the correct particle density is hard to
determine. Many coal fly-ash particles are hollow (cenospheres) and, so, of
uncertain density. Agglomerates of metal-fume particles contain much empty
space, and the handbook value for the density of the metal oxide would
greatly exaggerate the density of a loose cluster. Because the geometric
dimensions of aerosol particles are usually of little intrinsic interest compared
to their behavior in motion, it is customary to refer to aerosol particles in
terms of their aerodynamic equivalent diameter. This is defined as the diame-
.

Cc = 1 +
2 104
pDp
3
6.23 + 2.01exp 1095 pDp 8
p = pressure in cm Hg
Dp = particle diameter in cm

Figure 7.6 Slip-Correction Factor CC

Adapted from the data of L. Silverman, C. E. Billings, and M. W. First, Particle Size Analysis
in Industrial Hygiene, Academic Press, New York, 1971, p. 17.

#1999 Howden Buffalo, Inc.


CHAPTER 7 PARTICLES AND PARTICLE CLOUDS 7-11

ter of a homogeneous sphere of unit density (that is, 1 g/cm3 the density of
water) that has an identical terminal settling velocity in still air. By this
definition, a particle of lead shot having a density of 11.3 g/cm3 and a diame-
ter of 10 m has an aerodynamic-equivalent diameter of 10 113 . = 335
. m;
that is, the 10-m lead sphere will behave in all respects like a 33.5-m water
droplet when acted on by forces that do not act equally on the suspending air.
So far only a single, isolated particle in space, that is, unhindered settling,
has been considered. When particle concentration increases to the point
where a hydrodynamic interaction sets in between adjacent particles, substan-
tial entrainment of the surrounding air occurs, and the entire dust cloud settles
as a coherent body at a rate greater than would be expected for individual
particles of the same aerodynamic equivalent diameter in unhindered settling.
This phenomenon is important for clouds settling in still-air conditions but not
for clouds in turbulent motion.

Dynamic Behavior of Aerosol Particles in Moving Air Streams


Stokes' equation (7.6) has been derived for unhindered settling of homo-
geneous spheres in still air but can also be applied without serious error to
slowly moving aerosols. Although gravitational sedimentation will occur at
largely the rate predicted by Stokes' equation, the path followed by particles
that are simultaneously conveyed by a moving aerosol cloud will be a combi-
.

Figure 7.7 Diagrammatic Sketch of a Simple Settling Chamber

Adapted from the data of P. Drinker and T Hatch, Industrial Dust: Hygienic, Significance,
Measurement and Control, Second Edition, McGraw-Hill Book Company, Inc., New York,
1954, p. 284.

#1999 Howden Buffalo, Inc.


7-12 FAN ENGINEERING

nation of the sedimentation vector and the aerosol-flow vector. This interac-
tion can be illustrated by examining settling chambers, once widely used for
removing large particles from industrial-process air streams. Figure 7.7 is a
schematic representation of such a device. Assume a horizontal aerosol-
velocity vector of 60 fpm in the effective-settling section, a chamber length of
20 ft, and a chamber height of 7 ft. From these figures and Equation 7.6, it is
possible to calculate the aerodynamic-equivalent diameter of the minimum-
size particle that will just settle into the dust bin after entering the chamber at
the upper surface. (This is the most unfavorable entry position because it
requires the longest settling path to reach the dust bin.) The aerosol retention
time inside the settling chamber will be the length (20 ft) divided by the
aerosol horizontal velocity (60 fpm), or (20 / 60) 60 = 20 s. Therefore,
particles entering at the top have this maximum time to settle 7 ft into the dust
bin and be captured, a rate of 0.35 ft/s (chamber height divided by retention
time). Rearranging Equation 7.6 to solve for the diameter and substituting the
appropriate quantities (in U.S. customary units), the solution is

12
18Vg
12
18 0.35 1.225 105
d =
p g
= 25400 12 = 60 m.
62.3 32.2

For quartz quarry dust, the higher particle specific gravity (2.65) will hasten
settling and, thereby, reduce the minimum-size particle collected at 100%, but
the nonspherical shape will reduce settling velocity to 65% of the original and
so increase the minimum size collected at 100%. Making these corrections in
the calculation gives the correct value for quartz dust:

