Design of wire mesh mist eliminators


Brunazzi, E.; Paglianti, A.

AICHE Journal 44(3): 505-512

1998


Knitted wire mesh mist eliminators are used extensively in many industrial plants. Their widespread application is essentially due to the low cost and efficient removal of entrained liquid droplets from vapor and gas streams. Despite the broad range of entrainment removal applications, open literature on this topic is limited. All the available design relations are based on semiempirical equations with an uncertain range of application. In this work, a set of new experimental data was obtained by investigating the performances of commercial eliminators, and a mechanistic model is presented. Comparison between experimental data and the proposed model shows that it can be used to predict separation efficiency both for horizontal and vertical arrangements.

Fluid
Mechanics
and
Transport
Phenomena
Design
of
Wire
Mesh
Mist
Eliminators
Elisabetta
Brunazzi
and
Alessandro
Paglianti
Dept.
of
Chemical
Engineering,
Industrial
Chemistry
and
Materials
Science,
University
of
Pisa,
1-56126
Pisa,
Italy
Knitted
wire
mesh
mist
eliminators
are
used
extensively
in
many
industrial
plants.
Their
widespread
application
is
essentially
due
to
the
low
cost
and
efficient
removal
of
entrained
liquid
droplets
from
vapor
and
gas
streams.
Despite
the
broad
range
of
en-
trainment
removal
applications,
open
literature
on
this
topic
is
limited.
All
the
available
design
relations
are
based
on
semiempirical
equations
with
an
uncertain
range
of
appli-
cation.
In
this
work,
a
set
of
new
experimental
data
was
obtained
by
investigating
the
performances
of
commercial
eliminators,
and
a
mechanistic
model
is
presented.
Com-
parison
between
experimental
data
and
the
proposed
model
shows
that
it
can
be
used
to
predict
separation
efficiency
both
for
horizontal
and
vertical
arrangements.
Introduction
Separation
of
liquid
droplets
from
gas
or
vapor
streams
is
one
of
the
most
common
operations
in
chemical
plants.
Liq-
uid
separation
may
be
required
not
only
to
recover
valuable
products
or
to
protect
downstream
equipment
from
corrosive
liquids,
but
can
also
be
necessary
to
improve
emissions
con-
trols.
Droplet
size
is
a
critical
parameter
in
selecting
the
most
appropriate
liquid
separator.
If
droplet
size
is
between
500
and
1000
gm,
a
simple
gravity
settling
drum
may
be
suffi-
cient
to
obtain
high
liquid
separation
efficiency.
On
the
other
hand,
if
the
mean
droplet
size
is
less
than
1µm,
fiber
coa-
lescers
are
necessary.
The
goal
of
this
article
is
to
study
liquid
separation
if
the
droplet
size
is
in
the
range
of
5-100
gm.
In
this
case
knitted
wire-mesh
mist
eliminators
can
be
used
to
achieve
highly
effi-
cient
removal.
Knitted-mesh
mist
eliminators
can
be
classi-
fied
as
"impingement
type
separators";
their
working
princi-
ple
is
based
on
inertial
force,
and
they
can
guarantee
high
separation
efficiency
in
the
droplet-size
range
studied
in
this
work.
Wire-mesh
contactors
are
made
by
knitting
wires
to
form
a
layer,
shown
in
Figure
la,
that
may
be
rolled
spirally
to
form
cylindrical
elements,
commonly
used
for
small-diameter
ap-
plications,
or
folded
up
in
several
layers
to
form
a
pad
of
the
desired
thickness
(see
Figures
lb—ld).
The
wire
used
in
the
knitted
layer
typically
has
a
diameter
in
the
80-280-gm
range
and
the
typical
thickness
used
for
the
pads
is
in
the
65-150-
mm
range.
Correspondence
concerning
this
article
should
be
addressed
to
A.
Paglianti.
In
this
work
some
knitted
wire-mesh
mist
eliminators,
set
in
layers
with
different
geometrical
characteristics,
were
tested
in
horizontal
and
vertical
arrangements,
using
air
and
water
as
working
fluids
and
operating
at
ambient
conditions.
Available
Methods
for
Designing
Wire-Mesh
Mist
Eliminators
Few
articles
on
this
topic
have
been
published
in
the
litera-
ture
despite
the
extensive
use
of
these
separators.
Some
arti-
cles
suggest
how
to
install
mist
eliminators
in
some
common
equipment
in
order
to
avoid
trouble;
the
suggestions
are
mainly
aimed
at
ensuring
uniform
gas
and
liquid
velocities
to
avoid
overloading
a
part
of
the
unit, but
few
articles
show
how
it
is
possible
to
evaluate
the
efficiency
of
the
separator.
Proper
installation
is
certainly
necessary,
but
without
a
doubt,
it
is
more
important
to
have
reliable
design
tools
that
make
it
possible
to
select
the
proper
separator
for
each
case.
