Professional Documents
Culture Documents
D e p a r t m e n t o f C h e m i c a l E n g i n e e r i n g - U n i v e r s i t y C o l l e g e L o n d o n
Table
of
Contents
1.
Introduction.................................................................................................................... 3
1.1.
Process
intensification ................................................................................................. 3
1.2.
Microchemical
systems ................................................................................................ 3
1.3.
Applications ..................................................................................................................... 4
1.4.
Microdistillation ............................................................................................................. 5
1.5.
Fabrication
and
ancillaries ......................................................................................... 7
1.6.
Scope
of
research ........................................................................................................... 7
2.
Literature
review....................................................................................................... 9
2.1.
Zero-gravity
distillation
utilising
the
heat
pipe
principle................................ 9
2.2.
Modelling
Zero-Gravity
Distillation .......................................................................10
2.3.
Microdistillation
of
multicomponent
systems ...................................................10
2.4.
Micro-Scale
Distillation
I:
Simulation.................................................................11
3.
Model
development ............................................................................................... 13
3.1.
Assumptions ..................................................................................................................13
3.2.
Control
volume..............................................................................................................14
3.3.
Molar
balance ................................................................................................................15
3.4.
Vapour
and
Liquid
flow..............................................................................................18
3.5.
Feed...................................................................................................................................19
3.6.
Physical
Properties ........................................................................................................19
4
Case
study:
Seok
and
Hwang
(1985) ................................................................. 24
4.1
Equations .........................................................................................................................24
4.2
Results
and
discussion ................................................................................................26
4.3
Discussion........................................................................................................................29
4
Case
studies ............................................................................................................... 31
4.1
Equations .........................................................................................................................31
4.2
Assigned
Variables
and
Parameters ..................... Error!
Bookmark
not
defined.
4.3
Results
and
discussion ................................................................................................32
1.
Introduction
1.1. Process
intensification
Throughout
the
history
of
chemical
process
engineering,
since
the
discovery
of
distillation
as
a
method
of
obtaining
pure
water
almost
2000
years
ago,
engineers
have
been
continuously
improving
these
methods
to
reduce
size,
cost,
energy
consumption;
and
increase
safety,
yield
and
control.
However,
the
last
few
decades
have
seen
vast
developments
in
this
area,
and
this
has
come
to
be
known
as
process
intensification,
which
may
be
defined
as
a
strategy
for
making
dramatic
improvements
in
chemical
processes
with
regards
to
the
criteria
mentioned
above.
This
area
of
study
can
be
divided
into
two
parts,
Equipment
Processes
are
intensified
by
the
design
of
novel
reactors
such
as
monolithic
reactors
and
microreactors;
and
other
unit
operation
equipment
such
as
compact
heat
exchangers,
static
mixers
and
rotating
(high
gravity)
separators.
Methods
Intensification
is
achieved
by
the
use
of
new
techniques
such
as
multifunctional
reactors
(integrated
with
separation,
heat
exchange
or
phase
transition),
hybrid
separation
(such
as
membrane
distillation)
and
the
use
of
alternative
energy
sources
(solar,
microwaves,
ultrasound).
Other
methods
include
new
process
control
methods
such
as
intentional
unsteadystate
operation
and
the
utilisation
of
supercritical
fluids.
Safety:
Highly
exothermic
reactions
can
be
performed
safely
due
to
better
heat
transfer,
enabled
by
high
surface-to-volume
ratios.
Also,
if
there
is
loss
of
containment
of
hazardous
materials,
adverse
impact
will
be
minimal.
Cost:
Due
to
the
small
amount
of
volumes
involved,
much
lower
energy
demand
and
lower
fabrication
cost.
Increased
opportunities
for
automated
processes
also
reduce
the
need
for
manpower.
Automation:
Enabled
by
recent
advances
in
digital
microfluidics.
Provides
more
possibilities
for
automated
processing
of
arrays
of
materials
such
as
DNA
analysis,
catalyst
screening
(Fair
,2007)
High
throughput:
More
opportunities
for
automation
enables
analysis
of
a
wider
spectrum
of
materials
and
faster.
Other
key
developments
include
methods
for
continuous
separation
of
particles
based
on
their
size
using
dialectropheses,
electrowetting
on
dialectric
microfluidics
and
electrochemical
and
photochemical
activation
(Wiles,
Kralj
2006).
1.3. Applications
Microfluid
systems
have
many
applications
in
different
areas
of
technology
including
biology,
medical
science
and
chemistry.
Due
to
the
advantages
explained
above,
there
is
a
lot
to
be
gained
from
the
scale
down
of
laboratory
techniques
to
the
smallest
size.
