This manuscript has been reproduced from the microfilm master. UMI films the
text directly from the original or copy submitted.
Thus, som e thesis and
dissertation copies are in typewriter face, while others may be from any type of
computer printer.
The quality of th is reproduction is dependent upon th e quality of th e copy
subm itted. Broken or indistinct print, colored or poor quality illustrations and
photographs, print bleedthrough, substandard margins, and improper alignment
can adversely affect reproduction.
In the unlikely event that the author did not send UMI a complete manuscript and
there are missing pages, these will be noted. Also, if unauthorized copyright
material had to be removed, a note will indicate the deletion.
Oversize materials (e.g., maps, drawings, charts) are reproduced by sectioning
the original, beginning at the upper left-hand comer and continuing from left to
right in equal sections with small overlaps.
Photographs included in the original manuscript have been reproduced
xerographically in this copy. Higher quality 6" x 9" black and white photographic
prints are available for any photographs or illustrations appearing in this copy for
an additional charge. Contact UMI directly to order.
Bell & Howell Information and Learning
300 North Zeeb Road, Ann Arbor, Ml 48106-1346 USA
Centrifugal Membrane Process
by
David Steven L ycon
B .S c., Saint M ary’s University, 1986
B.E ng., Technical University o f N o v a Scotia, 1990 M .A .Sc., Technical U niversity o f N o v a Scotia, 1993 A Dissertation Submitted in Partial Fulfillm ent o f the
Requirements for the D egree o f DOCTOR OF PH ILO SO PH Y in the Faculty o f Graduate Studies W e accept this dissertation as conform ing
to the required standard
4 /
Dr. T.M . Fyles, Co-Supervisor (Department o f Chemistry)
Dr. V i^ r s ,\C o -S u p e r v is o r (Department o f M echanical Engineering)
Dr. N. Djilali, Member (Department o f M echanical Engineering)
dne, MemI
Dr. M B Hocking, \rcm b er (Departm ent o f Chemistry)
Dr. K.J. Smith, External Examiner (Departm ent o f Chem ical & B io-R esource Engineering, University o f British Colum bia)
© David Steven Lycon, 1999 University of Victoria
All rights reserved. This dissertation may not be reproduced in whole or in part by photocopying or other means, without the permission of the author.
C o-S u p erv iso rs: Dr. T hom as M. Fyles Dr. G e o ff W. V ickers
A B S T R A C T
The C en trifu g al M em b ran e and D ensity S e p a ra tio n (C M D S ) process is a novel type o f m em brane p ro ce ss that exploits the action o f a cen trifu g e to gen erate process pressu re for reverse o sm o sis and nanofiltration. T he c e n trifu g e co uld potentially en h an ce tlux and alleviate fo uling o f the m em brane as a resu lt o f th e hydrodynam ic e n v iro n m en t o f the centrifuge. A ll experim ental w ork has b een co n d u c te d on a pro to ty p e m odel o f the C M D S process. T h e apparatus allow s a m em b ran e m odule to be fixed in sp ace at a specified o rie n tatio n , w ith respect to the ro tatio n . T his o rientation in space is d e n o ted by the term s " p itc h , roll an d yaw " (p.r.y).
E xperim ents have been done using brin e feed so lu tio n s at various co n c en tra tio n s to d eterm in e i f th e C M D S process m in im izes th e effects o f co n cen tratio n p o larizatio n . An exam ple o f th is w as illustrated w ith a 54% tlu x e n h an cem en t relativ e to a con v en tio n al m em brane p ro ce ss u sing a 35000 ppm N aC l feed so lution. C olloidal feed so lu tio n s w ere also used to e x a m in e how the C M D S p ro ce ss e n h an ces tlux in a touting en v iro n m en t. T hese feed so lu tio n s include 21 g/L silica an d 300 m g/L hum ic acid, w ith typical relative tlux e n h a n ce m en t factors (k) found to be 0 .5 9 and 0.1 4 , respectively. T he final g ro u p o f
ex p erim ents e x a m in ed the use o f 50 g/L w h ey feed solutions w ith n an o filtratio n m em branes. R esults obtained here indicate th at the centrifugal a ctio n e n h an ced the tlux with an ab so lu te flux en hancem ent factor (ic') o f 17.5 L /m ‘ hr. T h e se ex p e rim e n ts have show n that a g iv en orien tatio n (9 0 .2 7 0 .0 ) b est en h an ces the flux o f a m em b ran e w ith respect to c o llo id a l fouling, w hile sh o w in g th at a n o th e r o rien tatio n (9 0 ,1 8 0 ,0 ) b est reduces the e ffe c ts o f co n centration p o larizatio n .
