Effectiveness of Vertical Mist Eliminators In a Cross Flow Scrubber


Bell, C G.; Strauss, W

Journal of the Air Pollution Control Association 23(11): 967-969

1973


range of applications of conventional wire mesh pads to be extended. In recent years, increasing research effort has been

Effectiveness
of
Vertical
Mist
Eliminators
In
a
Cross
Flow
Scrubber
Cohn
G.
Bell
and
Werner
Strauss
University
of
Melbourne,
Australia
Three
types
of
vertical
mist
eliminator—knitted
wire
mesh,
crushed
aluminum
turnings,
and
multiple
pass
louvers—have
been
studied
in
a
packed,
cross
flow
scrubber.
The
knitted
wire
mesh
had
a
maximum
efficiency
of
92%
with
a
pressure
drop
of
0.43
in.
W.G.
at
a
velocity
of
13
ft/sec,
while
at
the
maximum
velocity
possible
in
the
equipment,
16
ft/sec,
the
louver
elim-
inator
was
the
most
effective
(76%).
The
Souders-
Brown
equation:
v
=
KA/(p
L
p,)/
p
G
,
recommended
for
conventional
horizontal
mist
eliminators
is
equally
applicable
to
vertical
units,
with
10%
higher
velocities.
There
is
surprisingly
little
published
information
on
the
be-
havior
and
operating
characteristics
of
mist
eliminators,
and,
with
the
exception
of
the
review
by
Stearman
and
William-
son,'
these
deal
exclusively
with
wire
mesh
mist
elimhiators.
2-5
However,
with
the
high
throughputs
used
for
air
pollution
control
systems,
cross
flow
scrubbers
are
in
common
use,
and
these
have
to
be
followed
by
effective,
low-pressure
drop,
mist
eliminators.
Mr.
Bell
and
Dr.
Strauss
are
in
the
Department
of
In-
dustrial
Science,
University
of
Melbourne,
35
Royal
Pa-
rade,
Parkville,
Victoria
3052,
Australia.
In
general,
mist
eliminators
should
have
the
following
char-
acteristics:
low
cost,
ease
of
manufacture
and
installation,
low-pressure
drop,
and
high
efficiencies
over
a
wide
range
of
superficial
gas
velocities
and
mist
loadings.
The
units
should
be
self-draining
and
self-cleaning,
with
low
operating
and
maintenance
charges,
able
to
operate
for
long
periods
without
attention.
The
operating
principle
of
mist
eliminators
is
either
gravity
settling
or
centrifugal
separation,
or
some
form
of
inertial
impingement
separation.
This
last
is
most
common
because
of
the
small
space
requirement
and
ease
of
installation.
This
series
of
experiments
explores
the
effectiveness
of
three
types
of
impingement
separators
usually
recommended
for
cross
flow
scrubbers.
Apparatus
The
mist
eliminators
were
installed
in
a
section
following
a
pilot
plant
cross
flow
water
scrubber
packed
with
1
in.
poly-
ethylene
Tellerettes
to
a
depth
of
2
ft.
The
mist
elimination
section
was
2
ft
square
and
9
in.
long.
The
fluids
used
in
the
experiments
were
air
and
water.
Three
types
of
mist
eliminator
were
tested.
The
first
con-
sisted
of
simple
zig-zag
louvers
with
5
changes
in
direction
of
gas
flow
(2
V's
in
series)
with
a
small
lip
at
the
end
to
prevent
reentrainment.
The
sections
were
molded
from
0.1
in.
P.V.C.,
and
were
in.
deep,
with
90°
angles
and
3%
in.
between
points.
The
lip
was
V
I
in.
long,
and
the
total
length
was
8
in.
The
spacing
between
louvers
was
1
1
4
in.
and
21
blades
were
used.
The
second
mist
eliminator
was
made
of
knitted
0.01
diam
stainless
steel
wire,
with
apertures
and
crimped
in
to
h-in.
corrugations.
The
packing
density
was
10
lb/ft
3
and
the
depth
of
the
bed
was
4
in.
The
third
mist
eliminator
consisted
of
a
6-in.
bed
of
aluminum
turnings
(0.001
in.
thick,
averaging
0.1
in.
diam)
crushed
into
a
wire
mesh
box,
with
a
packing
density
of
4
lb/ft
3
.
November
1973
Volume
23,
No.
11
967
2
4
6
8
10
12
14
16
18
Superficial
gas
velocity,
ft/sec
Figure
1.
Efficiency
of
demisters
as
a
function
of
superficial
gas
velocity.
