Documente Academic
Documente Profesional
Documente Cultură
A = P
k
k
r
(5)
where k is ility, k
r|
is the relative the absolute permeab
velocity of phase |, and g is the acceleration due to
gravity. The fluid pressure in phase | (P
|
) is the sum
of a reference pressure P (assumed equal to the
pressure of the gas phase) and the capillary pressure,
i.e.,
| | C
P P P + = (6)
Multicomponent Diffusion
The Ficks law of molecular diffusion works well for
iffusion of tracer solutes that are present at low
mponent systems may
epend on all concentration variables, leading to non-
d
concentrations in a single-phase aqueous solution.
However, many subtleties and complications arise
when multiple components diffuse in a multiphase
flow system, as in AGMD [7].
Effective diffusivities in multico
d
linear behavior especially when some components are
present in significant (non-tracer) concentrations.
Additional nonlinear effects arise from the dependence
of tortuosity on phase saturations, and from coupling
between advective and diffusive transport. For gases
and vapors, the Fickian model has serious limitations
even at low concentrations, which prompted the
development of the dusty gas model [8,9], and
111
Downloaded 01 Nov 2012 to 182.255.2.5. Redistribution subject to AIP license or copyright; see http://proceedings.aip.org/about/rights_permissions
accounts for molecular streaming effects (Knudsen
diffusion) that become very important when the mean
free path of gas molecules is comparable to pore sizes
[8,9,10].
We have use
fl
d a pragmatic approach in which diffusive
ux of component (water or air) in phase (= liquid,
gas) is written as [7]
k
|
k
| |
k
|
X D f A = (7)
where
k
| |
k
|
t |t d D (8)
0
=
Diffusion coefficients ( )
k
|
d of
follows [11]
gases depend on
pressure and temperature as
( ) ( )
80 . 1
15 . 273
15 . 273
, ,
0
0 0
|
|
|
|
+
+
=
T
T
P
P
T P d T P d
k
|
k
|
0 . \
(9)
, for general two-phase conditions, th
flux for component k is written as
Equ ral
finite difference method [7,12]. Time is discretized
NUMERICAL MODEL
In this pap le AGMD
module, consisting of a feed channel, a condenser
te and at a specified temperature. Cooling water is
F
ngle
flow channels are initially
turated with water. Initial pressure in the entire flow
iscosity, and saturated vapor pressure) are calculated
=0 C), the diffusion coefficient for air-water vapor
RESULTS AND DISCUSSIONS
Figure 3 shows the temperature profile in the feed and
ndenser channels. For this simulation, we assumed a
Finally e diffusive
k k k k
g g g l l l
k
X D X D f A A = (10)
Spatial and Temporal Discretization
ations 2, 3, and 4 are discretized using the integ
using a fully implicit finite difference scheme. Such a
discretization approach yields a set of three coupled
nonlinear algebraic equations per volume element of
the overall flow system. These equations are solved
using a stabilized bi-conjugate gradient solver [7].
er, we consider only a sing
channel, a membrane and an air-gap, of the AGMD
design (see Figure 2). We assume that flow in both the
feed and condenser channels can be represented by
flow between two flat rectangular plates (see Figure
2). In other words, it is assumed that flow is one-
dimensional. Because the lengths of the feed and
condenser channels (1. 5 m) are significantly larger
than their widths (0.002 m), this is a reasonable
assumption. The hydrophobic membrane and the air
gap are 150 m and 0.001 m wide.
Seawater is input into the feed channel at a constant
ra
sent through the condenser channel countercurrent to
the flow in the feed channel. Simulations are
performed for both the scenarios where the flow rates
are same in the two channels or different. The
membrane and the air gap are initially saturated with
chematic representation of one-dimensional,
module of AGMD
igure 2. S
si
air, whereas the two
sa
system is assumed to be 0.5510
5
Pa, and initial
temperature is 30
o
C.
All water properties (density, specific enthalpy,
v
from the steam table equations as given by the
International Formulation Committee [13]. Air is
approximated as an ideal gas, and additivity is
assumed for air and vapor partial pressures in the gas
phase, Pg = Pa + Pv. The viscosity of air-vapor
mixtures is computed from a formulation given by
Hirschfelder et al. [14]. The solubility of air in liquid
water is represented by Henry's law.
At standard conditions (P
0
=1.0132510
5
Pa and
T
o
0
mixture is 2.3410
-5
m
2
s
-1
. Diffusion at any other
pressure or temperature is calculated using Equation 9.