Scanning e le c tro n m icroscopy (S E M ) an d a n e n e rg y dispersive x-ray (E D X ) d e te c to r have h elp ed to ex a m in e the nature o f th e fo u lin g layers and d eterm in e h o w w ell the layers ad h ere to th e su rface o f the m em b ran e . It w as determ in ed th at in som e ca se s, the
fouling layer adhered better to the surface o f a membrane used in the C M D S process. H ow ever, as the fluxes were typically higher in the dynam ic process, it leads to the conclu sion that the fouling layer on the C M D S membranes is more permeable.
From the experim ental work it has been concluded that the forces at work in the CM DS process create sufficient secondary flo w instabilities to reduce the effects o f fouling and concentration polarization on the membrane surface. The sign ifican ce o f this process with respect to industrial applications is considered, and the process is deem ed feasible for such applications.
Examiners:
Dr. T.M . Fyles, C o-Supervisor (Department o f Chem istry)
Dr. 'G.W .'wickers, Co-Supervisor (Department o f M echanical Engineering)
Dr. N. Djilali, M em ber (Department o f M echanical E ngineering)
ing. M en
Dr. M .B. Hocking, M ember (Department o f Chem istry)
Dr. K.J. Smith, External Exam iner (Departm ent o f Chem ical & B io-R esource Engineering, U niversity o f British Colum bia)
IV TABLE OF CO NTEN TS A b s t r a c t ...ü T a b le o f C o n t e n ts ... iv L is t o f T a b l e s ...vi L is t o f F ig u r e s ... vii N o m e n c la tu r e ...x A c k n o w le d g e m e n ts ... xi D e d ic a tio n ... xii C h a p t e r 1: In tro d u ctio n 1.1 M em brane P ro c e sse s ... 1 1.1.1 M em b ran e F o u lin g ...2
1.1.2 M em brane Fouling R eduction T e c h n iq u e s...5
1.1.2.1 C le a n in g ...5
1.1.2.2 Feed P re-T reatm en t...7
1.1.2.3 M em brane/P rocess D e sig n ... 7
1.1.3 E v o lu tio n o f C entrifugal M em brane P ro c esse s...10
1.2 C M D S P ro cess D e sc rip tio n ... 13
1.2.1 M em b ran e O rien ta tio n ... 16
1.3 P revious and O n g o in g A cadem ic W ork on C M D S P ro c ess... 19
1.3.1 FR .A C T... 19
1.3.2 C o m p u tatio n al Fluid D ynam ic M o d e ls...20
1.3.3 A n g u la r Influences in C M D S P ro c e s s ...22
1.4 P relim in ary C alib ratio n E x p erim en ts...22
1.5 P rin cip les o f E xperim ental W o rk ... 28
C h a p t e r 2: C o n c e n tra tio n P olarization 2.1 In tro d u c tio n ...30
2.2 E xp erim en tal P ro c e d u re ...33
2.3 R e su lts ... 35
2.3.1 O rien ta tio n - 0 ,r,y ...35
2.3.2 O rien ta tio n - 9 0 ,r,y ... 38
C h a p t e r 3: F ouling 3.1 In tro d u c tio n ... 47
3.2 H um ic A c id F o u lin g ... 48
3.2.1 In tro d u c tio n ...48
3.2.2 E xp erim en tal W o rk ...49
3.2.3.1 R oll 90/270 R esu lts... 53
3 .2.3.2 R oll 180 R e su lts... 60
3 .2 .4 M icro sco p ic A nalysis o f F ouling L a y e rs ...64
3.2.4.1 E xperim ental P ro c ed u re...64
3 .2.4.2 S E M and ED X R e su lts... 65
3.3 C o llo id al S ilica F o u lin g ...72
3.3.1 In tro d u c tio n ...72
3.3 .2 E xp erim en tal W o rk ... 73
3.3.2.1 P relim in ary E x p erim en ts...73
3 .3.2.2 P ro ced u re F or Fouling E x p e rim e n ts ...74
3.3.3 R e s u lts ... 75
3.3.3.1 O.r.O R e su lts... 76
3 .3.3.2 90.r.0 R e su lts... 78
3.3.3.3 9 0 .180,y R esu lts... 79
3 .3 .4 M icro sco p ic A nalysis o f F ouling L a y e rs ...81