It
can
be
used
to
cheaply
and
accurately
measure
parameters
such
as
transport
coefficients,
reaction
kinetics,
physical
parameters
and
catalyst
performance
(Yager,
2001).
It
is
easier
to
predicting
heat
and
mass
transfer
characteristics
due
to
mostly
laminar
flow,
which
enables
better
measurements.
It
also
allows
the
running
reactions
under
more
stringent
conditions
due
to
flatter
temperature
and
composition
profiles,
minimising
side
reactions.
It
is
useful
in
the
investigation
of
problems
with
recycle
streams.
This
is
due
to
the
small
cumulative
hold-ups
on
these
micro
plants,
in
the
order
of
1
day
(Sundberg,
2009).
The
small
volumes
involved
and
superior
heat
control
enables
1.4. Microdistillation
Separation
methods
are
integral
to
the
development
of
microchemical
systems.
These
include
liquid-liquid
extraction,
particle
sorting,
phase
separation,
absorption
and
distillation.
It
is
important
to
understand
how
these
scaled
down
units
behave,
particularly
due
to
the
increased
significance
of
interfacial
and
surface
phenomena.
Distillation
is
one
of
the
oldest
and
most
understood
unit
operations
in
Chemical
Engineering.
On
a
production
scale,
continuous
distillation
is
carried
out
in
columns,
either
filled
with
packing
or
plates,
with
the
distillate
coming
out
of
the
top
of
the
column
and
the
bottoms
coming
out
the
bottom.
Part
of
the
vapour
is
condensed
and
fed
back
to
the
column
as
reflux
while
part
of
the
liquid
at
the
bottom
is
vaporised.
This
creates
a
continuous
counter-current
stream
of
vapour
flowing
up
and
liquid
flowing
down,
enabling
mass
transfer
across
the
column.
In
this
case,
we
rely
on
gravity
for
liquid
to
flow
down
the
column
while
the
vapour
rises
to
the
top
naturally
due
to
the
pressure
in
the
reboiler.
The
idea
of
a
distillation
method
that
relies
on
these
capillary
forces
instead
of
gravity
(hence
the
name
zero-gravity
distillation)
has
been
around
since
the
80s
when
Seok
and
Hwang
(1985)
developed
the
method
by
utilising
the
heat-pipe
principle.
A
heat-pipe
is
a
heat
transfer
device
vapour
is
allowed
to
condense
at
the
cool
end,
and
the
condensate
flows
towards
the
hot
end
where
it
is
evaporated.
Fluid
flows
via
pressure
gradients
created
by
the
heating
and
cooling
of
opposite
ends.
At
the
heating
end,
analogous
to
the
column
reboiler,
vapour
is
generated,
increasing
pressure
and
forcing
vapour
down
to
the
cooling
end,
where
the
vapour
is
condensed
creating
the
same
effect
in
the
liquid
phase.
This
creates
the
same
counter
current
flow
of
vapour
and
liquid
seen
in
conventional
distillation
columns.
By
the
same
principle,
using
a
binary
mixture
would
result
in
the
lighter
component
accumulating
at
the
cool
end
and
the
heavier
component
at
the
hot
end,
as
confirmed
by
Cotter
(1965).
By
continuously
feeding
the
heat
pipe
with
a
mixture
and
withdrawing
products
from
both
ends,
a
continuous
distillation
unit
is
created.
Aside from the utilising capillary forces instead
of
gravity,
another
major
difference
with
microfluidic
distillation
is
that
reflux
occurs
internally
and
is
not
controlled
directly.
Hence
the
actual
reflux
rate
is
difficult
to
quantify
which
poses
a
challenge
to
understanding
the
influence
of
geometry
and
flow
rates
on
separation
performance.
In
microdistillation
devices,
uncontrolled
boiling
tends
to
occur
in
the
hot
as
the
liquid
tends
to
superheat.
This
is
due
to
the
increased
significance
of
surface
tension
and
low
nucleation
rate
(Hartman
and
Sahoo
2009).
To
counter
this,
some
researchers
have
used
a
carrier
gas
such
as
nitrogen
to
be
able
to
vaporise
the
liquid
below
boiling
point
(Wootton).
However,
this
results
in
a
poorer
separation
because
there
no
reflux
flow
due
to
the
unidirectional
flow
of
the
carrier
gas.
A
way
of
solving
this
problem
is
by
providing
a
stable
vapour
liquid
interface,
which
can
be
done
using
membranes
or
using
microstructures
to
exploit
surface
tension
effects
(Adiche
and
Sundmacher
(2010)).
2. Literature
review
2.1. Zero-gravity
distillation
utilising
the
heat
pipe
principle
D.