Liquid- and gas-phase relative permeabilities are
assumed to be linear functions of saturation.
co
flow rate of 1.08 lit hr
-1
for both the feed and
condenser channels. As indicated earlier, the whole
system was initially at 30
o
C. Hot water was fed at
85
o
C at the left hand side of figure 3, and was flowing
left to right (in Figure 3). Cold water at 30
o
C was
flowing from right to left (see Figure 3). After 1000
seconds of operation (as shown in Figure 3), the
maximum temperature within the feed channel was
72
o
C and declining gradually from left to right along
the length of the channel. As expected, temperature
was rising from right to left in the condenser channel,
with the temperature being about 65
o
C at the exit of
the condenser channel. In other words, a temperature
difference was created at any location between the
Condenser
Mem brane
Air-gap Air-gap
x
y z
Feed
y z
x
112
Downloaded 01 Nov 2012 to 182.255.2.5. Redistribution subject to AIP license or copyright; see http://proceedings.aip.org/about/rights_permissions
feed and condenser channel. After continuing with the
operation for a significant time, a steady state
temperature profile is achieved as shown in Figure 3.
When steady state conditions are reached, feed water
was entering the system at 85
o
C and exiting at about
35
o
C, whereas condenser water (that was entering the
system at 30
o
C) was leaving the system at about 80
o
C.
A constant temperature difference of about 5
o
C was
thus established across most of the lengths of the feed
and condenser channels.
Figure 3. Simulated temperature profile in the feed and
condenser channels
Fi d
lines) and gas (da on profiles in the
gure 4 shows the gradual evolution of liquid (soli
shed lines) saturati
feed channel at different times during operation, and
also at steady state. Note that gas saturation includes
contributions from both water vapor and air. With
continued operation more and more vapor is generated
(gas saturation increases and liquid saturation
decreases) until a steady state is reached, where liquid
saturation varies between 0.4 (at the feed water entry
side) to 0.2 (at the feed water exit side).
Figure 4. Liquid and gas saturation profiles in the feed
channel.
Ini n
of air is one and the tion of air in the gas
tially, the air-gap contained only air (gas saturatio
mass frac
phase is one). However, vapor generated in the feed
channel diffused through the membrane and entered
the air-gap, where it condensed, and the condensed
water drained out into a collection bucket
(conceptually). Consequently, the liquid and gas phase
saturation (plus the mass fraction of air and water
vapor) changed with time in the air-gap during
operation. This is illustrated in Figure 5, which shows
the transient change in saturation and mass fraction
profiles in the air-gap.
Figure 5. Temperature, liquid and gas saturation, and air
mass fraction profiles in the air-gap.
rate as a
nction of time. This figure shows results from three
Figure 6 s
fu
hows the freshwater collection
operation modes. For these three operations, all the
physical and operational parameters are identical
except the flow rates in the feed and condenser
channels. For Case 1, the flow rate in both the
channels is 1.08 lit hr
-1
. However, these flow rates are
2.16 lit hr
-1
and 3.24 lit hr
-1
for Case 2 and Case 3,
respectively. From Figure 6, it is clear that increasing
the flow rate increases the freshwater collection rates.
However, the freshwater collection rate as a percent of
feed water flow rate remains more or less similar
(because this is governed by thermodynamic
equilibrium between the liquid and vapor phases).
0.00
0.05
0.10
0.15
0.20
0.25
0.30
a
t
e
0 200 400 600 800
Time (hour)
C
o
n
d
e
n
s
a
t
e
F
l
o
w
R
(
l
i
t
/
h
o
u
r
)
Case 1
Case 2
Case 3
Figure 6. Freshwater collection rates as a function of time
for three different feed water flow rates
113
Downloaded 01 Nov 2012 to 182.255.2.5. Redistribution subject to AIP license or copyright; see http://proceedings.aip.org/about/rights_permissions
A
flow rate , and by
ults s in thi er are all based e
tion t diffusi efficie of ai r
apor mixture is only a function of temperature and
ARY AND CONCLUSIONS
Mem le
desa ess
of memstill as an energy-efficient water purification
The author houghtful
technical d Chris
Dotremont about membrane distillation.
1. C. Charcosset, D 1 (2009).
2. G. W. Meindersma, C. M. Guijt, and A. B. de Haan,
Environmental Progress 2 441 (2005).
emical
6.
roperties
13. International Form
Them
Feed Condenser
e (lit
Conde
Temp.
