вход по аккаунту


Use of Fluidized Bed for Intensification of Microfiltration on Ceramic Membranes.

код для вставкиСкачать
Use of Fluidized Bed for Intensification of
Microfiltration on Ceramic Membranes
P. Mikulaiek' and L. Filandrova
Department of Chemical Engineedng, University of Pardubice,
nim. Cs.legii565,532 10 Pardubice, CZECH REPUBLIC
The influence of a fluidized bed on permeate flux during the microjiltration of model
disperswns by ceramic membranes has been studied Following description of the
basic characteristic of alumina tubular membranes, model dispersions and the
spherical particles use4 some comments about the experimental system and
experimml resultsfor different microfiltration systems are presented From analysis
of the experimental results it is concluded that the use of a fluidized bed resulted in
a significant increase in permeate flux when compared with results obtained in an
empty tube system. This phenomenon is especially pronounced during the
microfiltration of polymeric dispersion when the permeate flux in a fluidized bed
system WQF nearly five times higher. I t was found that the optimum porosity was
approximately 0.8 for a fluidized bed exhibiting maximal values of p e m a t e flux
Cross-flow microfitration is a pressure-dn'ven membrane process used for the
separation of colloidal particles, such as paint pigments, synthetic polymer spheres and
emulsions, from dispersions. The process is applicable for water and wastewater
treatment processes, the processing of food and dairy products, recovery of
electrodeposition paints, treatment of oil and latex dispersions, and in
biotechnology-oriented applications. However, membrane processes for liquid feed
streams are complicated by the phenomena of membraue fouliag and by concentration
polarization in the liquid boundary layer adjacent to the membrane wall. Concentration
polarization (CP)and membrane fouling are major concerns for the successful use of
membrane-based separation operation in cross-flowmode as their net effect is to
reduce the permeate flw, thereby resulting in less productivity. Such Limitationshave
spurred research into the development of new membrane materials and the utilization
of novel hydrodynamic approaches in membrane Separation systems.
The use of ceramic membranes in separationtechnologies is relatively new and has
given rise to much interest in recent years. Ceramic membranes offer excellent
chemical, themnal, and pressure resistance to a wide variety of feed conditions which
are only partially developed (or are completely absent) with polymeric membranes.
Ceramic membranes represent a distinct class of inorganic membranes. Other classes
include membrane materials such as glasses, carbon and metals, and organic-inorganic
Polymers 111.
'Authorf i r correspondence.
P. Mikulris'ek and L Filandrova
TIE m e m h used were aldna-based, i m m a l - ~ t y p e cerarmic
micdiltrationmdmnes supplied by TERRONIC (Hra&c KAlovd, CzechRepublic).
T k y were COQfiglltcd as single cylindrical tubes,02mlaPg, inside diametct 12 mm,
aside diameter 17 mm, consisting of a thiu a-ahrmina layer on top of asuppart. In
our experimeas, middtration manbrans w a used with tbe mean pure diameter
equal to 0.2 pm. Tbe membmnc parasize distribucian was demmined by the liquid
displacement metbod [lo]. Tbe data presented in F i i 1 indicate that the.
doubblayemicaamicmembme d i n thesestudies basaveylrarraw pore size
Twodifferent types offixdwereu3ed eitbcz withtheesapty tube or tht fhridizedbed
system in arder to detedne the effecton flux. Rrsrly, tbe d i s p e h ofanaknnina
P & r , ~ Y , c C . a n m e r c l a l~ o f v i n y l a c e t a t e - e t h y l e n cupolymer
(Appaetaa 3588,Hoechst, Gemmy) in deiwater. The p?aide size ctisaibutioors
of the dispersions used (shown in Figure 1) weat determined by a particle sizer
&&Graph 5100 (Micromentics) and BI-90 @rookhaven Instr. Gorp), respectively.
concenttatianof solids m the dispersiorrs was 1 % (w/w).
