Re-entrainment in wave-plate mist eliminators


Azzopardi, B.J.; Sanaullah, K.S.

Chemical Engineering Science 57(17): 3557-3563

2002


Wave-plate mist eliminators are among the most effective devices to separate liquid from the gas. Liquid separation efficiency in these devices is largely dependent on the gas superficial velocity. However, the upper limit of gas throughput is restricted by the re-entrainment of the deposited liquid from the separator surfaces into the gas stream. There is some knowledge available of re-entrainment phenomena in vertical wave-plate separators. This paper presents experiments that have enabled the mechanisms of entrainment to be identified and the boundaries for air and liquid rates at which entrainment starts for horizontal wave-plate separators to be characterised. It is supported by modelling using an extension of earlier work.

Chemical
Engineering
Science
PERGAMON
Chemical
Engineering
Science
57
(2002)
3557-3563
www.elsevier.com/locate/ces
Re-entrainment
in
wave-plate
mist
eliminators
B.
J.
Azzopardi
*,
K.
S.
Sanaullah
Multiphase
Flow
Research
Group,
School
of
Chemical,
Environmental
and
Mining
Engineering,
University
of
Nottingham,
University
Park,
Nottingham
NG7
2RD,
UK
Received
24
October
2001;
received
in
revised
form
11
June
2002;
accepted
13
June
2002
Abstract
Wave-plate
mist
eliminators
are
among
the
most
effective
devices
to
separate
liquid
from
the
gas.
Liquid
separation
efficiency
in
these
devices
is
largely
dependent
on
the
gas
superficial
velocity.
However,
the
upper
limit
of
gas
throughput
is
restricted
by
the
re-entrainment
of
the
deposited
liquid
from
the
separator
surfaces
into
the
gas
stream.
There
is
some
knowledge
available
of
re-entrainment
phenomena
in
vertical
wave-plate
separators.
This
paper
presents
experiments
that
have
enabled
the
mechanisms
of
entrainment
to
be
identified
and
the
boundaries
for
air
and
liquid
rates
at
which
entrainment
starts
for
horizontal
wave-plate
separators
to
be
characterised.
It
is
supported
by
modelling
using
an
extension
of
earlier
work.
©
2002
Elsevier
Science
Ltd.
All
rights
reserved.
Keywords:
Mist
eliminator;
Gas/liquid;
Drops;
Re-entrainment
1.
Introduction
It
is
sometimes
important
to
remove
small
quantities
of
drops
from
gas
streams
as
this
liquid
could
cause
upset
or
damage
to
downstream
equipment.
Mist
eliminators
are
de-
vices
that
can
remove
entrained
liquid
from
a
gas
flow
effec-
tively,
usually
by
inertial
impingement.
In
mist
eliminators
of
the
wave-plate
type,
the
gas
is
forced
to
travel
in
a
zig-zag
manner
between
pairs
of
appropriately
shaped
plates.
Drops
with
their
higher
inertia
cannot
follow
these
changes
in
di-
rection
and
so
impinge
on
to
the
solid
surfaces.
There
they
adhere,
coalesce
and,
when
the
amount
of
liquid
is
suffi-
ciently
high,
form
a
film
which
drains
away
under
gravity.
For
wave-plate
units
where
the
gas
flows
vertically
upwards,
this
drainage
is
counter-current
to
the
gas
flow.
Where
the
gas
flows
horizontally
through
the
unit
the
drainage
is
per-
pendicular
to
the
gas
flow.
The
inertia
of
the
drops
and
the
drag
of
the
gas
control
the
motion
of
drops
through
the
zig-zag
passages.
This
led
Burkholz
(1989),
amongst
others,
to
describe
the
effi-
ciency
through
the
concept
that
the
turning
of
the
gas
could
centrifuge
drops
out
of
the
gas
stream.
Drop
size,
plate
spacing
and
bend
angle
as
well
as
fluid
properties
were
*
Corresponding
author.
Tel.:
+44-115-951-4160;
fax:
+44-115-951-4115.
E-mail
address:
(B.
J.