R.
Seok
and
Sun-Tak
Hwang
(1985)
The
concept
of
a
distillation
process
utilising,
capillary
forces
was
first
proven
by
Seok
and
Hwang
(1985),
who
conducted
experiments
on
a
54
cm
long,
1.05
cm
internal
diameter
pipe
with
a
fibreglass
wick
using
methanol-water
and
ethanol-
water
mixtures.
Along
with
experimental
results,
also
presented
is
a
model,
which
they
developed.
Theory
Mass
transfer
occurs
due
to
heat
transfer
and
diffusion;
and
is
controlled
by
the
vapour
phase,
with
a
constant
mass
transfer
coefficient
along
the
column,
independent
of
vapour
and
liquid
flow
rates.
The
model
assumes
no
concentration
and
temperature
gradients
in
the
radial
direction
and
that
vapour
and
liquid
are
in
constant
equilibrium.
The
overall
height
of
a
transfer
unit
(HTU)
is
used
to
evaluate
column
performance.
Liquid
flow
is
taken
to
be
at
its
capillary
limit,
which
means
the
axial
pressure
gradient
is
only
due
to
the
surface
tension
gradient,
and
liquid
flows
according
to
Darcys
law.
Experiments
were
conducted
for
both
mixtures
first
at
total
reflux
and
with
products,
varying
feed
rates
to
study
the
effects
of
varying
gas
to
liquid
ratios.
For
the
model
boundary
conditions,
it
is
assumed
that
vapour
and
liquid
compositions
are
equal
at
the
evaporator
and
condenser
ends,
which
is
not
true
in
reality
as
the
vapour
should
be
richer
in
the
lighter
component.
One
issue
with
the
model
was
evaluating
gas
to
liquid
ratios,
which
was
estimated
by
trial
and
error
since
it
could
not
be
measured
directly.
The
experimental
results
follow
the
modelled
column
profiles
quite
closely,
though
the
actual
gradients
were
steeper
at
the
middle
of
the
column.
Though
this
cannot
be
called
microdistillation
as
it
was
the
flow
channels
were
in
the
order
of
1
cm,
the
transport
equations
is
expected
to
be
the
same
when
brought
down
to
the
microscale.
10
11
first
assumed
mesh
flow,
with
a
constant
cross
sectional
area
perpendicular
to
direction
of
flow,
equal
to
the
mesh
cross
sectional
area.
The
second
case
assumed
a
falling
film
configuration,
where
the
area
of
flow
and
velocity
are
dictated
by
the
flow
rate,
with
fully
laminar
flow
conditions.
Simulation
results
show
that
the
first
case
is
a
more
accurate
representation,
showing
a
20%
average
deviation
with
experimental
results,
though
the
model
is
only
verified
based
on
HETP
calculations
as
opposed
to
concentration
profiles.
12
3. Model development
3.1. Assumptions
1. Vapour
liquid
interface
is
at
the
inner
surface
of
the
wick
area
of
flow
of
liquid
is
constant,
liquid
velocity
controls
the
flow
rate.
Simulations
by
Fanelli
(2007)
confirmed
that
this
assumption
leads
to
a
more
accurate
prediction
of
HETP
compared
to
area
of
flow
being
dictated
by
flow
rate.
However,
Sundberg
(2008)
reported
that
while
the
liquid
level
in
their
microdistillation
column
varied
with
flow
rate,
the
surface
of
the
wick
was
constantly
wet.
2. No
angular
concentration
or
temperature
gradients
angular
velocity
is
assumed
to
be
zero
and
this
vastly
reduces
the
complexity
of
the
model.
Inlet,
outlet
and
sample
ports
are
in
the
same
angular
position,
reducing
the
significance
of
angular
gradients.
3. Vapour
liquid
equilibrium
vapour
is
always
in
equilibrium
with
the
liquid
phase.
A
common
assumption
in
distillation
problems.
This
may
not
be
entirely
true
at
the
evaporator
end
where
the
liquid
is
superheated
as
reported
by
Ranjan
(2011)
in
his
studies
on
heat
pipes.
4. Heat
balance
is
neglected
Simulations
by
Ramirez-Gonzalez
(1992)
were
conducted
under
this
assumption
with
good
agreement
with
experimental
data.
Hence
it
is
deemed
unnecessary
to
add
further
complexity
to
the
model.
5. Equimolar
counter
diffusion
-
Another
common
assumption
for
distillation
problems.
This
is
probably
not
valid
at
the
condenser
and
evaporator
ends
where
extra
condensation
and
evaporation
occur.
6. Liquid
flows
at
its
capillary
limit
This
means
liquid
flows
at
the
maximum
allowed
by
capillary
forces.