3.3.4.1 S E M R e su lts... 81
3.3 .4 .2 E D X R e su lts... 84
3.4 N a n o filtra tio n ...8 6 3.4.1 In tro d u c tio n ...87
3.4 .2 E x p erim ental W o rk ... 90
3.4.2.1 P relim in ary E x p erim en ts... 90
3.4.2.2 P rocedure For Fouling E x p e rim e n ts ...91
3.4.3 R e s u lts ... 92
3.4.3.1 V ariable Pressure R e su lts ... 92
3.4 .3 .2 F ixed Pressure R e su lts... 95
3.4 .4 M icro sco p ic A nalysis o f F ouling L a y e rs ...96
C h a p t e r 4: C o n c lu sio n s 4.1 C M D S P ro cess C o n c lu s io n s ... 100
4.1.1 M em b ran e O rie n ta tio n ... 102
4 .1 .2 F o u lin g in C M D S and C o n v entional M em b ran e P ro cesses 104 4.1.2.1 F o uling Feed Stream D iffe re n c e s ... 104
4 .1 .2 .2 F o uling L ayer P e r m e a b ility ...106
4.1.2.3 A dh eren ce o f Fouling L a y e r...106
4.2 T o w ard In d u strial A p p lic a tio n s... 107
VI LIST OF T A BL ES T able 1 . 1 T able 1 . 2 T able 1.3 T able 1.4 T able 1.5 T able 1 . 6 T able 2 . 1 T able 2 . 2 T able 2.3 T able 3.1 Table 3.2 T able 3.3 T able 3.4 Table 3.5 T able 3.6 T able 3.7 T able 3.8 T able 3.9 T able 3.10 T able 3.11 T able 3.12 T able 3.13 T able 3.14 T able 4.1 Industrial ap p licatio n s o f m e m b ra n e s ...2 F o u lan t d iffe re n c e s...3 P hysical C lean in g m e th o d s...6 P re-treatm en t m e th o d s... 7 FR.ACT o p eratin g ra n g e s...20
S u m m ary o f calib ratio n e x p e rim e n ts...25
DS-3™ m em b ran e m aterial s p e c ific a tio n s... 33
9 0.r.y head e x p erim en ts o p eratin g pressure ra n g e s ... 35
R elativ e tlux v a lu e s... 46
F iltratio n effectiv en ess and particle s iz e ...51
H um ic acid fouling ex perim ental c o n d itio n s...53
a and ic v alues for roll 90/270 o rie n ta tio n s ...59
a and k values for roll 180 o rie n ta tio n s ...63
D u P o n t L udox p ro p e rtie s...73
C o llo id al silica fouling ex p erim en tal c o n d itio n s... 75
a an d ic v alues for O.r.O o rie n ta tio n s... 78
a an d k v alues for 90.r.0 o rie n ta tio n s...78
a and k v alues for 90.180.y o rie n ta tio n s ...80
T y p ical solids co n te n t o f w h e y ... 89
DS-5™ m em b ran e m aterial sp e c ific a tio n s ... 90
W hey so lu tio n fouling ex p erim en tal c o n d itio n s...92
ic' values and “ ro llo v er" pressu res for v ariable pressure e x p e r im e n ts 94
a
an d k v alues for fixed pressu re e x p e rim e n t (ex p e rim e n t # 4 ) ...96 S u m m ary o f m em b ran e o rien tatio n perfo rm an ce forLIST OF FIGURES Figure 1.1 Figure 1.2 Figure 1.3 Figure 1.4(a) Figure 1.4(b) Figure 1.5 Figure 1.6 Figure 1.7 F igure 1.8 Figure 1.9 F igure 1 .10 Figure 1.11 Figure 1.12 Figure 2.1 Figure 2.2 Figure 2.3 Figure 2.4 Figure 2.5 Figure 2.6 Figure 2.7 Figure 2.8 Figure 2.9 Figure 3.1 Figure 3.2 F ouling m e c h a n ism s...3
Forces actin g on a particle at the m em b ran e surface (S p in tek a p p a ra tu s)... 9
T ay lo r v o rtices in a rotating a n n u la r m o d u le ...10
C R O a p p a ra tu s ... 12
T ypical spiral w ound c a rtrid g e ... 12
C M D S A pp aratu s - d o o r c lo s e d ... 13
C M D S A p p aratu s - d o o r o p e n ... 14
M em brane m odule s ta c k ... 15
M em brane head c o n fig u ra tio n ...15
E xam ples o f m em brane o rie n ta tio n ... 18
Static m em b ran e a p p a ra tu s ... 23