Azzopardi).
important.
The
effect
of
multiple
bends
was
handled
by
as-
suming
re-dispersion
after
each
bend
and
repeating
the
cal-
culation
process.
The
cumulative
effect
is
then
determined
by
multiplying
together
each
of
the
single
bend
penetrations
(=1—efficiency).
More
recent
work,
by
Wang
and
Davies
(1996),
Wang
and
James
(1998,
1999),
has
employed
Com-
putational
Fluid
Dynamics
to
calculate
drop
deposition.
Both
experiment
and
theory
indicate
that
the
higher
separation
efficiencies
might
be
gained
by
increasing
the
drop
inertia
via
the
gas
velocity
though
at
the
penalty
of
increased
pres-
sure
drop.
However,
even
the
pioneering
work
of
Houghton
and
Radford
(1939)
reported
that
a
sharp
decrease
occurred
in
the
separation
efficiency
for
gas
velocities
greater
than
a
critical
value.
In
their
case
this
velocity
was
6
m/s.
This
was
attributed
to
re-entrainment
of
the
collected
liquid.
For
vertical wave-plates
(gas
upflow,
downwards
drainage
of
the
liquid)
the
mechanism
for
the
decrease
in
efficiency
has
been
attributed
to
flooding
of
the
draining
film
by
the
upward
shear
of
the
gas
(Verlaan,
1991).
Flooding
is
the
condition
at
which
an
upwards
gas
flow
starts
to
pre-
vent
the
downflow
of
liquid.
From
their
experiments
with
air/water
at
ambient
conditions,
Monat,
McNulty,
Michel-
son
and
Hansen
(1986)
proposed
that
re-entrainment
would
occur
when
a
critical
value
of
dimensionless
re-entrainment
number
(u
4
g
p
2
g
/p/go
-
)
was
exceeded.
Here
u
g
and
p
g
are
the
gas
velocity
and
density,
respectively,
p
1
is
the
liquid
density,
g
the
gravitational
acceleration
and
o
-
the
surface
tension.
This
group
is
the
fourth
power
of
the
dimension-
0009-2509/02/$
-
see
front
matter
©
2002
Elsevier
Science
Ltd.
All
rights
reserved.
PII:
S0009-2509(02)00270-1
3558
B.
J.
Azzopardi,
K
S.
Sanaullahl
Chemical
Engineering
Science
57
(2002)
3557-3563
less
velocity,
usually
called
the
Kutatelathe
number,
a
parameter
much
used
in
the
analysis
of
flooding
processes.
Though
there
is
information
on
critical
conditions
for
re-entrainment
for
vertical
wave-plate
mist
eliminators,
no
such
information
exists
for
horizontal
types.
This
paper
reports
on
part
of
a
wider
study,
which
has
also
mea-
sured
overall
efficiency,
variation
of
drop
sizes
along
the
demister
and
pressure
drop
(Azzopardi
et
al.,
2002).
Here
an
experimental
and
theoretical
study
that
sheds
light
on
the
phenomena
occurring
during
re-entrainment
of
liquid
in
a
horizontal
wave-plate
system
is
reported.
The
results
are
related
to
published
information
on
vertical
upflow
systems.
2.
Experimental
arrangements
Two
series
of
experiments
are
reported.
The
first
considers
the
effect
of
a
horizontal
gas
flow
across
a
film
flowing
under
gravity
on
one
wall
of
a
straight
channel.
This
provides
un-
derlying
information
on
the
interfacial
features
of
this
type
of
cross-flow.
The
second
employed
one
passage
between
two
wave-plates
where
the
effect
of
the
exact
geometry
could
be
observed.
2.1.
Straight
channel
The
cross-flow
experiments
were
carried
out
in
a
once-through
wind
tunnel.
A
pair
of
axial
flow
fans
drew
air
though
a
calming
section
and
then
the
test
section,
both
of
whose
walls
were
made
of
plate
glass.
This
section
was
0.305
m
high
and
0.0445
m
wide.
A
liquid
film
was
formed
over
0.56
m
of
one
wall
by
allowing
liquid
to
flow
over
a
knife-edge
weir
after
passing
through
a
calming
section.