7. Perfect
wetting
this
is
common
assumption
which
is
valid
for
porous
materials.
8. No
velocity
fluctuations
The
model
is
meant
for
steady
state
operation,
hence
velocity
fluctuations
are
assumed
to
be
negligible.
13
9. Constant
pressure
(1
atm)
along
z
Although
this
is
not
true
since
vapour
flow
is
enabled
by
pressure
gradients
along
the
pipe,
this
gradient
is
not
very
large
and
its
effect
on
physical
properties
is
assumed
to
be
negligible.
10. Temperature
is
a
function
of
composition,
constant
along
the
radial
and
angular
direction
This
is
a
simplification
due
to
the
fact
that
heat
balances
are
neglected
in
the
model.
In
reality
composition
(through
relative
volatilities)
are
a
function
of
temperature,
which
is
calculated
by
heat
balances.
Figure
1:
Schematic
of
vapour
and
liquid
phases
in
the
microdistillation
device
The
device
is
a
copper
cylinder
with
a
layer
of
wick
to
guide
liquid
flow
on
the
walls
while
vapour
flows
in
the
middle,
in
the
opposite
direction
of
liquid
flow.
In
the
axial
direction,
mass
transfer
is
mostly
by
convection
while
in
the
radial
direction,
we
have
equimolar
diffusion
across
the
interface
and
feed
entering
the
liquid
phase
from
the
outer
radius.
A
3D
diagram
of
our
control
volume
used
to
derive
the
molar
and
energy
balances
in
the
liquid
phase
is
shown
below
14
Figure
2:
Control
volume
diagram
In
the
z
direction:
Net
rate
of
addition
=
n A,z
( r + r) 2 r 2
2
n A,z
z +z
( r + r) 2 r 2
2
n A,z = CL DAB
x A
+ CL uz
z
In the r direction:
15
Due
to
assumption
4,
we
take
r
=
r1,
the
inner
radius,
and
r
+
r
=
r2,
the
outer
radius.
At
r1,
we
have
mass
transfer
across
the
interface,
which
is
given
by:
n A,r r r = kL r1 (CA CAi )
n A,r r r +r = n A,F r2
Where
nA,F
=
0
at
all
axial
positions
except
at
the
feed
point.
Calculation
of
this
value
is
explained
later
below.
It
should
be
noted
that
by
this
equation,
it
assumes
that
the
feed
enters
the
bulk
liquid
phase
with
a
radial
velocity
of
0,
hence
travels
along
with
the
liquid
its
axial
velocity.
This
is
valid,
because
simulations
show
that
the
feed
flow
rate
is
small
compared
to
the
liquid
flow
rate.
Furthermore,
it
is
not
possible
to
model
exactly
flow
of
feed
as
momentum
balances
are
not
considered
in
this
model.
In
the
direction:
Net
rate
of
addition
=
n A, rz + n A, rz +
Rate
of
generation
is
equal
to
zero
since
there
is
no
chemical
reaction.
The
rate
of
accumulation
within
the
c.v.
is
then
given
by:
2
CA ( r + r) r 2
z
t
2
Combining
the
above
equations
yield
r2 2 r12
r2 2 r12
CA r2 2 r12
dCA
dCA
z = DAB
+ v zCA
DAB
+ v zCA
2
t
2
dz
dz
z +z
z
+ kL r1 (CAi CA ) + n A,F r2 z
16
Since
CA
is
equal
to
xACL,
and
assuming
the
change
in
CL
over
length
z
is
small,
we
can
replace
CA
with
xA.
DAB
is
replaced
with
Deff,
given
by
equation
(x)
to
account
for
porosity.
As
z
approaches
zero,
dividing
by
whole
equation
by
the
control
volume
results
in
the
following
differential
equation
2
rn
x A
2 x A Deff x A
x
= Deff
v L A kL r1 ( x A x Ai ) + 2 A,Feed 2
=0
2 +
t
z
z z
z
CL r2 r12
(1)
Using
similar
methods
for
the
gas
phase
yields
the
following
equation,
where
the
feed
flux
term
is
dropped
because
feed
is
only
in
the
liquid
phase
and
the
partial
derivative
of
diffusivity
is
dropped
because
gas
diffusivity
is
more
or
less
dependant
of
composition
[6]:
y A
2 yA
y A
= DAB,G
+ 2r1kCG ( y Ai y A ) = 0
2 + vG
t
z
z
(2)
The
model
equations
(1)
and
(2)
are
related
by
mass
transfer
through
the
interface
by
equimolar
counter
diffusion.