The
usual
design
is
based
on
the
computation
of
maximum
gas
velocity,
which
according
to
the
Souders—Brown
relation,
can
be
expressed
as
P
P
g
(1)
p
g
where
the
constant
K
depends
on
the
deentrainment
height
(York,
1954)
and
on
the
physical
properties
of
the
working
fluids,
and
p
i
and
p
g
are,
respectively,
the
density
of
the
liq-
uid
and
the
gas
phases.
The
K
values
are
experimentally
de-
AIChE
Journal
March
1998
Vol.
44,
No.
3
505
(a)
(c)
(b)
(d)
Figure
1.
Wire-mesh
mist
eliminator:
(a)
single
layer;
(b)
two-layer
pad;
(c)
four-layer
pad;
(d)
eight-
layer
pad.
termined
by
vendors;
a
typical
value
of
0.107
m/s
is
com-
monly
used,
even
if
lower
values
are
suggested
for
high
en-
trainment
loadings
or
when
the
liquid
phase
is
dirty.
The
method
of
designing
wire-mesh
separators,
based
on
the
constant
K,
is
very
rough
because
it
does
not
take
into
account
the
drop
size,
on
which
the
collection
efficiency
is
strongly
dependent.
For
this
reason,
some
authors
have
ana-
lyzed
the
separation
phenomena
in
detail,
suggesting
differ-
ent
semiempirical
relations.
From
the
theoretical
point
of
view
the
collection
efficiency,
in
fact,
involves
three
different
separation
mechanisms
(Gerrard
et
al.,
1986;
Holmes
and
Chen,
1984;
Feord
et
al.,
1993):
Inertial
capture
involves
the
drops
that,
leaving
the
gas
stream
line
because
of
their
inertia,
impact
the
target
wire
of
the
mesh
and
are
collected.
Interception
capture
involves
the
drops
that
remain
on
the
gas
stream
line,
but
because
of
their
size,
brush
against
the
wire
of
the
mesh
and
are
collected.
Diffusion
capture
involves
only
submicron-size
particles
and
for
this
reason
is
negligible
in
the
study
of
this
type
of
unit.
Inertial
and
interception
capture
mechanisms
are
involved
in
the
particle-size
range
in
which
wire-mesh
mist
eliminators
work.
Holmes
and
Chen
(1984)
showed
that
for
this
type
of
equipment
the
primary
mechanism
responsible
for
efficiency
is
inertial
impaction.
This
implies
that
the
total
separation
efficiency
can
be
evaluated
by
taking
into
account
only
the
contribution
due
to
inertial
capture.
For
this
mechanism
sim-
ple
relations
have
been
published
(Langmuir
and
Blodgett,
1946;
Pich,
1966)
to
evaluate
the
inertial
capture
efficiency
for
a
single
wire
target,
-ri
sr
.
All
these
relations
agree
that
the
inertial
capture
efficiency
is
a
function
of
the
Stokes
number,
St,
defined
as:
p
i
.
u
d
i
;
St
18
18
g
g
D,,'
(2)
where
u
is
the
superficial
gas
velocity,
g
g
the
gas
viscosity,
and
d
d
and
D
indicate
the
droplet
and
wire
diameters,
re-
spectively.
The
relation
suggested
by
Langmuir
and
Blodgett
(1946)
is
important
from
a
theoretical
point
of
view
because
it
makes
it
possible
to
evaluate
the
capture
efficiency
of
a
single
tar-
get,
but
since
the
aim
here
is
to
predict
the
efficiency
of
in-
dustrial
mesh
collectors,
n
n
,
it
is
necessary
to
introduce
de-
pendence
on
the
geometry
of
the
wire
mesh
packing.
All
the
published
equations
refer
to
the
analysis
proposed
by
Car-
penter
and
Othmer
(1955),
who
suggested
the
following
semiempirical
equation:
2
Ti„
=
1
--
1
-
-
3
-
a
e
'
"ilsr
'
71"
(3)
where
a,
is
the
specific
surface
area
of
the
separator,
z
the
distance
between
two
successive
layers,
n
the
number
of
lay-
ers,
and
-ri
s
,
the
efficiency
of
a
single
target.
The
Experimental
Loop
Experimental
collecting
efficiencies
as
a
function
of
droplet
size
and
gas
velocity,
were
determined
in
atmospheric
work-
ing
conditions
in
two
experimental
loops
designed
and
built
at
the
Department
of
Chemical
Engineering
of
the
University
of
Pisa.
For
this
purpose,
industrial
and
nonindustrial
mist
eliminators,
furnished
by
Costacurta
S.p.A.,
were
tested
in
both
the
vertical
and
horizontal
loop.
These
rigs
are
mainly
made
up
of
a
spray-generation
circuit
and
a
carrier
air
cir-
cuit.
The
spray
is
generated
using
an
ultrasonic
nozzle
fed
by
a
volumetric
pump,
giving
liquid
flow
rates
ranging
from
0
to
2,600
L/h,
and
by
a
compressor
supplying
air
at
6
atm
at
flow
rates
up
to
280
Nm
3
/h.