Great
care
was
taken
to
ensure
that
the
liquid
flow
rate
was
uniform
over
that
length
of
wind
tunnel.
Quantitative
data
of
the
spatial
variation
of
film
thickness
was
obtained
using
a
liquid
absorption
technique.
A
camera
and
light
source
were
placed
on
either
side
of
the
test
sec-
tion.
A
soluble
dye,
nigrocine,
was
dissolved
in
the
water
to
increase
the
absorptivity.
The
photographs
were
developed
and
grey
scales
measured
using
a
densitometer.
Grey
scale
was
related
to
film
thicknesses
using
a
calibration
curve
ob-
tained
on
a
wedge
cell,
which
contained
films
of
known
thicknesses.
Full
details
can
be
found
in
Azzopardi
(1977).
2.2.
Single
channel
wave-plate
These
experiments
were
carried
out
in
a
simple,
open-circuit
wind
tunnel.
Air
from
a
blower
passed
through
a
calming
section
of
rectangular
cross-section
and
then
through
a
small-angle
convergent
section
before
entering
the
single
channel
wave-plate.
This
had
a
height
of
0.13
m,
a
spacing
of
0.0115
m,
a
bend
angle
of
and
an
individ-
ual
plate
width
of
0.023
m.
The
airflow
rate
was
altered
by
a
damper
at
the
blower
outlet
and
measured
using
a
Pitot-static
tube
at
the
end
of
the
wave-plate.
Normally
liquid
films
in
wave-plate
mist
eliminators
are
formed
by
deposition
of
drops
with
the
flow
rate
of
the
gravity-driven
film
having
a
cumulative
value
dependent
on
the
vertical
position.
Now,
industrial
wave-plate
mist
elimi-
nators
can
be
1
m
tall
or
more.
It
is
not
easy
to
reproduce
this
dimension
and
the
corresponding
film
flow
rate
in
the
labo-
ratory.
Therefore,
an
alternative
method
of
creating
the
film
was
employed.
Liquid
was
allowed
to
flow
over
the
top
of
one
of
the
flat
plates
forming
the
wave-plate
from
a
cavity
ar-
ranged
on
the
outside
through
a
pipe
in
the
bottom
as
shown
in
Fig.
1.
Water
was
supplied
from
a
constant
head
tank.
Tap
Water
AIR
-
F.!
1%1
s.
T.
1
\
-
Fig.
1.
Wave-plate
with
liquid
film
injection
arrangement.
Mean
gas
velocity
(m/s)
11.5
13.3
15
17.3
IN
-
*
o
9
B.
J.
Azzopardi,
K
S.
Sanaullah
I
Chemical
Engineering
Science
57
(2002)
3557-3563
3559
40
0
0
200
400
600
800
1000
1200 1400 1600 1800
1
S
Str
ia
t
ion
sp
a
c
ing
(mm
)
25
20
15
10
5
Fig.
2.
Falling
film
waves
at
zero
gas
flow
(4—
direction
of
gas
flow).
-fr
V
o
r#1
:
41
%,
-
Liquid
Reynolds
number
(-)
Fig.
4.
Effect
of
gas
and
liquid
flow
rates
on
striation
spacing.
(Closed
symbols)
water,
(open
symbols)
aqueous
glycerine
solution,
kinematic
viscosity
=
6
10
-6
m
2
/s.
Liquid
Reynolds
number
Water
Gylceml
solution
224
624
1080
1600
164
340
576
II
*
0
0
0
30
Nail%
k
1
4,
iae
t#
Aft
25
20
15
1
0
5
4—
Direction
of
gas
flow
0
0
100
200
300
400
500
Str
ia
t
io
n
ang
le
(
°)
Fig.
3.
(Striations)
Gas
Reynolds
number
=
39500,
Liquid
Reynolds
number
=
224.
3.
Results
3.1.
Straight
channel
Observations
were
made
of
the
behaviour
of
the
interface
using
photography
with
backlighting.