1.0
Entrainment
points
0.01
1
10
100
Superficial
gas
velocity,
ft/sec
Figure
2.
Pressure
drop
through
demisters
as
a
func-
tion
of
superficial
gas
velocity.
The
entrainment
velocity
is
clearly
discernible
for
the
knitted
wire
and
packed
turnings
eliminators
by
the
change
of
slope
in
the
plots.
The
pressure
loss
in
the
knitted
wire
demister
tends
to
be
greater
than
that
for
the
packed
turnings
below
the
entrainment
velocity,
but
this
changes
above
en-
trainment
velocities.
The
louver
demister
has
a
pressure
drop
considerably
lower
than
both
these.
In
general,
the
lower
pressure
drop
has
a
lower
efficiency
associated
with
it.
Mist
Sampling
and
Analysis
100
The
mists
were
sampled
isokinetically,
with
probes
facing
upstream
before
and
after
the
mist
eliminators.
The
gas
flow
was
measured
with
an
orifice
plate
(D
and
D/2
tappings)
upstream
of
the
pilot
plant
in
a
long
straight
section
of
duct
(14
in.
diam).
After
sampling,
the
mist
droplets
were
collected
either
on
the
stages
of
a
cascade
impactor,
or
in
U-tubes
packed
with
glass
wool.
In
addition,
particle
size
distributions
were
determined
by
microscopic
siting
of
droplets
collected
on
a
slide
coated
with
a
hydrophobic
oil
and
exposed
for
a
brief
period
(1-2
sec)
in
the
gas
stream.
The
cascade
impactor
was
of
the
multistage
type
developed
by
Brink.'
The
impactor
was
surrounded
by
a
water
jacket
maintained
at
90-95°F
(32-35°C),
5°F
(2°C)
warmer
than
the
temperature
of
the
duct
to
allow
for
heat
losses
and
mini-
mize
condensation
inside
the
impactor.
Sample
lines
to
the
impactor
were
kept
as
short
as
possible
and
warmed
with
electrical
resistance
tapes
to
avoid
condensation
of
the
vapor
and
mist
deposition
in
the
line.
The
pressure
drop
across
the
impactor
was
measured
by
a
mercury
and
a
water
ma-
nometer
respectively
before
and
after
the
impactor.
After
the
impactor
a
Millipore
filter
paper,
with
an
0.8-gm
pore
size,
in
a
Plexiglass
holder
was
used
to
retain
residual
droplets
and
particles.
The
collection
cups
in
the
cascade
impactor
were
carefully
weighed
immediately
before
and
after
each
run,
and
sampling
times
of
at
least
30
min
were
required
for
adequate
samples.
The
cascade
impactor,
while
it
gave
both
concentrations
and
size
distribution,
was
not
only
time-consuming
to
use,
but
required
steady
operation
of
the
pilot
plant
scrubber
for
long
periods.
_So,
overall
performance
of
the
mist
eliminators
was
evaluated
using
U-tubes
packed
with
fiber
glass
(10-15
gm
diam)
before
and
after
the
test
action.
Desiccants,
such
as
calcium
chloride
and
magnesium
perchlorate,
were
also
tried
as
a
U-tube
packing,
but
the
results
tended
to
be
irregular
because
of
channelling
of
the
moist
gases
through
the
packed
desiccants.
However,
the
glass
fiber
packed
U-tubes
were
followed
by
tubes
containing
a
dessiccant
(CaC1
2
),
in
which
increases
in
weight
were
detected
by
passing
or
reentrainment
from
the
glass
fiber
packing.
The
U-tubes
were
placed
in
a
water
bath
held
at
43-45°F
(6-7°C),
while
the
sampling
lines
were
warmed
and
kept
to
a
minimum
length.
Sampling
times
were
short
(5-1.5
min)
and
depended
on
the
inverse
of
the
flow
rate
through
the
sampling
system
(5-20
1/min).
This
avoided
reentrainment.
The
size
of
the
sample,
about
1
to
2
g,
could
be
accurately
determined
to
better
than
0.1%,
in
a
total
U-tube
weight
of
90
to
120
g,
when
special
precau-
tions
were
taken
to
avoid
both
entrainment
and
evaporation
of
the
sample.
Results
The
overall
(concentration)
efficiencies
of
the
three
types
The
droplet
size
distribution
of
the
mist
entering
and
leav-
of
demister,
as
a
function
of
superficial
velocity
derived
from
ing
the
louver
eliminator
with
different
superficial
gas
veloci-
measurements
with
packed
LT-tubes,
are
shown
in
Figure
1.
ties,
and
with
different
liquid
loadings
in
the
packed
tower
is
Although
the
experimental
points
exhibit
some
scatter,
the
shown
in
Figure
3.