18 ( 0.35 0.65 ) 1.225 105


12 12
18Vg
d =
p g
=
25400 12 = 30 m.
( 62.3 2.65 ) 32.2

The net effect is to decrease the minimum, aerodynamic-equivalent quartz-


particle size that will be 100% collectable to about one-half the diameter of
the minimum collectable water droplet. The settling time can be increased
either by lengthening the gas path or by reducing the gas velocity. For
practical reasons, gravitational settling is not used in air cleaning for particles
below 50 m in size.
Aerosol particles also respond to forces other than gravity, some much
more effective in influencing the differential behavior of air-borne particles in
motion. The most important include centrifugal, inertial, and electrical forces.
Centrifugal force. When a small, suspended particle travels in a circular
path, it is acted upon by centrifugal force and moves radially outward relative
to the path. Centrifugal force Fc equals the product of the centrifugal accel-
eration and the particle mass m . Centrifugal acceleration equals the square of
the tangential velocity Vt divided by the radius of turn R :

#1999 Howden Buffalo, Inc.


CHAPTER 7 PARTICLES AND PARTICLE CLOUDS 7-13

2
mVt
Fc = . (7.8)
R

Although the particle may be suspended in a rotating fluid that is in turbulent


flow, it is assumed that radial outward particle movement is resisted by
viscous drag and that Stokes' equation reasonably approximates this drag.
Therefore, the equation of equilibrium motion for a small, spherical particle in
a centrifugal-force field assumes the following form, whereby centrifugal
2
acceleration Vt R replaces gravitational acceleration g in a new equation to
define terminal radial velocity Vr :

d 2 pVt
2

Vr = . (7.9)
18 R

In a centrifugal-force field, the radial acceleration may be as high as


80000 ft/s2, which is about 2500 g's. The radial acceleration for a tangential
gas velocity of 5000 fpm and a radius of curvature of 3 inches is

(5000 60 ) = 27800 ft/s 2 .


2 2
Vt
=
R (3 12 )
For a 5-m particle in such a force field, the terminal radial velocity
according to Equation 7.9 is

2
5
25400 12

Vr = (150 )( 27800 ) = 5 ft/s.
(
(18 ) 1.225 105 )
Assuming a 1.0-ft-long gas path, the distance X that a 5-m particle will
travel radially can be approximated by

X=
(1.0 ) 5.06 = 0.06 ft.
(5000 60 )
This value is only approximate because it is based on a constant value of R ,
whereas R actually increases as the particle migrates radially. A more exact
value can be obtained by using a differential equation and integrating over the
entire path length or by an arithmetical method of dividing the path into short
incremental lengths, solving for X , adding this to R , and repeating the
process over the entire path. The total X can then be obtained by summation.

1999 Howden Buffalo, Inc.


7-14 FAN ENGINEERING

When in the same flow field, the trajectories of particles having the same
relaxation time will be identical if they start from the same point. The radial
migration distance of the particle can be increased by lengthening the gas
path, by decreasing the radius of curvature, or by increasing the gas velocity.
For practical reasons, centrifugal separation is not used to capture particles
much below 5 m in size.
Inertial force. Inertia can be considered a special case of centrifugal force
because its application always involves the production of sharply curved
streamlines for the conveying air and the migration of particles across stream-
lines to impinge on a target around which air flows. This process is illustrated
in Figure 7.8, which shows an aerosol stream issuing from a round or rectan-
gular nozzle where the gas velocity is greatly increased. The jet from the
nozzle is discharged against an adjacent flat surface, causing the air to diverge
sharply. The numbers indicate the velocity pattern in the impinging jet flow.
Particles in the airstream have more inertia than the air itself and tend to
continue forward as the air turns off to the sides, causing particles with the
requisite inertia to impact on the surface. Because the air velocity over the
collecting plate is high, the plate must be coated with a viscous material to
prevent particle bounce and re-entrainment. The distance from the jet outlet
to the stationary plate governs the sharpness of curvature of the fluid stream
and, with jet velocity, also controls the collection efficiency. This distance is
maintained about equal to the characteristic dimension of the nozzle (for
example, jet diameter) for capturing small particles. Efficiency of impaction
is usually presented as a function of the square root of the dimensionless
impaction parameter I :

d C V
2 12 12

18 D
=
p c j
I1 2 , (7.10)
j

Figure 7.8 lmpingement of a Free Jet on a Flat Plate

#1999 Howden Buffalo, Inc.