This
particular
nozzle
makes
it
possi-
ble
to
obtain
sprays
with
the
required
droplet
diameter
distri-
bution,
that
is,
a
great
number
of
drops
with
a
diameter
of
less
than
10
gm
and
an
overall
mean
diameter
of
less
than
20
m
The
test
section,
shown
in
Figure
2,
consists
of
a
4.5-m-long
Plexiglas
measuring
section
with
a
rectangular
cross
section,
120
mm
wide
and
190
mm
high.
The
use
of
a
Malvern
Parti-
cle
Sizer
instrument,
based
on
measurements
of
the
diffrac-
tion
of
an
He—Ne
laser
beam
by
droplets
moving
through
the
measuring
section,
allowed
accurate
measurements
of
total
concentration
and
of
the
volumetric
droplet
distribution.
Par-
ticular
care
was
given
to
the
acquisition
of
the
experimental
data
in
order
to
minimize
the
disturbances
due
to
the
mea-
suring
system
and
to
maximize
the
reproducibility
of
the
ac-
quired
data.
Each
datum
represents
the
mean
of
six
different
506
March
1998
Vol.
44,
No.
3
AIChE
Journal
ultrasonic
nozzle
Rae
Compressed
air
ater
vane—type
box
*P'rthr
Transmute
t
To
ton
Receive
unit
Laser
Drained
Figure
2.
Test
section
in
the
horizontal
experimental
loop.
acquisitions.
The
accuracy
of
the
measured
efficiency
is
quite
high
and
the
maximum
uncertainty
on
the
measured
effi-
ciency
is
below
5%.
Figure
3
shows
typical
droplet
distribu-
tions
upstream
and
downstream
from
the
mist
eliminator,
and
the
measured
efficiency
as
a
function
of
droplet
diameter.
This
article
analyzes
several
different
industrial
wire-mesh
mist
eliminators,
all
metallic.
The
main
geometric
character-
istics
of
each
packing
are
shown
in
Table
1.
It
must
be
em-
Packing
style
A
40
mm
pad
thickness
horizontal
setup
u
=
1
m/s
Liq
u
id
ho
ldup
(
%)
0.004
0.003
/
Inlet
0.002
Outlet
0.001
0
I-
0
20
40
60
80
Drop
diameter,d
d
(microns)
Figure
3.
Typical
separation
efficiency
(a)
and
droplet
distributions
before
and
after
the
mist
elimi-
nator
(b)
vs.
drop
diameter.
Table
1.
Geometric
Characteristics
of
the
Tested
Packings
Style
A
Wire
diameter
(mm)
0.27
0.27
0.27
0.15
0.15
Packing
density
(kg/m
3
)
145
190
200
128
190
Specific
area
(m
2
/m
3
)
267
360
363
459
643
Void
fraction
0.98
0.975
0.974
0.984
0.975
Pad
thickness
(mm)
150
[28]
150
[42]
65
[22]
150
[45]
65
[46]
[No.
of
layers]
65
[12]
100
[28]
40
[13]
40
[81
20
[7]
Equiv.
mesh
diameter
2.35
2.64
3.16
1.23
2.08
(mm)
phasized
that
wire
diameter,
specific
surface
area,
and
pad
thickness
are
varied
in
this
work
in
order
to
investigate
the
influence
of
each
parameter.
A
Model
to
Simulate
Wire-Mesh
Behavior
The
model
that
will
be
presented
is
based
on
the
following
assumptions:
(a)
no
reentrainment,
(b)
no
buildup
of
liquid,
and
(c)
no
mixing
after
passage
through
each
layer.
The
first
two
hypotheses
are
common
to
the
model
suggested
by
Car-
penter
and
Othmer
(1955),
whereas
the
last
represents
one
of
the
differences
between
the
present
model
and
the
works
published
to
date.
The
main
assumption
that
makes
the
present
model
differ-
ent
from
the
previous
works
is
the
new
schematization
of
the
separator.
It
goes
without
saying
that
using
the
real
geomet-
ric
characteristics
makes
it
necessary
to
develop
very
complex
models.
Unfortunately,
the
increased
complexity
of
computa-
tion
does
not
correspond
to
an
increase
of
accuracy
in
pre-
diction,
as
can
be
noted
by
comparing
the
present
experimen-
tal
results
with
the
theoretical
results
obtained
by
Feord
et
al.
(1993).
For
this
reason,
in
the
present
work
a
simplified
approach
is
suggested.
To
evaluate
the
separation
efficiency,
a
refer-
ence
cell
with
a
square
cross
section
whose
characteristic
length
will
be
defined
as
d
eq
and
a
thickness
given
by
a
num-
ber
of
layers
that
will
be
defined
as
being
equal
to
n,
has
been
introduced.
By
studying
the
behavior
of
one
of
these
cells,
which
are
identical
across
the
pad,
it
is
possible
to
eval-
uate
the
removal
efficiency
of
the
packing.
A
representation
of
a
single
layer
of
a
wire-mesh
pad
is
shown
in
Figure
I
a.