At
low
gas
flow
rates,
the
surface
is
covered
by
waves
travelling
vertically
down-
wards
as
shown
in
Fig.
2.
Beyond
a
critical
velocity,
the
appearance
of
the
interface
changes
dramatically
and
those
waves
are
replaced
by
a
regular
system
of
nearly
vertical
structures,
termed
striations.
These
are
similar
to
rivulets
but
have
thin
films
between
them
instead
of
dry
patches.
Fig.
3
shows
how
the
falling
film
waves
are
quickly
converted
to
striations.
In
what
follows,
the
gas
Reynolds
number
is
defined
as
p
g
u
g
sM
g
.
ri
g
is
gas
dynamic
viscosity.
The
film
Reynolds
number
is
defined
as
4FM/
where
F
=
W/B;
W
is
Square
of
maximum
gas
velocity
(m
2.
/s
2
)
Fig.
5.
Effect
of
gas
velocity
squared
on
striation
angle.
liquid
mass
flow
rate
(kg/s)
and
B
is
the
width
of
one
of
the
flat
surface
over
which
the
film
flows.
The
distance
down
into
the
wind
tunnel
that
is
required
for
the
occurrence
of
striations
was
found
to
increase
with
gas
flow
rate
and
to
de-
crease,
to
a
lesser
extent,
with
liquid
flow
rate.
The
regularity
of
the
striations
can
be
seen
in
Fig.
3.
The
measured
values
of
wavelength,
Fig.
4,
are
insensitive
to
both
gas
and
liquid
flow
rates.
The
striation
angle
increased
as
the
gas
velocity
squared
and
showed
a
smaller
dependence
on
liquid
flow
rate
Fig.
5.
The
variation
of
film
thickness
across
striations
was
found
to
be
very
regular
as
illustrated
in
Fig.
6.
The
thicknesses
increased
with
liquid
flow
rate
but
were
insensi-
tive
to
gas
flow
rate.
This
is
illustrated
in
Fig.
7,
which
also
shows
that
mean
film
thicknesses
were
slightly
less
than
the
values
predicted
by
the
theory
of
Nusselt
(1916)
for
films
15
20
25
30
35
Table
1
Conditions
at
inception
of
re-entrainment
Liquid
mass
Film
thickness
Gas
velocity
Weber
(mm)
(m/s)
number
flow
rate
(kg/s)
0.014
0.57
5.0
4.7
0.012
0.54
5.6
5.9
0.008
0.47
7.4
10.4
0.0072
0.46
8.2
12.3
1.6
1A
L2
1
Film
t
hic
kness
(mm
)
0.8
0.6
0.4
0.2
0
3560
B.
J.
Azzopardi,
K
S.
Sanaullahl
Chemical
Engineering
Science
57
(2002)
3557-3563
Distance
along
film
(mm)
Fig.
6.
Spatial
variation
of
film
thickness
in
direction
of
gas
flow-Gas
Reynolds
number
=
39
500,
Liquid
Reynolds
number
=
1600.
Mean
gas
velocity
(m/s)
11.5
13.3
El
0
500
1,000
1,500
2,000
Liquid
Reynolds
number
(-)
Fig.
7.
Mean
film
thickness
measured
for
a
falling
film
with
a
transverse
gas
flow.
(Closed
symbols)
Water;
(open
symbols)
aqueous
glycerol
solution-kinematic
viscosity
=
6
10
-6
m
2
/s.
Lines
are
predictions
from
Nusselt
model.
of
liquid
travelling
down
a
vertical
plate
under
the
influence
of
gravity.
Other
workers
who
studied
falling
films
without
gas
flow
have
found
a
similar
result.
The
thicknesses
at
the
crests
of
the
striations
were
found
to
be
twice
the
mean
film
thickness.
The
existence
of
striations
and
their
behaviour
can
be
ex-
plained
by
stability
analysis
and
other
physical
arguments.
Estimations
of
the
vertical
and
lateral
velocities
showed
that
the
film
flow
downwards
was
larger
than
that
across.