CHAPTER 7 PARTICLES AND PARTICLE CLOUDS 7-15

where D j is the jet diameter for a round jet or the jet width for a rectangular
jet and V j is the average air velocity at the jet outlet. The characteristic
grouping of d 2 p 18 , so prominent in Stokes' equation, is also clearly
visible in this equation. The Cunningham-Millikan slip-correction factor Cc
is added because, as jet velocity increases, smaller and smaller particles cross
the streamlines and strike the collecting plate. When sonic velocity is
reached, particles of 0.25 m and smaller can be collected very efficiently
although the energy expenditure is, necessarily, great. However, since energy
expenditure is only a minor concern for air-sampling instruments, impaction
.

Figure 7.9 Seven-Stage Cascade Impactor


Adapted from the data of ARIES, Inc., Box 111, Davis, California 95616.

#1999 Howden Buffalo, Inc.


7-16 FAN ENGINEERING

devices have been modified for use as particle-size measuring devices.


Characteristically, this is done by arranging several stages with decreasing
nozzle sizes in series. Such an arrangement, known as a cascade impactor, is
shown in Figure 7.9. As the same air volume passes through each nozzle
stage in turn, changes in V j are solely a function of jet size D j . For a circular
jet, V j = 4Q D j , where Q equals the air rate in consistent units. For a
2

series of jets of decreasing diameter, the square root of the impaction pa-
rameter I will change as the square root of V j D j or 4Q D j changes, that
3

32
is, inversely as D j . Rearranging Equation 7.10 to solve for D j gives

d C 4Q 1
13 13

18 I
2 23

D =
j
p c
1/ 2
.

To design an impaction stage capable of removing all particles of greater than,


for example, 5-m aerodynamic equivalent diameter, it is necessary to know
the relationship between I and impaction efficiency. This is given in Table
7.1. For 100% collection with a round jet, I 1 2 must equal 0.56.
Assuming an air-sampling rate of 2.8 Lpm (46.7 cm3/s or 46.7 10 6
m3/s), 10m (10 106 m) particles of unit density (1 g/cm3 or 1000 kg/m3)
. 105 Pa-s.
and an air viscosity of 182

Table 7.1 lmpaction Efficiencies

Impaction efficiency Parameter I 1 2


% Round jet Rectangular jet
0 0.20 0.31
10 0.28 0.40
20 0.32 0.46
30 0.35 0.51
40 0.37 0.53
50 0.38 0.57
60 0.40 0.60
70 0.42 0.63
80 0.43 0.67
90 0.50 0.71
100 0.56 0.83
The experimental values above were determined for a ratio of plate spacing to
characteristic jet dimension of 3.

Adapted from the data of L. Silverman, C. E. Billings, and M. W. First, Particle Size Analysis
in Industrial Hygiene, Academic Press, New York, 1971, p. 21.

#1999 Howden Buffalo, Inc.


CHAPTER 7 PARTICLES AND PARTICLE CLOUDS 7-17

210 10 7 1000 1 24 46.7 10 7 1


6
13
6
13

D = 0.56
23

18 182 5
.
. 10
j

D j = 0.00386 m = 0.39 cm.

The diameter of a jet at the same flow rate that impacts all particles 5 m and
larger would, by an analogous calculation, be 0.244 cm, and if the smaller jet
were placed downstream of the larger, the aerosol could be fractionated into
three particle-size groups: 10 m and larger, between 5 and 10 m, and less
than 5 m. Cascade impactors often contain as many as 12 coordinated stages
that make it possible to divide aerosol particles into narrow, overlapping size
fractions for accurate analysis of size and size distribution. Classifying a
particle cloud into discrete parts by using a cascade impactor measures
aerodynamic diameter directly and, so, makes the cascade impactor a highly
valuable aerosol-characterization device.
Electrostatic force. A particle carrying a charge in the region of an
applied electrical field experiences acceleration in a direction dependent on
the polarity of the charge. The magnitude of the acceleration a is the product
of the particle charge ne and the strength of the charging field E :