The
characteristic
length,
d
eq
,
has
been
evaluated
as
usu-
ally
suggested
for
the
equivalent
pipe
diameter
in
tubes
with
a
noncircular
cross
section,
cross
section
d
eq
4
.
wetted
perimeter
(4)
The
cross
section
is
given
by
the
product
between
the
pack-
ing
cross
section,
A
p
,
and
the
packing
void
fraction,
e,
while
the
wetted
perimeter,
P,
can
be
computed
as:
P—
(5)
where
1„
is
the
total
length
of
the
wires
in
the
packing,
z
is
the
distance
between
two
successive
layers,
and
V
p
is
the
vol-
ume
of
the
packing.
1
00
80
60
40
20
0
0.006
JJ
0.005
1„,•A
p
•z
V
AIChE
Journal
March
1998
Vol.
44,
No.
3
507
4
7r
D,„
a
e
z
(8)
n=1
n=2
n=3
C
o
P
M=0
From
the
definition
of
the
specific
surface,
a
e
,
it
follows
that
71"
D,„„•
a
e
V
P
and
therefore
a
e
Finally,
the
characteristic
dimension,
d
eq
,
can
be
evaluated
as:
A
P
d
eq
=
4—
A
z
V
P
P
Now
it
is
necessary
to
define
the
thickness
of
the
reference
cell.
A
mist
eliminator
pad
is
formed
by
a
large
number
of
layers
that
are
staggered
with
respect
to
the
others
and
that
are
set
in
such
a
way
as
to
cover
the
whole
cross
section
of
the
pad.
In
this
work
the
separator
is
schematized
as
com-
posed
of
wires
set
perpendicularly
to
the
gas-flow
direction.
With
reference
to
Figure
4,
it
follows
that
the
number
of
the
layers,
ri,
necessary
to
fill
each
cell
can
be
estimated
as:
=
d
e
q
D„,
(9)
To
evaluate
the
separation
efficiency
it
is
necessary
to
compute
the
concentration
of
the
particles,
C,„
in
the
gas
stream
after
a
generic
number
of
layers,
n.
Before
the
first
layer
the
concentration
of
the
carried
droplets
is
uniform
across
the
section
and
equal
to
C
o
.
As
underscored
previ-
ously,
in
the
present
model
we
assume
that
no
mixing
occurs
across
the
separator.
From
this
hypothesis
it
follows
that
only
the
particles
that
arrive
in
front
of
a
wire
can
eventually
be
separated
in
each
layer.
M=1
According
to
the
present
model,
to
predict
the
efficiency
of
the
separator
it
is
sufficient
to
analyze
the
behavior
of
the
reference
cell.
With
this
assumption,
the
fraction
of
the
cross
section
that
is
covered
by
wires
in
each
layer
can
be
evalu-
ated
as
1/
7t,
and
only
this
part
of
the
cross
section
partici-
pates
in
the
separation
process.
The
concentration
of
the
particles
after
the
first
layer,
C
1
,
can
finally
be
computed
as
—1)
1
C
i
=
C
o
+C
o
•—
--1
7s7
-
)
,
(10)
"ft
where
the
first
righthand
term
represents
the
particles
that
are
not
in
front
of
the
wires,
and
therefore
are
not
separated,
and
the
second
righthand
term
represents
the
particles
that
are
in
front
of
the
wires
but
are
not
separated,
the
latter
term
being
a
function
of
the
efficiency
of
the
single
target,
7
15T
Following
the
same
approach
for
the
second
layer,
Eq.
10
becomes
(n-2)
2
C
2
=
C
0
Tz
ri
C
(
1
7
),ST
),
(
11)
whereas
the
concentration
of
liquid
drops,
C„,
after
a
generic
number
of
layers
n
with
n
less
than
or
equal
to
the
number
of
layers
ri
necessary
to
cover
the
whole
cross
section
(see
Eq.
9),
is
n
n
C
n
=
C
o
+
C
o
—(1
71
5T
)
=
C
o
•(1
ns
T
)
(12)
When
the
number
of
layers
is
equal
to
n,
the
outlet
con-
centration
becomes:
c,=c0.(1
-
71sT)-
(13)
If
the
number
of
layers
is
greater
than
n
and
less
than
2.
the
droplet
concentration
in
front
of
the
wires
becomes
equal
to
C
r
,.
The
generic
concentration
of
droplets
after
a
generic
number
of
layers
n
with
rz
<
n
<
2
Tz
can
thus
be
computed
as
(
n
1
7
i)
(
n
C
n
=
C
h
-•[
_
n
+
(1
7
1,57)1
=
//sr
)
[n—(n—n)
+
(n
fi)
(1
77
sTd.
(14)
n
1
7
1
If
the
number
of
the
layers
is
absolutely
generic,
it
is
nec-
essary
to
evaluate
the
number,
M,
of
reference
cells
that
are
present
in
the
pad.
This
parameter
is
a
function
of
the
num-
ber
of
layers
that
form
the
separator,
n,
and
of
the
number
of
layers,
l
W
V
P
7r
*D
m.