T e m p era tu re dep en d en ce on perm eate tlu x ...27
M em brane process c o m p a riso n ...28
C o n cen tratio n profile du rin g c o n c en tra tio n p o la riz a tio n ...30
O.r.O o rie n tatio n co n cen tratio n p o lariza tio n e x p erim en t (35000 pp m N a C l)...37
O.r.O o rie n ta tio n co n cen tratio n p o lariza tio n e x p erim en t (70000 pp m M gS O a)... 38
90.90.0 o rie n tatio n co n cen tratio n p o la riz a tio n ex p erim ent (22500 pp m N a C l)...39
90.180.0 o rie n tatio n co n cen tratio n p o lariza tio n ex p erim ent (22500 ppm N a C l)...40
M em brane m odule o rien tatio n and flow d ire c tio n ... 41
A verag ed su rface m ass fraction a lo n g the flow channel for various o rie n ta tio n s... 42
R elative flux results for c o n c en tra tio n p o lariza tio n e x p e rim e n ts ... 44
9 0 ,I8 0 ,y o rie n tatio n effect o f C o rio lis forces e x p e rim e n t... 45
E xam ple o f h um ic acid m o le c u la r s tru c tu re ...49
V l l l F ig u re 3 F igure 3 F ig u re 3 F igure 3 F igure 3 F igure 3 F ig u re 3 F igure 3 F igure 3 F igure 3 F igure 3 F ig u re 3 F igure 3. F ig u re 3. F igure 3 F ig u re 3 F ig u re 3 F ig u re 3 F ig u re 3 F ig u re 3 F igure 3 .3 H um ic a c id ex p erim en t # 1 ... 55 .4 H um ic acid ex p erim en t # 2 ... 56 .5 H um ic acid ex p erim en t # 3 ... 57
. 6 H ypothetical fouling cu rv e w ith q u an tificatio n n o m e n c la tu re ... 59
.7 M em brane m o d u le o rie n tatio n and feed tlow d irectio n (9 0 .1 8 0 .y )...60
. 8 H um ic acid ex p erim en t # 4 ... 61
.9 H um ic acid ex p erim en t # 5 ... 62
. 10 L ocation o f deposit on m em b ran e m odule after 90 .1 8 0 ,9 0 e x p e rim e n t....63
. 1 1(a) S am ple m o u n ted on a lu m in u m stub for surface im a g e ... 65
. 11(b) Sam ple m o u n ted on m o d ifie d alu m in u m stub for edge im a g e ...65
. 1 1(c) Side v iew o f S E M a lu m in u m m o u n tin g s tu b ... 65
.12 S urface electro n m icro g rap h s ( lOOOx m ^.gnification) of: (a) unused m em brane: (b) static h u m ic acid fouled m em brane: and (c) 90.270.0 o rie n tatio n d y nam ic hum ic a c id fouled m e m b r a n e ...6 6 13 S urface electro n m icro g rap h s (lOOOx m ag n ificatio n ) of: (a) unused m em brane: (b) static hum ic acid fouled m em brane: and (c) 90.270.0 o rie n tatio n d y n am ic hum ic a c id fouled m e m b ra n e ... 67
14 Edge electro n m icro g rap h s (lOOx m ag n ificatio n ) of: (a) unused m em brane: and (b) static h u m ic acid fouled m e m b ra n e ...6 8 15 Edge e le c tro n m icro g rap h s (lOOx m ag n ificatio n ) of: (a) 90.180.0 o rie n tatio n d y n am ic hum ic acid fouled m em brane: and (b) 9 0 .180.90 o rie n tatio n d y n am ic hum ic a c id fouled m e m b ra n e ...6 8 16 H2O rin sed m em brane su rfa c e electro n m icro g rap h s ( 1 OOx m ag n ificatio n ) of: (a) static h u m ic acid fouled m em brane: (b) 90 .1 8 0 .0 o rie n tatio n d y n a m ic h um ic acid fouled m em brane: and (c) 9 0 .1 8 0 .9 0 orien tatio n d y n a m ic h um ic acid fouled m e m b ra n e ... 71
17 M em brane m o d u le o rie n ta tio n and feed flow d irectio n (O.r.O)...76
18 C o llo id al silica e x p erim en t # 1 ... 77
19 C o llo id al silica e x p e rim e n t # 2 ... 79
20 C o llo id al silica e x p erim en t # 5 ... 80