kL CL (x A x Ai ) = kG CG (y Ai y A )
(3)
Boundary conditions
The
model
uses
Danckwerts
boundary
conditions
for
at
the
evaporator
and
condenser
ends,
given
by
the
following
equations
x A
z
(4)
(5)
(6)
(7)
=0
z =length
y
vG y A,in z =length + = vG y A ,in DAB,G A
z z =length
x
v L x A,in z =0 = v L x A,in Deff A
z z =0 +
y A
z
=0
z =0
17
PL = PG
2L cos
,0 180
rp
(8)
Assuming
constant
pressure
in
the
gas
phase
and
a
constant
pressure
gradient
in
the
liquid
phase,
this
equation
can
be
used
to
find
the
pressure
drop
in
the
following
equation.
Liquid
velocity
is
described
by
the
Darcy
equation
for
flow
through
porous
media
(Dunn
and
Reay,1994).
vL =
K PL
L z
(9)
The contact angle; , can be taken as 0 under assumption 9, ie (cos 0 = 1). At the
vL =
2K (c e )
L rp Length
(10)
Where
e
and
c
are
surface
tension
taken
at
the
evaporator
and
condenser
respectively
(Dunn
and
Reay,1994).
By
assumption
1,
liquid
molar
flow
rate
is
given
by:
(11)
Since
reflux
is
not
controlled,
actual
flow
rates
are
difficult
to
quantify.
In
Seok
and
Hwang
(1985),
they
managed
to
estimate
the
gas
to
liquid
ratio
by
trial
and
error,
which
we
can
use
to
calculate
the
vapour
flow
rate.
We
can
then
calculate
the
vapour
phase
velocity
using
equation
(12).
18
( )
G = GL
vG =
(12)
G
r12CG
(13)
ratio
3.5. Feed
The
feed
enters
through
a
hole
perforated
in
the
device.
Since
our
model
only
considers
the
axial
dimension,
we
assume
that
it
enters
the
liquid
phase
through
the
outer
surface
of
the
liquid
phase
from
all
directions
of
,
for
a
length
dzF.
In
mathematical
terms,
the
area
of
flux
for
the
feed
is
given
by:
A feed = 2r2 dz F
(14)
Where
the
Afeed
is
equal
to
the
area
of
the
feed
inlet
in
the
experiment
and
dzF,
is
the
size
of
the
element
used
in
solving
the
differential
equation.
The
feed
molar
flux
is
then
obtained
from
equation
(14)
for
length
dzF
at
the
feed
location,
and
equals
to
0
at
all
other
points.
n A, feed =
3.6.
F
at z = z feed : z feed + dzF
A
(15)
Physical Properties
L = ( x i M i )
CL
1000
and
G = ( y i M i )
CG
1000
(16)
19
CL =
1
x
C i
L,i
CL,i =
B1
B
T 4
B21+1 B 3
(17)
(18)
Gas
concentration
is
obtained
from
the
ideal
gas
law,
which
is
valid
at
low
pressures
CG =
n
P
=
V RT
(19)
Liquid
viscosities
for
pure
components
are
known
to
vary
with
temperature
by
equation
(20).
Constants
for
the
equation
are
found
by
regression
of
experimental
data
or
data
obtained
from
ChemCAD.
(20)
(21)
J AB (T) = 1 (1 J AB,0 )
(573 T )
275
(22)
Gas
viscosities
for
pure
substances
are
obtained
from
the
Chapman-Enskog
equation,
where
/K
are
found
expressed
by
equation
(24)
with
constant
20
M wT
2 ,
(23)
(24)
= 0.841Vc1/ 3
(27)
G, pure = 2.6693 10 5
/K = 0.77TC
Viscosity of the gas mixture is then obtained from equation (28), where ij is
A,G
B,G
+
yB
yA
1+ AB 1+ BA
yA
yB
(28)
1/ 2
1/ 2
1/ 4
1 M i i M j
1+
ij =
1+
8 M j j M i
(29)
G,mix =
correlation:
x
B
A
0
0
DAB,L = DAB,L
DBA,L
Where
diffusivities
for
dilute
systems
are
given
by
the
following
equation
proposed
by
Wilke
and
Change
(1955).
1/ 2
0
AB,L
7.4 10 8 (M B ) T
=
0.6
BVA,boil
(30)
(31)
Deff =
DAB
2 DAB , where 1/
Gas
phase
diffusivity
is
obtained
from
equation
(32)
proposed
by
Wilke
and
Lee (1955) where AB and AB are given by equations (33) and (34)
21
DAB,G
[3.03 (0.98 M )] 10 T
=
1/ 2
AB
3/2
2
TM 1/AB2AB
D
(32)
AB = 0.5(A + B )
(33)
AB = A B
(34)
Mass
transfer
coefficients
are
obtained
from
the
following
correlations
for
mass
transfer
in
fixed
beds
(Treybal,
1981)
vG
kG =
v L
Re Sc 2 / 3
kL =
Sc 2 / 3
(35)
(36)
For laminar flow, = 1.09 and = 2/3, while for turbulent flow, = 20.4 and =
0.815.