,
(6)
(7)
c
o
C
o
C
o
CD
,
C
o
p,
d
eg
M
=
int
n
(15)
Figure
4.
Droplet
flow
within
the
packing.
508
March
1998
Vol.
44,
No.
3
AIChE
Journal
100
80
60
40
20
0.)
0
4
6
100
0
80
a
.
03
"
The
concentration
of
the
droplets
in
the
gas
stream
after
M
n
layers
can
be
computed
as
C
M
Tr
=
C
Al
T1
-
7
1ST
)
=
C
M
-2Fi
•(1—
T
IST
)2
=
C
M
3,74
1
1
7$0
3
"
=
C
0
4
1
%O
m
(16)
The
concentration
after
a
generic
number
of
layers,
n,
can
be
evaluated
as
(n
M
Tz)
(n
M
+
_
(1
77s01
Ft
n' n'
=
C
0
41
n
sT
)'"
•[
n
_
+
n
_
7
1sT)',
(17)
where
n
=
n
n
M
represents
the
number
of
layers
that
are
not
sufficient
to
form
a
complete
cell.
Finally,
the
wire-mesh
separation
efficiency
of
a
bed
with
a
generic
number
of
layers
can
be
evaluated
as
m
[h
.
n' n'
=
I
0
nsT)
_
+
_
(1
71sT)
n
n
to
an
array
of
targets
that
are
close
to
each
other.
This
is
the
case
of
wire-mesh
mist
eliminators,
and
for
this
reason
in
the
present
article
the
following
empirical
relation
has
been
in-
troduced
as
closure
equation:
if
St
<
1
then
7
1sT
=
St,
(19)
whereas
if
St
>
1
then
TN
T
=1,
(20)
where
the
Stokes
number
has
been
defined
according
to
Eq.
2.
This
simplified
hypothesis
is
justified
by
analysis
of
Figure
5,
where
all
the
experimental
data
obtained
in
the
horizontal
loop
have
been
plotted.
It
will
be
noted
that,
with
the
partial
exception
of
packing
type
A,
40
mm
thick,
all
the
experimen-
tal
data
obtained
at
Stokes
numbers
greater
than
1
display
separation
efficiencies
of
nearly
100%.
These
results
disagree
with
the
results
obtained
using
the
relation
by
Langmuir
and
Blodgett
(1946),
and
one
reason
for
this
discrepancy
may
be
the
complex
geometry
of
wire-mesh
mist
eliminators,
which
causes
a
mutual
influence
between
single
targets.
C„
=
C
m
.,
(18)
Equation
18
makes
it
possible
to
compute
the
separation
efficiency
if
the
efficiency
of
a
single
target,
n
sT
,
is
known.
In
the
literature,
many
different
equations
are
available
to
com-
pute
the
efficiency
of
a
single
target
and
one
of
the
most
commonly
used
has
been
suggested
by
Langmuir
and
Blod-
gett
(1946).
As
pointed
out
by
Lucas
(1983),
the
theoretical
approach
by
Langmuir
and
Blodgett
(1946)
can
induce
un-
derestimation
of
the
separation
efficiency
when
it
is
applied
100
80
ii
.
4
.
1
4.
'I
,
At
y•
Ai
A
A.
vAA
t
vlo
A
-
n
I
Packing
style
A
A
A
60
A
o
V
B
*•
C
I
A
D
73'
03
40
AV
*E
vp
A
20
0
0
1
1
10
Stokes
number,
St
Figure
5.
Separation
efficiency
vs.
Stokes
number;
all
the
experimental
data
obtained
in
horizontal
setup.
U
(a)
60
40
20
(b)
0
0
5
10
15
20
Drop
diameter,d
d
(microns)
Figure
6.
Separation
efficiency
vs.
drop
diameter:
(a)
packing
style
A,
2
m/s
superficial
gas
veloc-
ity,
pad
150
mm
thick;
(b)
packing
style
D,
1
m/s
superficial
gas
velocity,
pad
150
mm
thick.
Vertical
setup;
O
horizontal
setup.
AIChE
Journal
March
1998
Vol.
44,
No.
3
509
(b)
0
o,
00
0
0
0
0
0
0
0
c
9
a
I l
et:
i•
*
(a)
Horizontal
setup
S
(b)
100
90
80
70
100
Measured
efficiency,11,
(%)
Figure
7.
Separation
efficiency.
Comparison
between
experimental
measurements
with
(a)
Carpenter
and
Othmer
(1955)
model,
and
(b)
present
model
(vertical
setup).
Analysis
of
Experimental
Results
When
a
gas/liquid
separator
has
to
be
decided
on,
the
first
problem
is
to
choose
between
the
horizontal
and
the
vertical
configuration.
From
the
point
of
view
of
separation
effi-
ciency,
differences
arise
if
the
working
conditions
are
close
to
the
flooding
point
or
if
there
is
a
high
value
of
the
slip
veloc-
ity
between
the
gas
and
liquid
phases.