Be-
cause
the
film
thickness
was
close
to
that
from
Nusselt
the-
ory,
it
could
be
assumed
that
the
effects
of
gravity
and
shear
are
uncoupled.
3.2.
Single
channel
wave-plate
In
this
case,
a
digital
camera
(Kodak
DC120)
and
high-speed
video
were
used
to
record
the
behaviour
of
the
liquid
film
falling
down
the
plate.
Liquid
flow
rates
were
set
and
the
gas
velocity
through
the
zig-zag
was
increased
slowly
until
disruption
of
the
liquid
film
occurred.
The
Sub-critical
Critical
Re-entrainment
-0--
B
C
*
1000
2000
3000
4000
5000
6000
7000
Gas
Reynolds
number
Fig.
8.
Re-entrainment
boundary
for
a
horizontal
wave-plate
mist
elimi-
nator.
mean
air
velocity
is
varied
from
2
to
8.4
m/s.
The
flows
at
which
the
phenomena
were
recorded
are
given
in
Table
1.
Here,
the
Weber
number
is
defined
as
2
paU
g
S
We
=
(
1
)
tT
where
s
is
spacing
between
wave-plates.
The
capacity
of
a
wave-plate
mist
eliminator
is
usually
set
by
the
critical
liquid
and
gas
rates
above
which
the
efficiency
of
liquid
removal
begins
to
decrease.
There
is
a
gas
velocity
level
at
which
droplets
begin
to
be
stripped
off
the
internal
surfaces
of
the
wave-plate
separator.
Fig.
8
shows
conditions
at
which
such
a
phenomenon
occurs.
It
appears
that
break
up
of
the
liquid
film
depends
on
the
velocity
of
the
film
as
well
as
the
gas
flow.
It
is
expected
that
this
will
also
depend
on
the
physical
properties
of
the
gas
and
the
injected
liquid.
In
this
study
only
air
and
water
are
used
and
so
only
the
effects
of
the
gas
and
liquid
flow
rates
have
been
studied.
The
photographic
evidence
given
in
Fig.
9
supports
the
flow
conditions
expressed
in
Fig.
8.
Fig.
9a
marked
as
A
in
Fig.
8
and
shows
that
air
flow
(u
9
=
2.32
m/s)
is
not
sufficient
to
strip
liquid
off
the
film
and
all
the
film
liquid
flows
down
the
plate
uniformly.
The
next
case
at
the
same
liquid
flow
rate
and
a
higher
gas
velocity,
4.53
m/s,
is
marked
B
in
Fig.
8.
Observation
shows
that
the
film
is
bent
towards
the
next
plate.
However,
the
film
does
not
break
up
into
droplets
at
these
conditions.
When
the
air
rate
is
increased
further
to
L
iq
u
id
Rey
no
lds
num
ber
2600
2400
2200
2000
1800
1600
1400
1200
1000
0
B.
J.
Azzopardi,
K
S.
Sanaullahl
Chemical
Engineering
Science
57
(2002)
3557-3563
3561
t
i
IJu
1
11.
(a)
Ewalt
al
I
II
1
1
11
1
4
gyy
'3
671
I
Fig.
9.
Photograph
showing
the
liquid
film
(a)
below
and
(b)
above
the
critical
gas
velocity.
5.6
m/s
(C
in
Fig.
8),
the
liquid
film
could
not
hold
together
and
broke
into
larger
droplets
Fig.
9b.
This
is
then
the
critical
velocity
above
which
forces
interact
in
a
way
that
droplets
detach
from
the
plates
and
are
carried
downstream
by
the
fast
moving
gas.
Beyond
this
point,
the
re-entrainment
increases
rapidly
with
increasing
gas
flow
rate.
The
events
recorded
in
Fig.
9b
are
illustrated
schemat-
ically
in
Fig.
10.
The
behaviour
presented
in
this
fig-
ure
is
based
on
visual
and
video
observations.
The
film
cross-section
changes
from
a
flat
layer
with
a
thickness
around
a
millimetre
to
a
relatively
thicker
rivulet.
The
film
begins
to
bend
in
the
direction
of
the
gas
flow.