The
droplets
ranged
from
10
to
1000
A
m
general
trends
are
clearly
shown.
The
knitted
wire
mesh.
(1
mm)
and
were
mainly
in
the
region
of
40
to
500
A
m.
demister
reached
a
maximum
efficiency
of
92%
at
13
ft/sec
Doubling
the
water
loading
in
the
scrubber,
at
constant
gas
(0.43
in.
W.G.),
which
was
significantly
higher
than
for
the
velocity,
gave
somewhat
larger
numbers
of
droplets,
but
re-
packed
aluminum
turnings,
84%,
at
a
similar
velocity
(12.5
duced
the
mean
size
of
the
drops
entering
the
eliminator.
ft/sec,
0.33
in.
W.G.).
It
was
observed
that
at
higher
veloci-
The
eliminator
proved
more
efficient
in
removing
the
larger
ties
the
drainage
of
water
was
impeded
by
the
gas
stream,
and
droplets,
and
this
blurred
the
particle
size
distribution
of
the
liquid
Wilt
up.
At
slightly
higher
velocities
this
was
torn
drops
leaving
the
eliminator.
Thus,
at
the
lower
loading
(10
from
the
demister
and
reentrained.
gal/min)
the
mean
sizes
were
about
400
A
m
before,
and
300
Reentrainment
velocities
were
not
reached
for
the
louver
A
m
after
the
eliminator,
while
at
the
higher
loading
(20
gal/-
demister,
even
at
16
ft/sec,
which
was
the
maximum
which
min)
the
means
were
200
and
300
gm,
respectively.
With
could
be
achieved
in
this
plant.
At
this
velocity
the
louver
constant
liquid
loading
in
the
scrubber
(10
gal/min)
and
demister
was
76%
efficient,
with
a
pressure
drop
of
0.22
in.
increasing
the
superficial
velocity
by
50%
(from
10
to
15
W.G.
The
pressure
drop
through
the
demisters
as
a
function
ft/sec),
the
total
number
of
droplets
increased
by
about
20%
of
superficial
gas
velocity
is
shown
in
Figure
2
on
a
log-log
on
entering
the
mist
eliminator,
and
the
mean
droplet
size
plot.
was
decreased
from
400µm
to
about
100µm.
After
the
louver
968
Journal
of
the
Air
Pollution
Control
Association
t°80
0
60
-7
>
0
40
20
0
1
Knitted
wire
demister
o
O
O
X
Packed
aluminum
turnings
Louver
demister
Louver
demister
Packed
aluminum
turning
demister
Knitted
wire
demister
I
1
1000
500
M
is
t
p
ar
t
ic
le
d
iame
te
r
(m
icrons
)
200
100
50
1000
C
O
E
.(2
100
k
0
3
00c0
0
10
1
10
100
Number
of
mist
particles
Figure
3.
Drop
size
distribution
for
mists
entering
and
leaving
the
louver
eliminator.
A.
Entering
mist
10
ft/sec,
10
gal/min
loading
a.
Leaving
mist
10
ft/sec,
10
gal/min
loading
B.
Entering
mist
10
ft/sec,
20
gal/min
loading
b.
Leaving
mist
10
ft/sec,
20
gal/min
loading
C.
Entering
mist
15
ft/sec,
10
gal/min
loading
c.
Leaving
mist
15
ft/sec,
10
gal/min
loading
2
3
5
10
20
%
of
mist
particles
Figure
4.
Size
distribution
of
droplets
leaving
all
types
of
eliminators.
eliminator
there
was
approximately
a
30%
rise
in
droplet
number,
and
a
mean
droplet
size
of
about
150
pm.
The
size
distribution
of
droplets
leaving
all
types
of
elimina-
tors
is
.shown
in
Figure
4,
with
a
tower
loading
of
10
gal/min,
at
a
superficial
velocity
of
10
ft/sec.
It
may
be
noted
that
the
size
distribution
(maximum
penetration)
of
droplets
shows
maxima
in
the
range
of
150-400
pm.
The
knitted
wire
eliminators
and
the
packed
turnings,
which
rely
on
inertial
impaction
of
the
drops
on
their
small
cross
section
are
more
effective
in
the
removal
of
small
droplets,
while
the
louver
eliminator
is
more
effective
with
the
large
(greater
than
500
pm)
droplets.
The
temperatures
throughout
the
experiments
were
68-72°F
(20-22°
C)
.