a = ne, (7.11)

where n is the number of electron charges on the particle and e is the charge
of one electron (4.8 10-10 statcoulomb). Airborne dusts often become
naturally charged by such common processes as combustion, comminution,
and dispersion as well as by collision with air ions formed by cosmic rays,
lightning, and similar air-ion-producing processes. But, natural charging
seldom results in more than one or two elementary charges per particle, and
the proportion of positively and negatively charged particles is usually about
equal. The result is an electrically neutral cloud containing neutral particles
as well as particles containing weak charges of either polarity. For example,
freshly formed, pulverized-coal fly ash contains about 30% positively charged
particles, 25% negatively charged, and 45% neutral, whereas freshly formed,
copper-smelter dust contains about 40% positively charged particles, 50%
negatively charged, and only 10% neutral. Each of these aerosols contains
only slightly more charges of one polarity over the other.
When particles larger than about 1 m in diameter are passed through a
high-voltage corona discharge, they acquire charges from adsorbed electrons
and gas ions in proportion to the square of the particle diameter d p and the
strength of the charging field. For conducting particles, the saturation charge
(that is, the maximum possible number) is

3E 0 d p
ne = . (7.12)
4

#1999 Howden Buffalo, Inc.


7-18 FAN ENGINEERING

Particles of insulating materials acquire charges of 50 to 60% of this value.


For particles smaller than about 0.2 m, diffusion charging predominates,
and the number of charges acquired by a particle is given approximately by


d p cN 0e 2 t

d p kT
n= 2
ln 1 + , (7.13)
2e 2 kT

where n is the number of charges on an initially neutral particle after time t ;


k is the Boltzmann constant (1.371 10-16 erg/molecule-K); N 0 is the ion
density in ions/cm3; c is the ion velocity (root mean square) in cm/s; and T is
the temperature in K. Typical charges acquired by particles of various sizes
are shown in Table 7.2. Particles in a unipolar ion flux acquire 75% of the
saturation charge in less than 0.1 s. The migration velocity of a spherical
charged particle Ve in the direction of a collecting electrode of the opposite
sign can be obtained from the following expression, which utilizes the air-
resistance relationship given by Stokes' equation:

EneCc
Ve = , (7.14)
3 d p

where E is the field strength in the collecting space (esu/cm) and Cc is the
slip-correction factor.
Electrostatic force fields usually produce migration velocities of about 0.1
ft/s for 1.0-m particles. Corresponding velocities for both larger and smaller
particles are somewhat higher. For particles larger than 1 m, which can
receive a saturation charge ne proportional to their surface (that is
proportional to the square of the diameter), the migration velocity increases
directly .

Table 7.2 Number of Charges Acquired by Particies

Particle Field charging Diffusion charging


Diameter Exposure time (sec) Exposure time (sec)
m 0.01 0.1 1 0.01 0.1 1 10
0.2 0.7 2 2.4 2.5 3 7 11 15
2.0 72 200 244 250 70 110 150 190
20.0 7200 20000 24400 25000 1100 1500 1900 2300

Note: calculated under the following conditions typical of a wire-in-tube


assembly: T = 300K, N0 = 5 107 ions/cm3, E0.= 2 kV/cm, in air at
atmospheric conditions at 40 kV with a discharge current of 40 A/ft.

Adapted from the data of L. Silverman, C. E. Billings, and M. W. First, Particle Size Analysis
in Industrial Hygiene, Academic Press, New York, 1971, p. 19.

1999 Howden Buffalo, Inc.


CHAPTER 7 PARTICLES AND PARTICLE CLOUDS 7-19

with the particle size. However, for particles smaller than 1 m in diameter,
the saturation charge that can be imposed is proportional to the first power of
the diameter. So, it appears that the migration velocity for particles below 1
m should be independent of particle size but, because the slip-correction
factor increases for smaller particles, the velocity actually increases with a
decrease in size. Assuming a 1 m particle with a migration velocity Ve of
0.1 ft/s, the migration distance X through air traveling at 300 fpm along a 2-
foot boundary can be determined from

12.06101. 6 = 0.04 ft.


