Professional Documents
Culture Documents
Abstract
-J-hepresent investigation pertains to modeling of seawater desalination system. A simulation model was developed
and veriJied for a small-scale reverse osmosis system. The proposed model combines material balances on the feed
tank, membrane moclule and product tank with membrane mass transfermodels. Finally a comprehensive simulation
model has been developed incorporating the effect ofmass transfer inhibition The model is non-linear differential
equation representing the feed concentration as a function of operating time and space. The solution of the
simultaneous differential equations was obtained using the fourth order Runge-Kutta method, due to self starting
and stability. The model was veritied using the experimental data from the literature [17,24]. Parameter sensitivity was
carried out to select the proper step size. The simulation was run for over 1000 h enabling a prediction of operational
perfomrance at high overall system recoveries.
K~~vow!s: Mathematical modeling; Reverse osmosis: Mass transfer; Inhibition: Concentration polarization
001 l-91 WOG/S- See front matter 0 2004 Hsevier Science J3.V. All rights reserved
PII: SO0 I l-9 164(03)006 16-7
solventandsolutefluxes throughmembraneshave reduceconcentrationpolarization and therefore
beendevelopedandanalyzedneglectingthe effect fouling, by increasingthe friction factor andbulk
ofmasstransferinhibition. Concentrationpolariza- velocity.A model hasbeendevelopedby Chiolle
tion and fouling of the membrane are the two et al. [6] for the reverseosmosis with the turbu-
serious problems that would prevent the use of lencepromotingnetsfor theparallelwall channels
RO into many of the processes.Both of these module. The model developed by Drioli and
phenomenaarc flux inhibiting boundary layer Bellucci [8] showsthe effect of the interactionof
effects and that -foulingis almost always a result theconcentration polarizationandsolute-membrane
of*concentration polarization.Concentrationpolari- interactionon thepressuredrivenmembranewhen
zation may be defined asthe presenceofa higher working with multicomponent solution. The
concentrationof rejected species.at the surface modification ofthe membranepolymeric structure
of a membranethan in the bulk solution, due to plays animportantrole in thereductionof concen-
theconvectivetransportofboth soluteandsolvent tration polarizationthroughthe tluidized bedwas
[ 10,331.It is generallyconsidereda totally revers- developedby van der Waal [:20].Bhattacharyya
ible effect that c.anbe describedin a first appro- et al. [25] developeda Galerkin finite elements
ximation. by a variety of analytical models. The program to compute the concentration profile
fouling in the membraneis thecondition,in which throughouta reverseosmosismembranemodule
membraneundergoespluggingor coatingby some to predict the performanceof the module. The
elementin the streambeingtreated,in sucha way finite elementmethodallowed rapidevaluationof
that its output or its flus is reduced.A model for variousmembranemoduleconfigurations,suchas
colloidal membranefouling hasbeenreportedby taperedcell geometry and channelscontaining
GreenandBelfort [9], which allows thetheoretical spaces.Basedon the available work.,a compre-
prediction of membraneflux declination.A lot of hensivesimulation model hasbeendevelopedfor
work has been done on the membrane foulants seawaterdesalinationsystem.
by thevariousinvestigatorsanda detailedanalysis
is availablein the literature[ 15.161.The measure-
ment and control in reverseosmosisdesalination 2. Models for solvent and solute transport
havebeenstudiedby Mindler and Epstein [22]. in reverse osmosis
The reductionof concentrationpolarizationis
importantfor theimprovementof theperformance Variousmodelsandmechanismfor the solvent
of osmotic type membrane, which will lead to and solute transport through reverse osmosis
reductionin thefouling ofthe membrane.Several membranehavebeendevelopedandproposedby
measuresto reducethe concentrationpolarization a number of investigators [12,19]. The flow of
to controlthe fouling havebeenadoptedand pro- solventthroughthe membraneis definedin terms
posed.The techniquesusedto reducethe concen- of flux:
tration polarization are increasing flow rate,
assemblinganintensifierfor turbulentflow, impulse J,,,= <Qj~s,>
Clv,:, (1)
methods, agitating methods, periodic depres- The solvent flus of the permeatedependson
surizationof membranetube, flow reversal,pre- the hydraulic pressureapplied acrossthe mem-
coatingof membranesurfaces.enzymeimmobili- brane,minusthedifferencein theosmoticpressure
zation, modification of membrane polymeric of the solutions of the feed and permeateside of
structure,mechanicaland ultrasonic vibration o! the membrane[2].