The
Reynolds
number
for
flow
through
porous
media
and
the
Schmidt
number
are
shown
below.
Re =
L v L d p
L
Sc =
DAB
(37)
(38)
Surface tensions at evaporator and condenser ends are found from literature or
ChemCAD.
Since
our
model
is
meant
for
steady
state
simulations,
surface
tensions
at
these
ends
can
be
input
as
single
parameters
Vapour
Liquid
Equilibrium
Compositions
at
the
interface
follow
the
relationship
yi =
i yi
at
r
=
r1
i yi
22
(39)
Relative
volatility
for
component
i
is
given
by
the
following
equation
where
the
least
volatile
component
should
be
chosen
as
nc,
although
this
is
just
a
matter
of
convention.
Pivap i
i = vap
Pnc nc
(40)
Vapour
pressures
for
pure
components
are
obtained
from
correlations
found
in
E
Pivap = exp E1 + 2 + E 3 ln(T ) + E 3T E 4
(41)
By
the
modified
Raoults
law
and
the
assumption
of
bubble
point
at
all
positions,
the
f
ollowing
e
quation
m
ust
b
e
o
beyed.
(P
vap
i xi ) = P
(42)
The activity coefficients are a function of composition, which can be found from
i = f ( x1, x 2 ...x nc )
23
(43)
4.1 Equations
Mass
Balances
2
rn
x A
2 x A Deff x A
x
= Deff
uG A kL r1 ( x A x Ai ) + 2 A,Feed 2
= 0
2 +
t
z
z z
z
CL r2 r12
y A
2 yA
y A
= DAB,G
+ 2r1kG ( y Ai y A ) = 0
2 + uG
t
z
z
yi =
i yi
at
r
=
r1
i yi
i =
Pivap i
Pncvap nc
E
Pivap = exp E1 + 2 + E 3 ln(T ) + E 3T E 4
(P
vap
i x i ) = P i = f (x A )
Fluid
flow
8rp 3 cos (Tc ) (Te )
uL =
150(1 ) 2 L
Leff
u (r 2 r 2 )C
uG = G L L 2 2 1 L
rC
( )
n A, feed =
F
at z = z feed : z feed + dzF
2r2 dz F
24
uG
ScG 2 / 3
kL =
uL
Re Sc 2 / 3
-
alpha
=
20.4,
beta
=
0.815
Re =
L uL d p
L
Sc L =
L
L DAB,L
ScG =
G
G DAB,G
Physical properties
Deff = f (x A )
(573 T )
275
NC
= x i i
i=1
CG =
n
P
=
V RT
Ancillary
relationships
x A + x B = 1
y A + y B = 1
x Ai + x Bi = 1
y Ai + y Bi = 1
25
Assigned
Variables
DAB0
DBA0
C
From
eq
30
at
average
T
From
eq
30
at
average
T
ChemCAD
at
condenser
T
and
x
ChemCAD
at
condenser
T
and
x
From
eq
32
at
average
T
From
eq
19
at
average
T
and
x
E
DAB,G
CG,ave
F
xin
yin
L,ave
CL,ave
Device
Parameters
r1,
r2
Lt
Feed
location
101325
bar
From
literature
0.0007
m,
adjusted
parameter,
starting
point
estimated
from
Dunn
and
Reay
(1994)
At
z
=
0.12
m
x methanol (mol)
0.1
0.2
0.3
0.4
0.5
0.6
Length
(m)
Current
Model
Experimental Results
Total
reflux
experiments
for
methanol
water
were
used
as
a
base
case
for
estimating
pore
size,
dP,,
which
is
why
the
model
shows
an
almost
perfect
fit.
26
Ramirez-Gonzalez
(1992)
did
not
present
a
case
study
for
this
experiment
so
we
are
unable
to
compare
our
model
with
theirs.
For
the
boundary
conditions,
xin
was
taken
as
0.999
and
yin,
was
0.001.