Here
the
waves,
which
are
caused
by
the
instabilities
on
the
film
interface
and
are
driven
by
the
airflow
under
the
influence
of
aerodynamic
forces
such
those
that
come
into
effect
as
the
air
passes
round
the
sharp
corner.
Drop
diameters
are
of
the
order
of
the
initial
film
thickness.
These
drops
redeposit
onto
downstream
bends
of
the
wave-plate,
adhere
on
the
surface,
coalesce
and
form
a
rivulet,
which
drains
under
gravity.
Fig.
10.
A
schematic illustration
of
the
disruption
of
the
liquid
film
on
a
wave-plate.
4.
Modelling
of
re-entrainment
The
adhesion
or
detachment
of
a
liquid
film
travelling
around
corners
has
been
studied
by
Owen
and
Ryley
(1985)
who
considered
both
gravity-
and
shear-driven
films
and
cases
in
which
gravity
assisted
or
opposed
film
retention.
For
the
case
of
flow
in
a
geometry
such
as
a
wave-plate
where
the
liquid
is
acted
on
by
two
forces,
a
body
force
acting
vertically
downwards
and
a
horizontal
shear
acting
on
the
interface,
the
problem
appears
to
be
more
difficult
to
handle.
However,
as
shown
above
the
effect
of
the
horizon-
tal
shear
does
not
appear
to
affect
the
film
thickness
so
let
us
assume
that
the
two
effects
are
independent
and
that
we
can
consider
a
horizontal
interfacial
shear
acting
on
a
film
translating
downwards.
The
geometry
for
the
present
prob-
lem
is
shown
in
Fig.
11.
If
an
element
of
unit
depth,
sub-
tending
an
angle
AA,
at
a
radius
r
within
the
area
ABCD
is
considered.
Applying
Newton's
second
law
(f
df
)(r
dr)Afi
frAfi
pi
r
Aficl.r—
u2
(2)
r
As
in
the
direction
of
interest
the
film
is
shear
driven,
after
manipulation
and
equating
to
the
surface
tension
force
which
act
to
retain
the
film,
a
force
balance
is
obtained.
The
worst
case
for
this
is
at
the
wall
for
which
F
=
4piU
2(
Ri
R2)
+
()-
MI
R2
(3)
Bend
radius
1
(mm)
3
.........
1
cS
..
.........
.............
6
cS
f+df
R2
Fig.
11.
Geometry
for
film
separation
analysis.
100
200
300
400
500
Liquid
Reynolds
number
(-)
r
it
ica
l
g
as
ve
lo
city
(m
/s
)
60
50
40
30
20
10
,1)
,,,
•••••••••Ar.1
"a
ir
I
I
Present
work
Azzopardi
Verlaan
—4--
Houghton
Monat
et
al.
Model
3562
B.
J.
Azzopardi,
K
S.
Sanaullahl
Chemical
Engineering
Science
57
(2002)
3557-3563
200
100
50
20
10
5
2
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
Film
thickness
(mm)
Fig.
13.
Comparison
of
experimental
and
theoretical
transition
to
re-entrainment
for
air/water
systems.
We
ber
num
ber
(-
)
Fig.
12.
Predicted
critical
gas
velocities
for
re-entrainment.
Values
of
F
below
zero
indicate
separation
of
the
film
from
the
wall.
This
compressive
force
becomes
negative
when
4p/US
(4)
3R1
Ri
+
6'
where
5
=
R2
R1.
However,
the
mean
transverse
velocity
can
be
written
in
terms
of
the
gas
velocity
as
Ti
6
CfPg14
2
g5
U
=
=
(
5
)
2r/
i
4r/
i
Substituting
into
Eq.
(4)
yields
3QR
1
v
i
u
=
c
f
p
g
5
2
(12
1
+
5)
or
in
terms
of
a
Weber
number
3R
1
vis
We=
(
7
)
5
2
(R
i
+
5)e
f
The
Weber
number
was
defined
above
in
Eq.
(1).
Examples
of
the
dependence
of
the
critical
gas
velocity
on
liquid
flow
rate
is
given
in
Fig.