X=
1300 606
This migration distance can be increased by lengthening the gas path, by
increasing the field strength, or by decreasing the gas velocity. For practical
reasons, electrostatic precipitation is not used in air cleaning unless high
efficiencies are required for particles smaller than 1 m in size. Larger
particles that may be present will also be collected.

Brownian Motion and Diffusional Processes


Aerosol particles smaller than 0.1 m in diameter exhibit a significant
random movement called Brownian motion due to collisions with individual
gas molecules. Brown first described this random motion in 1827 in connec-
tion with small particles suspended in liquids, and later it was confirmed for
aerosol particles. The average linear displacement X of such a particle in a
time interval in seconds is a function of the particle size d p in ft and vari-
.

Table 7.3
Particle Displacements in Standard Air Due to Various Force Fields

Particle Diameter Displacements in 1 sec - ft


1 2
m Feet Cc Grav. Cent.3 Elect.4 Brown.5
10.0 3.28 10 -5 1.016 .024 20.5 0.98 .0000057
1.0 3.28 10 -6 1.165 .00027 .235 0.11 .0000194
0.1 3.28 10 -7 2.93 .0000069 .0059 0.27 .0000972
0.01 3.28 10 -8 22.6 .00000053 .00046 2.12 .0008540
1
Stokes-Cunningham slip-correction factor - dimensionless.
2
Gravitational force field - downward linear displacements based on 32.2 ft/s2 acceleration.
3
Centrifugal force field - outward radial displacements based on 862 g's acceleration.
4
Electrostatic force field - normal linear displacements based on 7500 volts/in. field strength
and a saturation charge on the particles.
5
Brownian movement - random linear displacements based on average values.
6
All data based on 150 lbm/ ft3 of particle density.

1999 Howden Buffalo, Inc.


7-20 FAN ENGINEERING

ous gas properties, as indicated by an equation Einstein developed in 1905:

4 R TC
X =
12

3Nd
u c
, (7.15)
p

in which the universal gas constant Ru is approximately 1545 ft-lb/lbm-mol-


R, the gas temperature T is in Rankins, the viscosity is in lbm/ft-sec, and
Avogadro's number N is 2.76 1026/lbm-mol. The factor Cc is the
Cunningham-Millikan slip-correction factor previously discussed.
Brownian motion will lead to diffusion, that is, a net streaming of particles
through the carrier gas in the direction of the lower concentration when a
concentration gradient exists. Concentration gradients are produced when the
streaming particles are removed from the gas at a boundary. Values for the
diffusion coefficient Dv (cm2/s) can be derived from the relationship

Dv = kBT , (7.16)

where T is the absolute temperature. The Boltzmann constant k is defined


as the universal gas constant divided by Avogadro's number. It has a value of
1.371 10-16 erg/molecule-K and is unaffected by pressure or gas
composition. The particle mobility B is defined as equal to Cc 3d p .
Values for the diffusion coefficients of various sizes of particles are given in
Table 7.3.
When the concentration of diffusing particles remains constant beyond a
certain distance from a deposition surface or boundary (as by vigorous eddy
.

Table 7.4 Velocity of Deposition

Particle Velocity of deposition (cm/sec) through boundary layer


Radius by gravity by molecular diffusion
m 1 m* l m* 10 m* 100 m* 1000 m* 10000 m*
10-4 negligible 7000 700 70 7 0.7
10-3 negligible 130 13 1.3 0.13 0.013
10-2 negligible 1.4 0.14 0.014 0.0014 0.00014
10-1 0.000225 0.022 0.0022 0.00022 0.000022 negligible
1 0.0128 0.0013 0.00013 0.000013 negligible negligible
10 1.2 0.00012 0.000012 negligible negligible negligible
100 70 0.000012 negligible negligible negligible negligible
*boundary layer thickness

Adapted from the data-of C. N. Davies, 'Deposition from Moving Aerosols, Chapter XII,
Aerosol Science, C. N. Davies (Editor), Academic Press, London and New York, 1966, p.
409.