The
velocity
of
the
liquid
is
0.16
m/s,
which
is
quite
large
but
since
the
data
was
taken
directly
from
Ramirez-Gonzalez
experiment,
this
was
assumed
to
be
true,
though
a
possible
error
may
have
been
made
in
assuming
that
the
area
of
flow
was
constant
and
equal
to
the
cross
sectional
area
of
the
wick.
x methanol (mol)
0.1
0.2
0.3
0.4
0.5
0.6
Length
(m)
Experimental
Results
Ramirez-Gonzalez
Current Model
x methanol (mol)
0.1
0.2
0.3
0.4
0.5
0.6
Length
(m)
Experimental
Results
Ramirez-Gonzalez
27
Current Model
Using
the
dp,
found
from
total
reflux
simulations,
the
cases
of
G/L
>
1
were
simulated.
The
feed
term
is
not
thought
to
make
a
large
difference
in
the
composition
profiles
as
the
liquid
flow
rate
is
much
larger
than
feed
flow
rates,
which
may
be
why
Ramirez-Gonzalez
neglected
this
term.
The
current
model
shows
a
better
fit
for
G/L
=
1.09
but
worse
for
G/L
=
1.21
compared
to
Ramirez-
Gonzalez
(1992).
It
should
be
noted
that
the
current
model
shows
a
smoother
curve
because
Ramirez-Gonzalez
only
used
8
elements
in
the
discretisation
of
length
while
the
current
model
uses
50.
The
boundary
conditions
for
the
vapour
phase
inlet,
yin,
were
significantly
larger
than
the
liquid
composition
at
the
evaporator.
This
will
be
discussed
at
the
end
of
the
chapter.
Model
results
for
ethanol-water
experiments
are
also
shown
below.
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
0
0.1
0.2
0.3
0.4
0.5
0.6
Length
(m)
Experimental
Ramirez-Gonzalez
28
Current Model
1
0.8
0.6
0.4
0.2
0
0
0.1
0.2
0.3
0.4
0.5
0.6
Length
(m)
Experimental
Current model
The
results
for
ethanol-water
experiments
do
not
show
as
good
a
fit
as
that
of
methanol-water.
One
reason
may
be
that
dp
was
estimated
using
the
methanol
water
experiments
as
a
lower
dP
would
result
in
a
sharper
curve
at
the
evaporator
end.
It
may
have
been
better
to
estimate
dP
using
both
ethanol
and
methanol
experiments
to
find
a
compromise
between
fitting
the
two
cases
at
total
reflux.
More
significantly,
ethanol-water
forms
an
azeotropic
mixture
and
is
highly
non-ideal
at
higher
concentrations
of
ethanol.
The
estimation
of
activity
coefficients
by
regressing
data
found
in
ChemCAD
did
not
show
as
good
a
fit
for
ethanol
compared
to
that
of
methanol.
4.3 Discussion
1. The
vapour
inlet
compositions
were
not
reported
in
either
of
the
papers
used
in
this
case
study
and
hence
had
to
be
estimated.
Ramirez-Gonzalez
assumed
x
=
y
at
the
condenser
and
evaporator
ends
but
this
does
not
make
much
sense
since
the
vapour
should
be
richer
in
the
lighter
component
(ethanol/methanol)
although
it
was
found
that
this
assumption
holds
for
total
reflux
methanol
water
experiments.
It
was
first
though
that
the
vapour
composition
would
correspond
to
the
liquid
composition
by
equilibrium
relations.
However,
it
was
found
that
a
larger
value
of
yin
had
to
be
used
for
cases
with
G/L
>1
to
fit
the
experimental
data,
and
this
value
was
larger
with
higher
G/L.
This
is
likely
to
be
due
to
the
fact
that
the
column
is
modeled
as
a
rectification
section,
neglecting
29
30
5.1 Equations
kL r1 ( x A x Ai )
x A
2 x A Deff x A
x A
= Deff
+
G
t
z 2
z z
z
r2 r1
y A
2 yA
y A kG ( y Ai y A )
= DAB,G
+
2 + uG
t
z
z
r1
yi =
i yi
at
r
=
r1
i yi
i =
Pivap i
Pncvap nc
E
Pivap = exp E1 + 2 + E 3 ln(T ) + E 3T E 4
i = f (x A )
Fluid
flow
u (r 2 r 2 )C
uG = G L L 2 2 1 L
rC
( )
Re =
L uL d p
L
Sc L =
L
L DAB,L
31
ScG =
kG =
uG
ScG 2 / 3
kL =
uL
Re Sc 2 / 3
G
G DAB,G
Physical properties
Deff = f (x A )
(573 T )
275
NC
= x i i
i=1
CG =
n
P
=
V RT
Ancillary
relationships
x A + x B = 1
y A + y B = 1
x Ai + x Bi = 1
y Ai + y Bi = 1
5.2 Results
and
discussion
32
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
Length
(m)
Current
Model
Fanelli
A
porosity
of
0.7
was
assumed
for
this
case,
which
is
typical
of
steel
mesh
used
in
heat
pipes.