Use of Fluidized Bed for Intensificationof Microfiltrationon Ceramic Membranes
Figure 1. Particle size d i s t d d o n of alumina dispersion (I), VAc-Et copolymer
&perasion (2), and p r e size disbibvction of 0.2 pm membrane used (3)
Equipment detaik
The microfiltration studies were canied out in a memkane filtratian unit equipped
with ceramic ltmhmes. Tbe unit allowed smdies in which tbe transaembrane
p s m e and the cross-flow velocity wete iadependently varied A schematic diagram
of the experimeatsl apparanrs is shown in Figure 2 It consisted o f a microfiltratian
module (l),a pump (2),a storage tank (3) equipped with a t h e d regulatian system
(4), and a tempemure and presslae cantrolsystan. Exprimems were carried out at
25 OC, either with (fluidized bed expriments) or without (empty tube experiment)
particles. The liquid was fed into tte bottom of& vertical membrarre ata canstant
volumetric flow rate regulated by ammo1 valves located at the pump outlet. The
permeate was collected &om the outer phiglass tube s u t r e n g the membrane.The
suppogts of tbe m b r a n e consist of two stainless steel meshes c o m p r e s s e d between
rubber packing. A calming Section COLltaining 2 mm diameter spheres was placed
immediately below the support.
Initially microfiltration of prefiltered deionized water was tried under various
transmembrane pressures in the range 0 - 150 kPa, while the superficial velocity was
fixed at a constant value. The dispersion was then treated at an intermediate
transmembrane pressure AP 50 kPa. F a various fluid velocities m the range 0.03
- 0.28 me',the permeate flux (I)
was evaluated. A new membme was used in each
experiment, and before the run the pure water flux was measured with prefiltered
deianized water. The dispersion was tben introduced into the unit, the pump was
tumed on, and operating pressure and superficial velocity wete adjusted by the
regulation system. The flux through the membrane was measured by collecting the
permeate in a graduated cylinder and timing the collection period. Both permeate and
retentate were recycled to the storage tank to maintain a relatively constant dispersion
P. Mikulriiek and L Filandrova
R . e 2.ScJmtalic of eaperinaenial appnms: ( I ) rniaofilrretion d e ; (2)pmp;
(3)stwage tmtR; (4) thermal ngukrzing system; (5) @-pass; (6) permeate owflow;
(7)rete&te o@w.
ccmcenrralim Every experiment was rrntimvAUPSi1the flux became amstant
The fluidized particles w a e eitba glass beads (1.465mm diamew, 2506
density) ar stainlm steel beads (1.001 mm diameter, 7506 Irgm" density). par each
experiment with a fluidized bed, the cohrmn was filled with solids so that the total
expanded bed height was equal to tbe membrane length. The amspmdipg solids
loadhg andsupedi&veloCities were knownfrom prehhryhydrodynamicsresults
obtaid inatramspent* with tksamcdiamemas themnnirmr..
Results and Discussion
) water flow
Typical pure water (pfiltexed deionized water) fluxes as a fuaaian of the
transmemlxane presslrte are shown m Figure 3 fur different ex@men~ straight line
relationships were obtained according to the flux quatian:
J =
The hydrodpamicmembmnere&mce&)isamembrmep.opertyand
depend on the feedcoanpositionaantbe mnsmeanbranepresslpc(AP).A O m b
slope of the plot o f J verms AP, the hydraulicdstawes for empty tube eJQerimeot
& 4.19~10'* d)and tbe flbed experhm
= 4.30~10"m") m y be
is close tounity. These findings indicate
estimated The value of the ratio
that additim of panicles to mate flow does not f l e a the pameate flux of dean
water, as has been reported [6J.However, the absolute value of the m e m b m
hydraulic resistance was slightly higher for the fluidized bed exprima%
Use of Fluidized Bed for Intensijcation of Microjiltration on Ceramic Membranes
Figure 3. Efeet of permeute flrures of Mlter on transmembrane pressure.