membraneshave been tried also but with little
success13.71.The turbulence promoter acts to J, =A@-An) (2)
?I
t-ration gradient
ROMJZMBUNE
J, = B,AC (3)
Continuau
x:=(+dV)RT (32)
dcpidt=acpiat+(ac,iay)ub (35)
Using Eq. (2 i) for material balance around the
feed tank with the concentration as a function of
time and space, we get
dc, ldt = [e, k’, - C, )dv,o - Q,t))l (36) Fig. 9. Solute transfer on the boundary layer of the Inem-
Substitution of Eq. (35) into Eq. (36) gives brane.
s, = 0. I8 1 111:.
AZ' = 4.02x 10" kg/m h’, A,,, = 4.2~ 10 I7 h/m, 13
Bc= l.l2rlO-.lm/h,
Y = 1.02x IO” m2/h2, Re = 30 -1 OjT~71f/Z.,
= 0.0 15,
0.62 mm < d I: 0.9 mm, 450 < SC < 2650,
0.55 <E -C0.9&0.0623 <X,< 1.455
a value of I.027 kg/m: after 1000 h. The average Fig. 4. Simulation results of feed concentration vs. time.
1.2 ..-
a
b.6
8c
E
‘:
.j”’
;
P
$0.2
I
e”
0
0 loo 200 300 400 500 ml 700 800 %M ml0
Time, t (br) 2 ^ .
decreases almost linearly at a slow rate with time Fig. 7. Simulation results oE feed concentration vs. rime
up to around 400 h of operation of the system. with concentration polarization.
After this period,the feedconcentrationincreases 1. The model developedwithout concentration
exponentially with time. In Fig. 4, the feed con- polarizationwaseffectively usedfor the prediction
centrationgot increasedup to 400 h of operation of feed concentration,permeateconcentration,
without concentrationpolarization, which is the rejection andflux asa function of operatingtime.
contrastwith concentrationpolarization. 2. The proposedmodel developedwith con-
Fig. 8 shows the variation of permeatecon- sideration of spatial dependenceof solute feed
centrationand averageproduction concentration concentration(hencethe permeateconcentration,
with time. The permeateconcentrationC?P remains averagepermeateconcentrationand rejection as
almost constantwith time up to around450 h of a functionoftime andspace)on simulationshowed
operation as is the case without concentration good agreementwith the actual reverseosmosis
polarization. Beyond time t - 450 11,the CP processand improved the permeateflux.
increaseslinearly at a high rate. This trend was 3. In theproposedmodel thebulk feedvelocity
observedalmost sameasexplainedin Fig. 5. The shows the important role which is the remedial
variation of averagepermeateconcentrationwith factor for improvementof permeateflux.
time remains same as observedfor the case of
without concentrationpolarization.
Symbols
u,-n, - Constantsin model equations
7. Conclusions
4” - Solventpermeabilityconstant.h/m
The following conclusionscan bedrawn from B.% - Solutepermeabilityconstant.m!h
the presentstudy: c - Concentration,kg/m’
c, - Solutebulk concentrationflow, kg/m)
C; - Solute feedconcentration,kg/m’
C,, - Initial solutefeedconcentration,kg/m?
c;, - Solute concentrationin the feed tank,
kg/m”
c - Solutepermeateconcentration,kg/m’
c:”,,Y” - Averagepermeateconcentration,kg/m3
CT - Soluteretentateconcentration,kg/m”
C/J - Concentrationasa flinction of time and
space,kg/m’
q,, - Solutewall concentration,kg/m:’
c w/r - Solventpermeateconcentration(water):
kg/m3
D - Diffusion coefficient,m/h
de - Equivalent diameter,(Wtl;)i’(H+ WY),m
4,. - Equivalent diameter of fluidized bed,
d,/[ 1 + (3/2) (1 --E)(djc$J]t m
- Diameterof the fluidized particles,m
0 loo .200 300 4lm 500 600 700 flcn 900 loo0 $ - Tube diameterof the fluidized bed.m
Time, t (iv)
.p’ - Volumetric flux rate,m-ilm2li
- Friction factor
Fig. 8. Simulation results of permeate concentration and
average permeate concentration vs. time with concen- H - Height of the channel,m
tlation polarization. j,, - Chilton-Colbumj factor, (/&?‘)lU,,
A’. .Jand et al. ,” Iksaiinrrlion 160 (2004) 29. -/2 31