12.
5.
Discussion
Studies
of
the
falling
film
on
a
flat
surface
with
a
hori-
zontal
gas
flow
have
indicated
that
the
flow
can
be
treated
as
uncoupled,
i.e.,
that
the
thickness
is
determined
from
the
Nusselt
equation
for
gravity
flow.
Moreover,
given
the
characteristic
spacing
of
the
striations
reported
in
Fig.
4,
wavelength
=
20
mm
and
the
width
of
the
individual
plates
in
the
wave-plate
unit,
—23
mm,
it
is
no
surprise
that
striations
are
not
seen
in
the
latter
case.
If
they
occur,
they
would
probably
sit
on
the
downstream
corner
of
that
plate.
In
which
event,
a
striation
would
be
masked
by
any
edge
effects.
However,
if
the
average
film
thickness
in
the
wave-plate
is
take
from
Nusselt
theory,
then
the
thickness
at
the
corner
might
be
expected
to
be
more
like
the
film
thick-
ness
at
the
striation
crest,
i.e.,
twice
the
Nusselt
thickness.
The
locus
of
conditions
at
which
detachment
of
the
film
is
expected
from
the
predictions
of
the
model
described
in
Section
4
has
been
compared
with
that
determined
experi-
mentally.
Fig.
13
shows
that
there
is
good
agreement.
Also
shown
is
the
transition
line
from
experiments
reported
by
Azzopardi
(1977).
However,
this
was
for
a
single
flat
plate
without
bends.
In
this
case,
the
re-entrainment
occurred
from
the
leading
edge
of
the
film
and
not
surprisingly,
the
detachment
occurs
at
higher
gas
velocities.
Also
shown
in
Fig.
13
are
data
taken
from
publications
reporting
tests
with
vertical
up
flow
of
the
drop-laden
gas,
i.e.,
where
the
crests
of
wave-plates
were
horizontal
(Houghton
&
Radford,
1939;
Monat
et
al.,
1986;
Verlaan,
1991).
In
these
cases
the
films
were
much
thinner.
However,
the
Weber
numbers
for
the
in-
ception
of
re-entrainment
were
similar
to
those
determined
in
the
present
work
albeit
with
much
less
dependence
on
film
thickness.
It
is
noted
that
the
mechanism
for
re-entrainment
for
those
cases
could
be
different
and
related
to
the
(6)
water
soap
solution
glycerol
(0.011
N/ms)
glycerol
(0.019
N/ms)
*
d
i
ME
*
*
*
iii
. le
30
25
20
15
10
5
We
ber
nu
m
ber
(-
)
0.6
0.5
0
0
0.1
0.2
0.3
0.4
40
30
......,
at
..sz,
20
o
s"
..o
;
10
0
Surface
tension
(N/m)
0.0695
0.032
0.0695
water
soap
solution
soap
solution
B.
J.
Azzopardi,
K
S.
Sanaullah
I
Chemical
Engineering
Science
57
(2002)
3557-3563
3563
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
Film
thickness
(mm)
Film
thickness
(mm)
Fig.
14.
Effect
of
value
of
surface
tension
attributed
to
the
soap
solution.
Fig.
15.
Effect
of
liquid
physical
properties
on
critical
Weber
number
for
a
vertical
wave-plate-data
of
Verlaan
(1991).
occurrence
of
flooding,
the
down
flow
of
liquid
being
obstructed
by
the
up
flow
of
gas.
The
effect
of
physical
properties
on
the
re-entrainment
limit
can
be
examined
using
the
data
of
Verlaan
(1991).
He
employed
four
different
liquids,
water,
an
aqueous
soap
solu-
tion
(surface
tension
=
0.032
N/m),
and
two
water/glycerol
mixtures
(viscosity=0.011
and
0.019
Pa
s,
surface
tension=
0.041
and
0.055
N/m).
Addison
(1945)
has
shown
that
the
actual
surface
tension
of
a
solution
at
new
surface,
such
as
that
created
at
re-entrainment
depends,
on
the
concentration
of
the
solute
at
the
surface.