Collection
hr )
dditional simulations were performed varying the
s in the feed and condenser channels
changing the temperature of the water into the
condenser channel. These simulation scenarios and the
freshwater collection rates from them additional are
shown in Table 1.
TABLE 1. Different Simulation Scenarios and Freshwater
Collection Rates From
The res hown s pap on th
assump hat on co nt r-wate
v
pressure and not of saturation. Diffusion coefficient
has a strong influence on diffusive flux of vapor and
freshwater collection rates. Additionally, we
considered operations only for one fixed pressure.
Changing the operational pressure will also change the
freshwater collection rates. Changing other parameters
(such as porosity and tortuosity, which are functions of
spacer geometries) will also change the diffusive flux
and hence freshwater collection rates. Finally, the
presence of dissolved solid will also influence the
extent of vaporization and the amount of freshwater
that can be collected. These different operational
scenarios will be investigated in a future paper with an
objective to optimize the performance of memstill
operations.
SUMM
brane distillation is emerging as a viab
lination technology. To establish the effectiven
technology and to help optimize the design and
operation of a memstill facility, there is a definite need
for developing a better understanding of the
underlying physical processes associated with
memstill and building a robust mathematical model for
it. The conceptual and mathematical models of fluid
flow and heat and mass transport for memstill to date
have been based on the assumption of steady-state
conditions and/or utilizing some empirical relationship
for heat and mass transport coefficients. These
previous modeling approaches do not aim for actually
solving the temperature and the saturation fields in the
various subcomponents of the overall system. In this
paper, we present an alternative modeling approach,
where the transient temperature and liquid and gas
saturation fields in the subdomains are actually
determined by solving the coupled mass and energy
balance equations. The rate at which freshwater can be
collected is determined from these transient
temperature and saturation fields. Freshwater
collection rates under different operational conditions
were determined in this paper. In future papers, we
intend to investigate further operational and design
issues associated with memstill using the approach
described in this first paper. While this present paper is
concerned about a single AGMD module, we plan to
extend the mathematical model to a system involving
multiple AGMD modules.
ACKNOWLEDGMENTS
Scenario Water
Rate (lit
Water
Rat
nser Rate (lit
-1
hr
-1
) hr
-1
) (
o
C)
Case 3 3.24 3.24 30 0.22
Case 4
Case 5 0.31
3.24
3.24
3.24
6.48
20
20
0.26
wish to acknowledges the many t
discussions with Lou Jing an
REFERENCES
esalination 245, 214-23
4, 434-
3. A.Criscuoli, M.C. Carnevale, and E. Drioli, Ch
Engineering and Processing 47, 1098-1105 (2008).
4. C.M. Guijt, G.W. Meindersma, T. Reith, A.B. de Haan,
Separation and Purification Technology 43, 233-244
(2005).
5. L.-H. Cheng, P.-C. Wu, and J. Chen, Ind. Eng. Chem.
Res. 48, 4948-4959 (2009).
J.H. Hanemaaijer et al., Desalination 199, 175-176
(2006).
7. K. Pruess, C. Oldenburg, and G. Moridis, TOUGH2
Users Guide LBNL-43134, Berkeley: Lawrence
Berkeley National Laboratory, 1999.
8. E.A. Mason and A.P. Malinauskas, Gas Transport in
Porous Media: The Dusty Gas Model, Amsterdam:
Elsevier, 1983.
9. S.W. Webb, J. Por. Media 1, 187 199 (1998).
10. A. Cass, G.S. Campbell and T.L. Jones, Soil Sci. Soc.
Am. J. 48, 25 32 (1984).
11. N.B. Vargaftik, Tables on the Thermophysical P
of Liquids and Gases, 2nd Ed., New York: John Wiley
& Sons, 1975.
12. T.N. Narasimhan and P.A. Witherspoon, Water Resour.
Res.12, 57 64 (1976).
ulation Committee, A Formulation of
the Thermodynamic Properties of Ordinary Water
Substance, Dsseldorf: IFC Secretariat, 1967.
14. J.O. Hirschfelder, C.F. Curtiss and R.B. Bird, Molecular
Theory of Gases and Liquids, New York: John Wiley &
Sons, 1954.
114
Downloaded 01 Nov 2012 to 182.255.2.5. Redistribution subject to AIP license or copyright; see http://proceedings.aip.org/about/rights_permissions