Discussion
While
this
is
the
first
report
of
the
performance
of
vertical
mist
eliminators
following
a
cross
flow
scrubber,
meaningful
comparisons
can
be
made
with
measurements
on
conventional
horizontal
two-stage
eliminators.
Both
York
and
Poppele
4,5
and
Massey
2,3
achieved
efficiencies
of
90%
at
superficial
veloci-
ties
of
12-15
ft/sec
and
pressure
drops
of
1
to
1
in.
W.G.
In
this
work
similar
efficiencies
have
been
achieved
for
a
vertical
knitted
wire
single
stage
eliminator,
with
similar
superficial
velocities,
but
less
than
half
the
pressure
drop
(0.43
in.
W.G.).
Superficial
velocities
of
7-10
ft/sec
are
recommended
by
Stearman
and
Williamson'
for
louver
eliminators,
but
this
work
indicates
that
optimum
values
may
be
higher.
Manufacturers
of
wire
mesh
demisters
suggest
that
a
form
of
the
Souders-Brown
equation
be
used
for
predicting
the
optimum
velocity,
and
suggest
further
that
the
units
will
work
satisfactorily
in
the
range
of
30
to
110%
of
the
optimum.
This
equation
is
generally
expressed
as
v
=
K
-
\./(p.b
PG)
/PG
where
v
is
the
superficial
gas
velocity,
PL
the
density
of
the
liquid,
and
pc
the
density
of
the
gas
or
vapor.
K
is
an
empirical
constant,
which
is
generally
taken
as
0.35
for
standard
types
of
wire
mist
eliminators,
and
0.40
for
high
throughput
demisters.
In
the
present
case,
the
value
of
K
at
entrainment
is
found
to
be
0.47
for
the
knitted
wire
demister,
and
0.44
for
the
crushed
aluminum
turnings;
i.e.,
the
lower
of
these
is
110%
of
the
maximum
recommended
value.
Thus,
the
Souders-Brown
equation,
using
a
K
value
of
about
0.45,
would
be
the
optimum
value
for
a
cross
flow
scrubber
fitted
with
a
wire
mesh
or
similar
eliminator.
Ten
per
cent
overload
would
only
give
a
slight
amount
of
entrainment,
but
efficiencies
at
velocities
of
less
than
70%
of
the
optimum
are
much
lower
than
those
at
the
optimum
velocity.
Conclusions
The
knitted
wire
and
packed
aluminum
turnings
types
of
mist
eliminators
are
most
effective
at
their
optimum
super-
ficial
velocity
of
about
13
ft/sec,
while
'the
effectiveness
of
the
louver
eliminator
is
improved
with
higher-velocities,
even
above
16
ft/sec.
The
reason
for
the
greater
efficiency
(92%)
of
the
knitted
crimped
wire
eliminator
compared
with
that
of
the
packed
aluminum
turnings
(84%)
is
probably
due
to
channeling
through
the
uneven
packing
of
the
latter,
by
pass-
ing
collecting
surfaces.
Higher
efficiency
is
accompanied
by
higher
pressure
drop,
which
adds
to
operating
costs.
The
present
study
shows
that,
provided
that
superficial
gas
velocities
do
not
exceed
13
ft/sec,
the
knitted
and
crimped
eliminators,
made
of
fine
wire,
are
the
most
effective
units
for
cross
flow
scrubbers,
but
when
significantly
higher
velocities
are
used,
multi-V
louver
eliminators
are
preferable.
References
1.
F.
Stearman
and
G.
J.
Williamson,
"Spray
Elimination,"
Chap.
16
in
Gas
Purification
Processes,
ed.
G.
Nonhebel,
Geo.
Newnes
Ltd.,
London,
1964.
2.
0.
D.
Massey,
"How
well
do
filters
trap
stray
stack
mist,"
Chem.
Eng.,
66
:
143
(July
13,
1959).
3.
0.
D.
Massey,
"Demisters
for
sulfuric
acid
plant
stacks,"
Chem.
Eng.
Progr.
55
(5):
114
(1959).
4.
0.
H.
York
and
E.
W.
Poppele,
"Wire
mesh
mist
eliminator,"
Chem.
Eng.
Progr.
59
(6):
45
(1963).
5.
0.
H.
York
and
E.
W.
Poppele,
"Two
stage
mist
eliminators
for
sulfuric
acid
plants,"
Chem.
Eng.
Progr.
66
(11):
67
(1970).
6.
J.
A.
Brink,
"Cascade
impactor
for
adiabatic
measurements,"
Ind.
Eng.
Chem.
50
:
645
(1958).
20
10
1
November
1973
Volume
23,
No.
11
969