Parameter
dP,
was
then
adjusted
to
fit
the
model
results
as
reported
by
Fanelli
and
a
value
of
0.045
m
was
found
to
be
the
best
value.
This
is
much
higher
than
the
usual
>0.001
m
found
in
literature.
The
model
uses
the
low
mass
transfer
coefficient
correlation
for
the
liquid
side,
consistent
with
the
recommendation
that
it
be
used
for
low
Re,
which
is
around
10
along
the
column.
33
0.14
0.12
0.1
0.08
0.06
0.04
0.02
0
-0.02
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
Length
(m)
Liquid
Side
-
Current
Model
nj =
x Ain, j
dx A , j
Length
and HETP j =
x A,ij x A, j
nj
HETPtotal = HETPvap +
mG
HETPliq
L
34
Case
1
Case
2
HETP
(cm)
Liquid
Vapour
Total
Liquid
Vapour
Total
Fanelli
Current
Model
1.27
0.03
1.3
1.95
0.03
1.98
1.2
0.5
1.7
1.7
0.9
2.4
Experimental
-
1.3
-
2.1
It
is
observed
that
the
current
model
overpredicts
the
proportion
of
HETP
in
the
liquid
phase
to
the
vapour
phase
compared
to
Fanellis
(2007)
model.
Fanellis
(2007)
also
concluded
that
since
the
flow
rates
had
more
of
an
effect
on
the
vapour
HETP,
mass
transfer
was
controlled
by
the
vapour
phase.
However,
the
current
model
shows
that
flow
rates
had
no
effect
on
the
vapour
HETP
and
mass
transfer
was
controlled
by
the
liquid
phase.
We
may
conclude
then
that
the
current
model
overpredicts
the
mass
transfer
capability
of
the
vapour
phase,
though
this
cannot
be
confirmed
since
the
current
model
actually
better
predicts
the
total
HETP.
35
6 Conclusion
The
model
was
verified
against
case
experiments
as
well
as
their
simulations
with
moderate
success
and
this
was
after
degree
of
parameter
estimation.
Hence,
the
model
is
not
suitable
for
simulations
without
experimental
work.one
of
the
main
issues
is
with
the
area
of
liquid
flow,
which
is
likely
to
be
the
source
of
errors.
The
question
of
whether
mass
transfer
coefficients
for
packed
beds
are
valid
for
predicting
microdistillation
devices
remain
unsolved.
A
more
accurate
study
could
be
made
given
more
experimental
data,
particularly
porosity
and
particle
diameter.
For
future
work
there
are
several
recommendations:
1) Implementation
of
heat
balances
to
enable
a
more
accurate
modelling
of
the
evaporator
and
condenser
ends.
2) A
proper
comparison
between
this
one
dimensional
model
with
mass
transfer
coefficients
and
the
two
dimensional
model
such
as
the
ones
used
by
Fanelli
(2007)
and
Tscherjaew
(1997)
3) Extension
of
the
model
to
multicomponent
systems
since
binary
systems
are
quite
rare
in
real
life
applications.
Work
by
Tscherjaew
would
be
a
good
starting
point
for
this
work.
36
7 Notation
Letters
Area (m2)
Concentration
(mol/m3)
Cp
Heat
capacity
(J/mol.K)
D
Mass
diffusivity
(m2/s)
dp
Pore
diameter
(m)
e
Molar
energy
(J/mol)
F
Feed
flow
rate
(mol/s)
G
Vapour
flow
rate
(mol/s)
hL,hG
Specific
enthalpy
(J/mol)
Hvap
Heat
of
vaporisation
hT
Heat
transfer
coefficient
(W/m2.K)
kc
Mass
transfer
coefficient
(m/s)
kT
Heat
conductivity
(W/m.K)
K
Permeability
(m2)
L
Liquid
flow
rate
(mol/s)
Mw
Molecular
weight
Density (kg/m3)
Molar
flux
(mol/m2.s)
Pressure
(Pa)
Surface
tension
(N/m)
Constant
rP
Re
Reynolds number
Relative
volatility
Subscripts
Sc
T
Schmidt
number
Temperature
(K)
A,
B
G,L
Species
A
or
B
Gas
or
liquid
Internal
energy
(J/mol)
Velocity
(m/s)
Feed
Species i
Liquid
mole
fraction
Vapour
mole
fraction
Axial
position
(m)
z,
r,
eff
Directions
v
x
y
z
Symbols
effective
Porosity (-)
Viscosity
(Pa.s)
Angle
(
)
Constant
8 References
37
38
39
40