Microfiltration of model diqmshm
Analysis of the permeate composition showed complete rejectiaa of the dispersion
particles in the retentate. For all expimental runs, the d i a ~ ~ eofm the dispedon
particles was always larger than the memttme pore size (see Fignre 1) in order to
plevat particle penemtian into of through tbe IneHlbram. values of pemeate flux
vezsus time for various meao feed velocities in the empty tube system are plotted m
F i g u r e 4 , f a r a l u m i n a d i s p e t s i o n a t t h e f e e d ~ ~ 0 fI96 wf solids. The
000 0000000000000
vvwvvvvvvvv 3
3- v
4 - 0
u =0.10 ms-l
u =0.20 ms-1
u =0.28 ms-'
8ooooooooooooo 1
P.Mikulhiek and L Filandrova
steady-statefha was usually reachedad was lower than the plre w a x flra, ranging
fram 30% to 95% oftbe pne water ma f a analumina dispersioa,and from 1% to
2% for a VAc-Et aqolymcr dispersia pigtln 4 shows the wry dependentnatme of
the ma an tbe saperficial velocity ofthe feed Tbe flux decay m tbe initial periods
d be explained by the firmatim o f 8 au&r a d r e (gel) layer QI tbemmrhcnne
surface, which tben offem b cantrouing hydraulic resisrroce to penneation
Figure 5 illusaates the effect of the wrperfickrl velocity -,the steady-state
permeate flux (an a log-log scale) far bob dispersion samples. An increase in
cross-flow velocity from 0.05 to 0.28 ms-l resulted in a permeate flux iocrease by a
faaat of*
3, lmda o t h e d s e d m apaating amTbe Slape of the plot
far an alumina dispersion is 0.34 which is very close to the om-third power
dependency of a system operating uader laminar d t i o n s , and is predicted by tbe
of ( m 3 c e ~ tpolatizatioa
Aheznatively, tbe dope
of the plot for VAc-Et copolymer disperSian is 0.77, which indicates that multiple
1nechad5 fbr pezticle Wdifhskm adgmtimaway frcm t b c b occm
[ll]. On the basis of the meawnemeng mpntinrrpd above, we suspect that Brownian
diffush is d
m both Qlscs relative to sbear-ioduced diffusicq Which ipcreases
withpartick size. Tbe h e r p
edepe&ence fortbe alumina particles may be due
to relatively thick and pemeable deposits in this case.
- o VAc-Et copolymer dispersion
- 0 A1203 dispersion
figure 5. Skodysrote permeate puX vs. supvelocity for empty at#
experiments. (Concenlratwn of so&& in the dispersion 1 96 (W),mnsmembrane
pressure 50 P a )
ToeMhEatethepdbIenwbamms oftbefhadecay,wehavealsoconsklered
the rinsing behaviour ofceramic micmfiluationmembraaes after their exposure to d~
dbpezsbns [la.In
teans, irreversible membrrure fouling qresem the
decawtpe m flux (cumpared with the original pure waaerflux) observed when the feed
Use of Fluidized Bed for Intensification of Microjiltration on Ceramic Membranes
stream is replaced by puce water, suppoeedly leaving the irreversible fauling deposit
intact IIbe flux iacreased quickly with rinsing, then flattened out and appeared to
asympmicaUy approacha final value after 30 minutes. The value of h i n g flux
reached up to 90% of the pne water flux, indicating low irreversible fouling for an
alumina dispersiolL This d be auributed to dnnrinantCPand/or looser packing of
the alumina dispersion deposits. In comrast with these results, the rinsing data for
VAc-Et copolymer dispersion showed higber flux reduction, giving values m the range
from 12 to 35% of the original pure watez fha In the preseat expe&wm with
VAeEt dispersion the irreversible f d g &posit resistance (gel layer) is dnmin?rnt
The addition of particles does not chaage the shape ofthe J(z) curves (see
Figure 6) when cumpared with those for an empy tube expehent (see Figure 4).
Howevex, a strong and Iyp-monotonic depeadence of permeate flux at steady state
(JA upon the superficial fluid velocity and/or porosity may be observed. In general,
the fluidized bed values for the steady-state permeate flux are supeliol to the ones
obtained m an empty tube,withamaximurnratio of 15 far analurnina dispetsion and
5 for VAc-Et copolymer dispessh (see Figure 7). ?hese results agree well with
scattered infom#ior~ on the iatensificatian ofmicrofiltratianby a fluidizedbed [5,6,8].