1999 Howden Buffalo, Inc.


CHAPTER 7 PARTICLES AND PARTICLE CLOUDS 7-21

diffusive mixing) and when the particle concentration decreases from this
value to zero within the diffusion boundary layer (on the basis that all parti-
cles contacting the deposition surface are permanently removed from the
aerosol), the rate of deposition is

Dv C
R= , (7.17)
h

where R is the deposition rate per unit area, Dv is the coefficient of diffusion,
C is the concentration of particles outside the boundary layer, and h is the
thickness of the boundary layer through which particle diffusion occurs. The
velocity of deposition Vd (the rate of deposition per unit of area for a unit of
airborne concentration) is R C and equals Dv h . Table 7.4 shows values of
the velocity of deposition for particles of differing sizes through boundary
layers of graduated thickness.
Thermophoresis. This is another characteristic of particles substantially
below 1 m for which the driving force is the gas-molecule bombardment of
suspended particles. An aerosol particle subjected to a temperature gradient
between a hot and a cold surface will tend to move toward the colder surface.
This motion is caused by a thermal force arising from differential interaction
of the particle with the gas molecules. Those approaching from the hot side
have a higher average velocity (momentum) than those approaching from the
cold side, producing a net force in the direction of the flux of thermal energy.
Thermal precipitation of dust particles is used to obtain samples for analysis
by light/optical or electron microscopy but has not yet been employed for gas
cleaning.
Coagulation. Coagulation of aerosol particles occurs continuously and
spontaneously when airborne particles in Brownian motion collide with one
another and stick together, forming larger particles in the process. When the
particle size becomes large enough, rapid gravitational settling occurs, but
when small particles are being formed or added continuously, the aerosol
particle size tends to remain constant, that is, larger particles settle out and
new agglomerates form to take their place. If no new particles are added,
however, coagulation ceases when the particle growth has produced particles
too large to have appreciable Brownian motion and when too few particles
remain to make possible frequent collisions. Consequently, the rate at which
particles disappear by coagulation follows a simple law:

dn
= k n 2 , (7.18)
dt

where n is the number of particles, t is the time, and k is a coagulation


coefficient equal to 4 kTCc 3 .

#1999 Howden Buffalo, Inc.


7-22 FAN ENGINEERING

The full expression becomes

dn 4 kTn 2 Cc
= . (7.19)
dt 3

Integrating gives the number of particles nt remaining at time t :

n0
nt = , (7.20)
4 kTn0Cc t
1+
3

where n0 is the initial particle concentration at t = 0 . Turbulence enhances


coagulation by the eddy diffusion it produces within the aerosol. Eddy
diffusion is independent of particle Brownian motion and enhances it because
mechanical mixing increases contact between particles by improving the
chances of collision and by correcting local regions where particle diffusion (a
relatively slow process) has depleted the airborne dust concentration.
Coagulation is an important phenomenon in particle collection because an
increase in particle size usually results in either greater efficiency or a lower
energy requirement. A special application of coagulation to fume collection
occurs in the smelting industry where an important fraction of the product
coming from the furnaces is in the form of condensing vapor in a very hot
carrier offgas. Before collection by filters or electrostatic precipitators is
possible, the temperature must be greatly reduced. This is done by slowly
passing the hot, condensing aerosol through large, steel flues exposed to the
air. Equation 7.19 shows that the coagulation rate is directly related to the
temperature, the number concentration, and the Cunningham-Millikan correc-
tion factor, which increases as particles become substantially smaller than 1
m. All three factors are greatly enhanced in the very hot furnace offgases,
where freshly formed fume particles are in the range of 0.01 to 0.05 m and
concentrations may exceed 10 g/m3. This produces vigorous coagulation to
the point where large flocs form and rapid gravitational sedimentation occurs
.

Table 7.5 Rates of Coagulation

No. per cm3 W - mg per L t 0 - sec


1010 5236 3
109 523.6 30
108 52.36 300
107 5.236 3000
t 0 . equals the time required to reduce the number of particles to one tenth of the initial
number. W equals mg per L, assuming 1 m, and density equals 1 g/m3.

#1999 Howden Buffalo, Inc.


CHAPTER 7 PARTICLES AND PARTICLE CLOUDS 7-23

in the slowly moving aerosol. As much as 50% of the smelter product is


recovered from the floor of the large cooling flues as a byproduct of gas
cooling. The rate at which coagulation occurs at ambient temperature with
increasing particle numbers is illustrated in Table 7.5, which gives
approximate times necessary to reduce the number of particles to one-tenth of
their original number. It is clear from Table 7.5 that the size characteristics of
small-particle concentrations over 107/cm3 will not be stable, and as a result of
coagulation, even monodisperse aerosols rapidly become polydisperse unless
the particle-number concentration is quickly diluted with dust-free air to
below 107/cm3. No other method is known to be effective in preventing rapid
coagulation in submicrometer-particle clouds.