In
horizontal
flow
the
force
of
gravity
does
not
modify
the
axial
gas
velocity,
whereas
in
the
vertical
configuration
it
induces
a
difference
between
the
gas,
flowing
faster,
and
the
carried
liquid
droplets.
If
analysis
is
restricted
to
low
gas
velocity,
in
a
range
of
working
conditions
where
reentrainment
can
be
neglected,
no
differ-
ence
seems
to
arise
in
the
experimental
separation
efficiency.
This
experimental
behavior
can
be
explained
if
the
slip
veloc-
ity
between
the
gas
and
liquid
phases
is
computed.
In
the
case
of
an
air—water
system
working
at
room conditions,
droplets
with
diameters
less
than
30-35
ktm
obey
Stokes'
law.
In
this
case
the
limiting
drop
velocity
can
be
easily
computed,
and
it
is
possible
to
deduce
that
the
limiting
drop
velocity,
and
consequently
the
slip
velocity,
remains
below
the
value
of
0.04
m/s
and
therefore
can
be
neglected.
This
observation
allows
us
to
conclude
that
from
the
point
of
view
of
separa-
tion,
outside
the
range
where
flooding
phenomena
occur,
the
horizontal
and
vertical
configurations
should
yield
the
same
performances.
This
theoretical
result
is
confirmed
by
experi-
100
90
80
70
60
50
c.)
40
-c
100
c.
90
80
70
60
50
40
40
50
60
70
80
90
100
Measured
efficiency.%
(%)
Figure
8.
Separation
efficiency.
Comparison
between
experimental
measurements
with
(a)
Carpenter
and
Othmer
(1955)
model,
and
(b)
present
model
(horizontal
setup).
mental
evidence,
as
shown
in
Figure
6.
where
experimental
data
relative
to
two
different
packings
are
plotted.
It
is
now
possible
to
compare
the
experimental
data
ob-
tained
in
this
work
and
the
computed
values
obtained
using
both
the
present
model
and
the
model
suggested
by
Carpen-
ter
and
Othmer
(1955).
Figure
7
shows
the
parity
plot
relative
to
the
vertical
setup.
Analysis
of
the
figure
shows
that
the
model
makes
it
possible
to
evaluate
separation
efficiency
with
a
slightly
higher
accuracy
than
with
the
model
suggested
by
Carpenter
and
Othmer
(1955).
The
same
results
were
ob-
tained
for
the
horizontal
setup
(see
Figure
8).
Also
in
this
case,
the
model
suggested
by
Carpenter
and
Othmer
(1955)
introduces
a
systematic
error,
overestimating
the
experimen-
tal
separation
efficiency.
Higher
accuracy,
with
respect
to
the
results
obtained
using
Carpenter
and
Othmer's
model,
is
not
the
most
important
goal
achieved
using
the
present
model.
In
the
last
few
years
a
new
tendency
in
separation
design
has
emerged,
and
nowa-
days
packings
made
up
of
two
mesh
pads
in
series
are
often
preferred
to
thick
packings
made
from
a
single
type
of
mesh.
The
first
pad,
made
of
fine
wires,
works
beyond
the
flood
point.
Because
of
high
gas
velocity,
it
captures
droplets
of
small
size
that
coalesce
in
larger
drops
and
some
of
them
are
reentrained
in
the
gas
phase.
These
large
drops
are
easily
separated
in
the
second
pad
working
below
the
flood
point.
Because
the
first
pad
works
beyond
the
flood
point,
at
high
60
50
40
ta.
100
90
80
70
60
50
40
40
50
60
70
80
90
Vertical
setup
0
C
0
0
0
Cl
IT]
(a)
510
March
1998
Vol.
44,
No.
3
AIChE
Journal
100
707
5-6
7
79—e--ee
-----------
°
0
0 °
100
95
90
85
80
75
70
100
U
0
0
65
mm
(a)
pad
thickness
28
mm
pad
thickness
(a)
Square
mesh
20
GS
)
90
80
-
40
mm
(b)
pad
thickness
0
0
0
U
20
mm
(c)
pad
thickness
10
20
30
40
50
Drop
diameter,d
d
(microns)
Figure
9.
Separation
efficiency,
effect
of
packing
length.
Packing
style
C,
superficial
gas
velocity
2
m/s.
(a)
packing
65
trim
thick;
(b)
packing
40
mm
thick;
(c)
packing
20
mm
thick
(present
model
continuous
lines,
Carpenter
and
Othmer
(1955)
model
dashed
lines).
gas
velocity,
it
does
not
require
a
large
number
of
layers;
thus,
for
proper
design,
the
separation
efficiency
must
be
predicted
with
high
accuracy
for
thinner
packings
than
those
tested
so
far.
Figure
9
shows
a
comparison
between
the
experimental
data
obtained
in
this
article,
the
present
model,
and
the
model
proposed
by
Carpenter
and
Othmer
(1955).
The
acquisitions
refer
to
three
packings
with
the
same
geometrical
properties,
differing
only
in
packing
thickness.