Solids act as obstacles to the fluid flow m the same way as if they were fixed w i t h
au identical pattern; because oftheirpaesence,turhlence is increasedandCPand/or
calre layers are considerably thinner tban in the empty tube. However, particle motion
may be thought of as being respona'ble for strong and oontirmosrs d u n of particle
deposits at the waU Thenumbez ofmllisic~~
of the particles (permit membrane area
per mit time) is inversely praportional to bed parosity. PartiClecaLe contacts are
certaidy effected mooe than the turbuknce when E is varied from a value of 0.8 (see
Figure 6 Permeate jk vs. time for an alumina dispmion @ui&ed
upcriments). (concentration of solids m the dispersion 1 R; (I@)), transmembrane
pressure 50 P a )
P. MikulriSek and L. Filandrova
- 40
- m
M '
" -
R p r e 7. compCrrison of steadpstate penneatejluws with supc$cial velocity for
&#erent mkrofltration systems. VAc-Et copolymer &psion: I 0 empty acbe
experiments: 1'- m fluidized bed experiments; AhO,: 2 0 empty t u h experiments;
2'- ~Iuidizedbed expcrintenkF. (wncennatwn of solids in the dispersion I R;
irMsmenbrane pressure 50 &a)
Figure 8), where the fluidized bed is less stable and because tbe lxidge-lilre particle
agglmerates are fanned, and slugging flow is observed Particles farm layers, or
slugs, betwen the large liquid pochets rmd rend to move upward in a piston-like
Afiarherof the etosive action of the particles may be h i n e d
when rht results offlnirlilrA bed m
' expezimeatsarecanpadwithsame
visual observations,and witb rhe prediction ofthe stability of a homogeneas fluidized
bed. The following stability criterionfor homogeneously expanded liquid-solid bed has
been presented 1131:
There exists for each bed a certain porosity above which a bamogenew fluidized bed
changes to a plug flow bed. The uitid bed porosity at which the possible uansitian
fram plrrticulate to unstable liquid fluidization begins to d e s t itself, was found to
ic= (Z
- l)/@- 0 5 )
Use of Fluidized Bed for Intensificationof Microfiltration on Ceramic Membranes
Figure 8 V&wn
of steady-state penneate fIuxes with flrridi’zed bed porosity:
1 - VAc-Et wpolymer dispersion, stainless steel beads; 2 - A AhO, &pision,
grclss be&; 3 - Al, 0, dispersion, stuinless steel be&. (wncentr&.on of soliris
in the dispersion 1 S (wfi), transmembrane pressure SO P a )
Based on set of out experiments m a tramparent tube with tbe same diameter as the
microfiltration membme, tbe cridcal bed parosity is close to 0.75 which is slightly
lower that the values obtained m fluidized bed miaduation (seeFigure 8). However,
thepcediaicnrsofuitical bedparosity provided by the E q u a t h (3) are d y valid for
mn-poears tubes, and are very sensitive to the panicle Reynolds number (Re3 and
the density
(Ps /PI.
Attempts have also beea made to explain the process behavbur m terms of tbe
energy efficiency of a fluidized bed. A good comparison between a fluidized bed and
an empty tube system can be given on the basis of required enesgy to obtain a certain
steady-state flux [3]. This energy per mit of time is expressed as the product of
praure-drop and quantity of displaced liquid. The ratio of the required energy for a
fluidized bed system to the energy of an empty tube system is largerthan 1 for both
&persions used, a saving in cjrmlation energy is not achieved for the fluidized bed
systems. However, the comparison of the effect of a fluidized bed with results of
empty tube experiments at velocities of 0.050.3 ms-’ is controversial, while in
industrial instahtim velocities up to 10 ms” are normal at the moment. In addition,
it is wishful rhinkiq to hope the fluidized bed microfltration d d compete with
empty tube microfiltration m those cases where a small-scale operation are used that
produces an extremely sensitive and valuable product. So, a fluidized bed system
necessarily intcoduces a mong limitation CM the quantity of fluid that m a y be treated
per m i t time in an apparatus of a given volume.
P. Mikulriiek and L Filundrova
The results demmsuate that the ceramic mia0filtratic.m membranes used can remove
bscahpninaandcopolymercobid dispasiarrs from wastewateraprocess streams.
Tbe \rpedMbulen#-prcHnotiag fluidid particles d t e d m a s i g n i f i i increase
m permeate flux through the membrane. However, the flux readus a maximum for
the OptirnMl bed porosity and is dependent oil the superficial velocity ofthe feed and
oil tbe behaviaur of
the fluidized bed.