21 S urface ele c tro n m ic ro g ra p h s of: (a) un u sed m em b ran e ( 1 OOOx): (b) static silica fouled m em b ran e (lOOOx): and (c) static silica fouled m em b ran e (5 0 0 x )... 82
Figure 3.22 Figure 3.23 Figure 3.24 Figure 3.25 Figure 3.26 Figure 3.27 F igure 3.28 Figure 3.29 Figure 3.30 Figure 3.31 Figure 3.32 Figure 4. l Figure 4.2
S u rfa c e electron m icro g rap h s (1 OOOx m ag n ificatio n ) of: (a) static silica fouled m em brane; (b) 9 0 .1 8 0 .9 0 o rie n tatio n dy n am ic silica fouled m em brane; an d (c) 9 0 .1 8 0 .0 o rie n tatio n dynam ic silica
fouled m em b ran e ... 83
E dge e le c tro n m icro g rap h s (IOOx m ag n ific a tio n ) of; (a) unused m em b ran e; and (b) static silica fouled m e m b ra n e ...84
H2O rin sed m em brane su rface e le c tro n m icro g rap h s (IOOx m ag n ificatio n ) of: (a) static silica fouled m em brane; (b) 90.180.0 o rie n tatio n dynam ic silica fouled m em b ran e; and (c) 90.180.90 o rie n tatio n dynam ic silica fouled m e m b ra n e ...8 6 G el p o larizatio n c o n c en tra tio n p ro file ...8 8 G e n e raliz e d NF tra n sp o rt re la tio n sh ip ...89
W hey ex p erim en t # I ... 93
W hey exp erim en t # 3 ... 94
W hey ex p erim en t # 4 ... 95
S u rfa c e electron m icro g rap h s (2 5 0 x m ag n ificatio n ) of: (a) unused m em b ran e; (b) static p rotein fouled m em b ran e; and (c) 90.1 8 0 .9 0 o rie n tatio n dynam ic protein tb u led m e m b ra n e ...97
E dge electro n m icro g rap h s (IOOx m ag n ificatio n ) of: (a) unused m em b ran e and (b) static p rotein fouled m e m b ra n e ... 98
E dge e le c tro n m icro g rap h s (lOOx m ag n ificatio n ) of: (a) unused m em b ran e and (b) dy n am ic (9 0 .1 8 0 .9 0 orien tatio n ) protein fouled m e m b ra n e ... 98
9 0 .1 8 0 .0 orientation w ith referen ce a x i s ...103
N O M EN C LA TU R E
9- feed c o n c en tra tio n
Cp p erm eate concen tratio n
d particle d iam e te r
dp pore d ia m e te r
jv v o lu m e flux across m em brane
k m ass tra n sfe r co efficien t
P pitch angle
Pr roll o v e r p ressure
r roll angle
y yaw angle
Cb b ulk fluid concen tratio n
Cg gel lay er fluid co n centration
Cw w all fluid concen tratio n
D o r Ds force d u e to back diffusion
Fd force a sso c ia te d w ith the m em brane tlux
Pc cen trifu g a l force
G g rav itatio n a l force
Jv m ass flux acro ss m em brane
4 m em b ran e perm eability
R m em b ran e rejectio n
S;Si su lp h u r to silica ratio
a relativ e slo p e factor (fouling)
P relativ e tlux (co n cen tratio n polarizatio n )
Ô b o u n d ary lay er thickness
AP hyd ro static p ressu re difference
A n o sm o tic p ressure d ifference
K relativ e o ffse t factor
K m o d ified o ffse t factor
t r radial sh e a r force
A C fC N O W L E D G E M E N T S
I w o u ld like to thank Dr. T om F yles for his g u id an ce and su p erv isio n th ro u g h o u t the c o u rse o f this d eg ree program . 1 e sp ec ially ap p reciate his tolerance w ith respect to m e
perio d ically tak in g tim e o f f to pu rsu e m y N aval R eserve career. His goo d sense o f h u m o u r and easy -g o in g nature m ad e this project a little easier. T h an k s are also w arranted to the o th e r g u id in g faculty m em bers o f the C M D S project, nam ely Dr. G e o ff V ickers an d Dr. N ed D jilali; their input has been appreciated.