Optical Properties of Aerosols


Particles as small as molecules and as large as raindrops scatter light.
They comprise a size range of from 10 angstroms to a few millimeters, or six
orders of magnitude. Light scattering also depends on particle shape and
orientation with respect to the incident light and the direction of observation.
Most of the theory of light scattering is limited to spherical particles, as it is
impossible to derive exact solutions for other geometric shapes.
When light is incident on a small particle, it is scattered in all directions,
and if the particle is made of an absorbent material, some of this light is also
absorbed. When the particle is large compared to the wavelength of the
incident light, the scattering process can be described by geometrical optics
laws, and one can identify individual light rays that are either reflected from
the surface of the particle or penetrate into it and that emerge from the particle
in a direction different from the incident beam after refraction and internal
reflection. The distribution of light around the particle by refraction and
reflection can be obtained by summing the intensities of the individual light
rays. When the particle is partially opaque for the wavelength of the incident
light, some of the light penetrating into the particle will be absorbed and
converted into other forms of energy, mostly thermal. Theory predicts and
observations confirm that the scattered light is composed of two, incoherent,
plane-polarized components whose planes of polarization are mutually
perpendicular. Figure 7.10 shows the angular distribution of light intensity
scattered by a 0.2-m-diameter water droplet illuminated by unpolarized light
having a wavelength of 0.524 m. The "i1"-light-intensity curves in Figure
7.10 are for light vibrations perpendicular to the plane of observation; the "i2"
curves are for the intensity of light vibrating parallel to the plane of
observation. The forward-scattering component is clearly the strongest. For
particles having a diameter larger than the wavelength of light, the ratio of
forward to backward scattering may be 1000 or more. For this reason,
forward scattering is used in most optical particle counters and sizers, as well
as in total-scattering photometers. For particles larger than the wavelength of
the incident light, the scattering pattern will have prominent side lobes at the
90 and 270 positions.

1999 Howden Buffalo, Inc.


7-24 FAN ENGINEERING
.
Equations for single-particle light scattering are extremely complex and
vary with particle size, shape, and composition. Consequently, single-particle
counting and sizing instruments require calibration with monodisperse aero-
sols of known size and refractive index. A schematic for an aerosol-particle-
size analyser is shown in Figure 7.11. A small sensing volume of the order of
.

Figure 7.1 0
Angular Distribution of Intensity of Light Scattered
by a Spherical Particle i1 and i2 vs

Adapted from the data of D. Sinclair, "Optical Properties of Aerosols," Chapter 7, Handbook
on Aerosols: Chapters from the Summary Technical Report of Division 10, National Research
Committee, Selected and published by the United States Atomic Energy Commission,
Washington, D.C., 1950, P. 83.

#1999 Howden Buffalo, Inc.


CHAPTER 7 PARTICLES AND PARTICLE CLOUDS 7-25

1 mm3 is illuminated from a source that may, for the smallest particle-size
measurements, be a laser instead of the visible-light lamp shown. Aerosol
particles in the flowing aerosol pass, one by one, through the sensing volume
enclosed in a thin, outer sheath of air made dust-free to avoid contaminating
the sensing chamber. Scattered light from each particle is then collected by a
lens and directed to an electronic photomultiplier tube as a discrete light
pulse. The resultant photocurrent is amplified, and each pulse is counted and
passed through a pulse-height analyzer calibrated to display particle-size
units. Using laser and conventional-lamp illumination sources, it is possible
to measure aerosol particles in the range of 0.1 to 20 m with reasonable
accuracy, but instrumentation for this technique is difficult because the light
intensity scattered per particle is about 10-10 to 10-12 watt/steradian when the
.

Figure 7.11
Optical and Sampling Systems of Sinclair-Phoenix
Aerosol Particle Size Analyzer

Adapted from the data of D. Sinclair, "A New Photometer for Aerosol Particle Size Analy-
sis," Journal of the Air Pollution Control Association, Volume 17, Number 2,1967, p. 107.