This
value
will
be
nearer
to
that
of
the
solvent
(0.0695
N/m)
than
the
equilibrium
value
of
the
solution
(0.032
N/m)
because the
concentration
at
the
new
surface
will
be
lower.
Fig.
14
shows
that
those
two
values
make
the
Weber
numbers
lie
on
either
side
of
the
water
data.
Fig.
15
shows
that
the
results
for
all
the
liq-
uids
lie
at
a
constant
Weber
number,
which
is
defined
in
Eq.
(7).
Here
the
surface
tension
for
the
soap
solution
was
taken
to
be
0.051
N/m.
The
data
shows
a
small
trend
with
film
thickness.
Verlaan
(1991)
also
examined
the
effect
of
gas
density.
This
data also
showed
that
the
critical
condi-
tions
for
re-entrainment
are
well
represented
by
a
constant
Weber
number.
6.
Conclusions
A
mechanism
for
the
re-entrainment
of
liquid
deposited
on
the
walls
of
wave-plate
mist
eliminators
operating
in
the
horizontal
mode
has
been
identified
and
quantified.
The
disruption
of
the
film
was
seen
to
occur
at
the
downstream
corner.
Acknowledgements
The
work
was
carried
out
under
EPSRC
Research
Grant
GR/K52911.
References
Addison,
C.
C.
(1945).
The
properties
of
freshly
formed
surfaces:
Part
IV-the
influence
of
chain
length
and
structure
on
the
static
and
dynamic
surface tensions
of
aqueous
alcoholic
solutions.
Journal
of
the
Chemical
Society,
98-106.
Azzopardi,
B.
J.
(1977).
Interaction
between
a
falling
liquid
film
and
a
gas
stream.
Ph.D.
thesis,
University
of
Exeter.
Azzopardi,
B.
J.,
James,
P.
W.,
Wang,
Y.,
Hughes,
J.,
Layton,
J.
D.,
Walters,
J.
K.,
&
Sanaullah,
K.
S.
(2002).
Performance
of
wave-plate
eliminators,
in
preparation.
Burkholz,
A.
(1989).
Droplet
separation.
New
York:
VCH
Publishers.
Houghton,
H.
G.,
&
Radford,
W.
H.
(1939).
Measurements
on
eliminators
and
the
development
of
a
new
type
for
use
at
high
gas
velocities.
Transactions
of
the
American
Chemical
Engineers,
35,
427-433.
Monat,
J.
P.,
McNulty,
K.
J.,
Michelson,
I.
S.,
&
Hansen,
0.
V.
(1986).
Accurate
evaluation
of
Chevron
mist
eliminator.
Chemical
Engineering
Progress,
12,
32-39.
Nusselt,
W.
(1916).
Die
Oberflachenkondensation
des
Wasserdampfes.
Zeitschrift
des
Verines
Deutschen
Ingenieur,
60,
541-546;
569-575.
Owen,
I.,
&
Ryley,
D.
J.
(1985).
The
flow
of
thin
liquid
films
around
corners.
International
Journal
of
Multiphase
Flow,
11,
51-62.
Verlaan,
C.
C.
J.
(1991).
Performance
of
novel
mist
eliminators.
Ph.D.
thesis,
University
of
Technology,
Delft.
Wang,
W.,
&
Davies,
G.
A.
(1996).
CFD
studies
of
separation
of
mists
from
gases
using
vane-type
separators.
Transactions
of
the
Institute
of
Chemical
Engineers,
74A,
232-238.
Wang,
Y.,
&
James,
P.
W.
(1998).
The
calculation
of
wave-plate
demister
efficiencies
using
numerical
simulation
of
the
flow
field
and
droplet
motion.
Transactions
of
the
Institute
of
Chemical
Engineers,
76A,
980-985.
Wang,
Y.,
&
James,
P.
W.
(1999).
Assessment
of
an
Eddy-interaction
model and
its
refinement
using
predictions
of
droplet
deposition
in
a
wave
plate
demister.
Transactions
of
the
Institute
of
Chemical
Engineers,
77A,
692-698.