From analysis of hydraulic resistaacts it may be d u d e d that fluidized solids
ezlsllte asiificantreduction mthe concentrationp o h h i m , as well as acaimous
the improved permeate flux that is achieved is due to the d i n e d action of
turbulence and particle 0lOtjc.m (collision of particles with tbe membme wall). This
reduces the thickness d the surface boundary layer and reduces the hydraulic
resistance of 8 adja cake lay=. 'fkrefore, the pmpexties, formation, and
movement ofparticles are domima fadoas that willdetermine the overallbchaviour
of a fluidized bed h
Axhimedes slumber [d3P(P, P)glPl
Diametex of particle or membrane pore (m)
Permeate flux @%-~)
Ste9dy-state permeate flux (kn3h'')
Hydraulic membrane resistance (m-I)
Terminal Reynolds number [drqplpl
Superficial velocity of the feed (IDS-')
Unhindered panicle settling velocity (m-')
Eqonmt in the Ricbiuds~Zaki-typeequation
Greek symbols
Bed porosity
Dynamic viscosity (pa s)
Liquid density Org m S
particle density (kg m",
Empty tube
Fluidized bed
This research work was finaacially supported by the Grant Agency of the Czech
Republic (Grant Project No. 104/93/2306).
Use of Fluidized Bed for Intensification of Microfiltration on Ceramic Membranes
1. Bbave, RR 1991. Inorganic Membmnes, pp. 10-63. Van Nostrand Reinhold, New Yo&.
2. hfikuliihk, P. 1994. Methods to reduce Concentntion polrrization and fouling in membrane filtration.
CoU Czech Chem Commun, 59,737-755.
3. Van der Wad, M.J., Van der Velden, P.M.,Koning, J., Smoldas, CA. and van Swpay, W.P.M. 1977.
Use of fluidised beds as turbulence promoters in tubular membrane system. Dcsalinerion, 22,465-483.
4. De Bar,R, Zonaman, JJ., Hiddink, J., Adderhey&, J., van Swaay, W S . U and Smolders, C.A.
1980. Fluidized beds as turbulence promotersin the wncentmtion of food liquids by reverse OSmOSiS.
J Food Sci, 45, 1522-1528.
5. Montlahuc, G., Tarodo 6 la Fuente, B. and Rios, G.M. 1985. Mass transfer h e e n a homogeneous
fluidized bed and a porous wall. Ennopic, 124,24-27.
6. Rim, G.M.. Rakotoarisoa, H. and Tarodo de la Fuente, B. 1987. Basic transpolt mechanisms of
uluafihtion in the presenoe of fluidized panicles. J Membrune Sci, 34,331-343.
7. Xuesong, W . 1987. Mass transfer and the fluidizedbed intensification of revem osmosis. Desalination,
8. Qavagura. F., Rjimat, E, Elmaleh, S. and Grssmiclc, A. 1991. Intdcatim of microliltrationby
a circulating bed. Kcy 15ngineering Materia,%, 61-62,569-572.
P. d cnkl, L 1994. Ranoval ofinduslrial latex aispasians from waste rmtas wing
miaoDorous alumha mcmbnurs. Dt~~linrrrion.
10. Mila&&, P. and Dolekk, P. 1994. Chammmm
. . *Onofcenmictu~arm~~activepore-size
diseibution Sep Sci T a b &29,1183-1192.
11. Belfort, G., Davis, RH. and Zydney. A.L. 1994. The behavior of SUspeMions and maaamolecular
solutions in crossflow microfiltration J Membrane Sci, 96, 1-58.
12. Brandani, S. and Foscolo, P.U. 1994. Analysis of diswntirmities arising from the onedimensional
equations of change far fluidization. Chem fig Sci, 49,611620.
13. Cakl. J. and MdaWek, P. 1995. Flux and fouling in the crossflow d
c membrane microfiltration
of polymer colloids. Scp Sci Technol, 30,3663-3680.
Rxeived: 13 July 1995; Accepted olpcr revidn: 15 December 1995.
Без категории
Размер файла
544 Кб
fluidized, bed, intensification, ceramic, membranes, use, microfiltration
Пожаловаться на содержимое документа