It
will
be
noted
that
the
two
models
agree
for
experimental
data
for
pad
thicknesses
greater
than
65
mm,
a
dimension
commonly
tested
in
papers
published
so
far;
if
the
packing
size
is
smaller,
the
relation
suggested
by
Carpenter
and
Othmer
(1955)
introduces
sys-
tematic
errors
that
increase
with
decreasing
pad
thickness.
This
result
must
be
emphasized
because,
as
pointed
out
ear-
lier,
the
trend
is
toward
packings
composed
of
different
pads
that
will
be
thinner
than
the
pads
used
and
tested
so
far.
All
the
experimental
data
illustrated
earlier
were
obtained
in
the
present
work.
Biirkholz
(1970)
performed
a
systematic
experimental
study
on
wire-mesh
mist
eliminators.
He
ana-
lyzed
the
behavior
of
a
commercial
mesh
pad
and
the
behav-
ior
of
two
other
homemade
packings
with
a
perfectly
square
mesh
set
perpendicular
to
the
direction
of
gas
flow.
The
ex-
perimental
results
obtained
on
the
second
type
of
packing
0
0
0
100
op
80
58
mm
pad
thickness
60
-;
40
'
Predicting
models:
Carpenter
&
Othmer,
1955
---
Present
model
0
0
0
2
4
6
Stokes
number,
St
Figure
10.
Separation
efficiency
vs.
Stokes
number.
Comparison
between
experimental
measurements
with
Carpenter
and
Othmer
(1955)
model
(dashed
lines)
and
present
model
(continuous
lines).
Experimental
data
ob-
tained
by
Biirkholz
(1970),
square
mesh
(a)
packing
28
mm
thick;
(b)
packing
58
mm
thick.
are
extremely
important
because
in
this
case
the
equivalent
mesh
diameter,
d
eq
,
can
be
easily
measured
and
assumes
the
dimension
of
the
square
mesh
side.
Figure
10
shows
the
com-
parison
between
the
present
model,
the
model
by
Carpenter
and
Othmer
(1955),
and
the
experimental
results
obtained
by
Biirkholz
(1970)
using
two
pads
of
square
mesh
with
different
thicknesses.
It
can
be
seen
that
the
present
model
predicts
experimental
results
with
high
accuracy,
whereas
the
model
by
Carpenter
and
Othmer
(1955)
induces
systematic
errors,
especially
for
the
thinner
pad.
Figure
11
shows
experimental
data
obtained
by
Biirkholz
(1970)
using
a
wire-mesh
mist
eliminator
and
the
theoretical
trends
obtained
with
the
pre-
sent
model
and
with
the
model
by
Carpenter
and
Othmer
(1955).
Also
in
this
case,
in
the
working
range
of
the
separa-
tor
where
the
separation
efficiency
is
greater
than
50%,
higher
accuracy
is
obtained
using
the
present
model
than
with
the
model
by
Carpenter
and
Othmer
(1955).
The
last
parameter
that
must
be
analyzed
for
correct
de-
sign
of
wire-mesh
separators
is
dp
95
(Biirkholz,
1986).
By
fix-
ing
the
geometrical
characteristics
of
the
unit
and
the
work-
ing
conditions,
dp
95
represents
the
smallest
droplet
diameter
that
can
be
separated
with
an
efficiency
greater
than
95%.
Figure
12
shows
a
comparison
between
experimental
data
ob-
tained
in
this
work
and
by
Biirkholz
(1970),
and
computed
70
65
I
00
90
80
70
60
50
0
20
0
(b)
8
10
AIChE
Journal
March
1998
Vol.
44,
No.
3
511
4
a
Gas
velocity
(m/s)
01
20
A
12
6
2.8
1.4
0
0.7
0.35
—Present
model
Carpenter
&
Othmer,
1955
/
0
ri,
/O
I
0
40
Present
experimental
data
Present
model
30
Carpenter
&
Othmer,
1955
Experimental
data
by
Biirkholz,
1970
Commercial
pad
20
Homemade
pads
0
10
0
0
-ct
0
0.
0
0
100
80
Sep
ara
t
io
n
e
ffic
iency
:
9,
(
%)
60
40
20
0
0
2
4
6
8
10
Stokes
number,
St
Figure
11.
Separation
efficiency
vs.
Stokes
number.
Comparison
between
experimental
measurements
with
Carpenter
and
Othmer
(1955)
model
(dashed
lines)
and
present
model
(continuous
lines).
Experimental
data
ob-
tained
by
Biirkholz
(1970),
mesh
pad
with
a
specific
sur-
face
area
of
77
m
2
/m
3
and
a
thickness
of
65
mm.
data,
according
to
the
present
relation
and
to
the
relation
by
Carpenter
and
Othmer
(1955).
It
must
be
pointed
out
that
the
comparison
between
experimental
data
obtained
by
Biirkholz
(1970)
and
theoretical
models
has
been
limited
to
the
present
model
because,
with
the
model
of
Carpenter
and
Othmer
(1955),
an
efficiency
of
95%
is
not
achievable.