#1999 Howden Buffalo, Inc.


7-26 FAN ENGINEERING

light intensity incident on a scattering particle is 1 watt/cm2. The instruments


are relatively insensitive to particle shape, at least in the range below 5 m.
Other types of optical instruments measure the light extinction of clouds
rather than single-particle scattering. These instruments, called total-
scattering photometers, are widely used for measuring atmospheric haze and
for filter testing. They use forward scattering, are of simple design, and are
intended only for relative measurements. Although they cannot make meas-
urements of absolute concentration, many of these instruments have a light of
fixed intensity that can be used as an approximate reference point for absolute
concentration measurements. Light scattering by aerosols is complicated by
the response variations produced by changes in particle size, surface texture,
index of refraction, particle color, and wavelength of illuminating light. But,
when relative measurements of concentration up-stream and downstream of
filters are made with light, these effects are nullified, since they are equal for
both, and only changes in concentration are detected.

Figure 12 Flow Chart for a Forward-Scattering Photometer

Adapted from the data of Air Techniques, Inc., 1717 Whitehead Rd., Baltimore, Md. 21207.

#1999 Howden Buffalo, Inc.


CHAPTER 7 PARTICLES AND PARTICLE CLOUDS 7-27

The basic equation for light extinction (total scattering and absorption) by
fine-particle clouds is often called the Lambert-Beer Law and has this general
form:

I T = I 0 e k snax , (7.21)

where I T and I 0 are the transmitted and incident light intensities, respec-
tively; x is the path length through the particle cloud; a is the projected area
of the average particle in the beam; n is the particle concentration per cubic
centimeter, and k s is a particle-extinction coefficient dependent on the index
of refraction, the surface roughness, and the color of the particle.
Figure 7.12 is a schematic showing the essential elements of a total
scattering photometer, called a penetrometer because it is used to measure
filter penetration. When air is drawn through the scattering chamber by the
vacuum pump, the sampled aerosol passes through the focal point of the cone
of light and causes that light to be scattered forward through the dark area.
The phototube, which has been exposed to darkness up to now, is activated by
the forward-scattered light and sends a signal to the amplifier. The amplifier
then augments the signal linearly and indicates it on the microammeter.
(Detectable particle sizes range from approximately 0.1 m to 100 m.) This
photometer has a threshold sensitivity of at least 10-3 g/L for aerosols
containing particles with a count-median diameter of 0.75 m, and it is
capable of measuring concentrations 105 times this value. Its sampling rate is
about 1 cfm for rapid response. In service, a sample of the aerosol mixture is
taken from the upstream side, close to the filter, to measure the concentration
of the challenge aerosol. With the challenge aerosol flowing through the
scattering chamber, the apparatus is then adjusted by setting the needle on the
100% reading of the 100% scale. Next, a stray-light adjustment is made to
compensate for any signal caused by dark current or reflections off internal
surfaces of the scattering chamber. The equipment is now ready to read the
downstream concentration, and the result will be the penetration (1 - effi-
ciency) in percent.
Figures 7.13 and 7.14 summarize many of the properties discussed in this
chapter. They have been included here for easy reference to often-sought
information.

#1999 Howden Buffalo, Inc.


7-28 FAN ENGINEERING

Figure 7.13 Particle Characteristics

#1999 Howden Buffalo, Inc.


CHAPTER 7 PARTICLES AND PARTICLE CLOUDS 7-29

Figure 7.13 (Cont.) Particle Characteristics


Adapted from the data of C. E. Lapple: Reprinted from Stanford Res. Inst. J. in "Nonviable
Particles in the Air," Air Pollution, Volume 1, Second Edition, A. Stern (Editor), Academic
Press, New York, 1968, pp. 50-51.

#1999 Howden Buffalo, Inc.


7-30 FAN ENGINEERING

Figure 7.14
Common Particle Dispersions and Methods of Size Measurement

Adapted from the data of M. W. First and P. Drinker, "Concentrations of Particulates Found
in Air," Archives of Industrial Hygiene and Occupational Medicine, Volume 5, Number 4,
April 1952, p. 388.

#1999 Howden Buffalo, Inc.

You might also like