The
figure
shows
that
the
present
relation
allows
predicting
dp
95
with
acceptable
accuracy.
Conclusions
The
experimental
data
on
droplet
removal
efficiency
pre-
sented
in
this
article
were
obtained
using
a
laser-based
droplet
sizer,
the
Malvern
Particle
Sizer.
This
technique
makes
it
possible
to
obtain
nonintrusive
measurements
and,
above
all,
it
allows
measuring
both
the
concentration
and
the
size
of
the
droplets.
New
experimental
data
on
vertical
and
horizon-
tal
setups
were
obtained
showing
that,
if
no
reentrainment
occurs,
the
horizontal
and
vertical
configurations
display
sim-
ilar
removal
efficiencies.
This
article
has
presented
a
new
model
for
predicting
re-
moval
efficiency.
Analysis
of
the
experimental
data
obtained
in
this
article
and
of
the
limited
published
data
shows
that
the
present
model
and
the
model
published
by
Carpenter
and
Othmer
(1955)
agree
for
packings
with
thicknesses
greater
than
65
mm.
For
thinner
pads
the
model
suggested
by
Car-
penter
and
Othmer
(1955)
systematically
underestimates
the
experimental
efficiencies,
whereas
the
present
model
enables
good
prediction
of
experimental
removal
efficiencies.
This
is
an
important
improvement
because
separation
units
will
be
formed
by
two
mesh
pads
in
series
to
optimize
pressure
drop/removal
efficiency
and
the
size
of
these
pads
is
gener-
ally
smaller
than
the
critical
dimension
of
65
mm.
Acknowledgments
The
work
on
which
this
publication
is
based
was
supported
by
Costacurta
S.p.A.,
Via
Grazioli
30,
20161
Milan,
Italy.
The
authors
thank
Ing.
A.
Vitaletti
and
Ing.
M.
Vettori
for
their
helpful
assist-
ance,
and
Ing.
B.
Mondello
for
some
useful
discussions.
0
0
2
4
6
8
10
12
14
d
95
,
measured
value
(microns)
Figure
12.
Comparison
between
experimental
measure-
ments
of
d
95
obtained
in
this
work
and
by
BOrkholz
(1970)
with
Carpenter
and
Othmer
(1955)
model
and
present
model.
Comparison
between
present
measurements
and
pres-
ent
model;
0
comparison
between
present
measurements
and
Carpenter
and
Othmer
(1955)
model;
comparison
between
experimental
data
by
Biirkholz
(1970),
commer-
cial
pads,
and
present
model;
comparison
between
ex-
perimental
data
by
Biirkholz
(1970),
homemade
pads,
and
present
model.
Notation
C
o
=
inlet
liquid
drop
concentration
7T
=
3.
1
4159
.
.
.
Literature
Cited
Bilrkholz,
A.,
"Die
Beschreibung
der
Partikelabscheidung
durch
Tragheitskrafte
mit
Hilfe
einer
dimensionsanalytisch
abgeleiteten
Kennzahl,"
Chem.
Ing.
Tech.,
58,
548
(1986).
Biirkholz,
A.,
"Tropfenabscheidung
an
Drahtfiltern,"
Chem.
Ing.
Tech.,
42,
1314
(1970).
Carpenter,
C.
L,
and
D.
F.
Othmer,
"Entrainment
Removal
by
a
Wire-Mesh
Separator,"
AIChE
J.,
1,
549
(1955).
Feord,
D.,
E.
Wilcock,
and
G.
A.
Davies,
"A
Stochastic
Model
to
Describe
the
Operation
of
Knitted
Mesh
Mist
Eliminators,
Com-
putation
of
Separation
Efficiency,"
Trans.
Ind.
Chem.
Eng.,
71,
282
(1993).
Gerrard,
M.,
G.
Puc,
and
E.
Simpson,
"Optimize
the
Design
of
Wire-Mesh
Separators,"
Chem.
Eng.,
93(21),
91
(1986).
Holmes,
T.
L.,
and
G.
K.
Chen,
"Design
and
Selection
of
Spray/Mist
Elimination
Equipment,"
Chem.
Eng.,
91(21),
82
(1984).
Langmuir,
I.,
and
K.
B.
Blodgett,
U.S.
Army
Air
Forces
Tech.
Rep.,
5418
(1946).
Lucas,
R.
L.,
"Gas-Solid
Separations,"
Chemical
Engineers'
Hand-
book,
5th
ed.,
Chap.
20,
R.
H.
Perry
and
C.
H.
Chilton,
eds.,
Mc-
Graw-Hill,
New
York,
p.
81
(1983).
Pich,
J.,
Aerosol
Science,
Chap.
9,
C.
N.
Davies,
ed.,
Academic
Press,
New
York
(1966).
York,
0.
H.,
"Performance
of
Wire-Mesh
Demisters,"
Chem.
Eng.
Prog.,
50,
421
(1954).
Manuscript
received
Dec.
6,
1996,
and
revision
received
June
4,
1997.
512
March
1998
Vol.
44,
No.
3
AIChE
Journal