S everal co lle ag u e s have also h elped to m ake the tim e sp en t at U V ic better both p ro fessio n ally and personally. T hese p eo p le include, in no p articu lar order. Jon Pharoah. A lv in B ergen, L ynn C am eron. D arryl B rousm iche. D ave R obertson and S cott M urphy. T w o o th e r gen tlem en , w ho helped out sig n ifican tly in the progress o f this research w ith th e ir m ach in in g ex p ertise, are Roy B e n n e tt and R ichard R obinson.
F in ally . 1 w o u ld like to thank m y fam ily and A nnette for th eir su p p o rt and patience th ro u g h o u t the pursuit o f this deg ree. T h e y can rest assured that this is finally the end. an d th ere w ill be no m ore deg rees forthcom ing.
XI I
D E D IC A T IO N
T o Dr. W illiam B ridgeo. D ean E m eritus o f S cience at Saint M a ry 's U niversity. T his is the m an w ho to o k a chan ce on a young, so lid C stu d e n t and sh ow ed him the w orld o f research and in tro d u ced him to th in g s that a C h em ical E n gineer m ight do. T hank you.
In th e field o f m em b ran e sep aratio n , fouling has long been co nsidered o n e o f the m ajor d raw b a ck s w ith respect to increased use o f the pro cesses. S im ply stated, fouling is the process resu ltin g in loss o f perfo rm an ce o f a m em b ran e due to th e d e p o sitio n o f su sp en d ed o r d isso lv ed su b stan ces o n its external surfaces, at its pore o p en in g s and w ith in its p ores [1 |. F ouling ty p ically leads to hig h er operatin g costs becau se o f the red u c tio n in th ro u g h p u t, and becau se o f the increased need for cleaning. T h e goal in fouling reduction, a n d su b seq u e n t tlux e n h an cem en t, is to create an e n v iro n m en t w here su sp en d e d a n d /o r d isso lv ed particles are d isco u rag ed from m igrating onto the surface o f the m em brane. O ne way to do this is to create unstable flow near the su rface o f the m em brane. T his m ay be acco m p lish ed by u sing hydro d y n am ic m eth o d s that will im prove m ass tran sp o rt o f the p articles from the m em b ran e surface, back to the bulk so lution. T he w ork c o n tain ed in this d issertatio n e v alu ates perfo rm an ce o f a system w h ere u n stab le flow is d e v e lo p e d in a cen trifuge. T h e process is called C en tritu g al M em brane and D ensity S ep aratio n (C M D S ). T h e goal o f w ork is to d eterm in e the nature a n d extent o f flux e n h a n ce m en t and fouling red u ctio n d u e to the rotating en v iro n m en t.
1.1
Membrane Processes
M em b ran e tec h n o lo g y is a relativ ely recent d e v e lo p m en t in the a re a o f separation science. It w as the d e v e lo p m en t o f the sy n th etic a sy m m e tric m em brane by S ourirajan a n d L oeb in 1960 that sp aw n ed the future d e v e lo p m en t o f m em brane p ro cesses [2]. T h o u g h in itially used in the d e salin atio n o f seaw ater, m em b ran e processes have now found th eir w ay into o th er pro cess industries (e x a m p le s in T able 1.1). T hey have the a d v a n ta g e o v e r o th er types o f sep aratio n , in that there is no phase change n ecessary for the se p ara tio n to o c c u r and o p e ra tin g c o sts are g en erally low er.
It m ay be first useful to d efin e som e o f the te c h n o lo g y th at will be e x a m in ed in the fo llo w in g ch apters. R everse o sm o sis is described as a p ressu re driven pro cess th at rejects so lu te from its c a rrie r so lv e n t v ia a se m i-p e rm e a b le m em brane. T his d iffu siv e pro cess h as the ab ility to sep arate p articles d o w n to 5 Â in size from a solvent. N a n o filtra tio n refe rs to the type o f m em b ran e th at relies on size se le c tiv ity to achieve a se p ara tio n o f
solutes. Size selectivity can also re fe r to the m olecular w eig h t o f so lu te com ponents. T he essential d ifferen ce b etw een R O an d N F is that RO in v olves so lv e n t diffusion th rough the b ulk m em brane m aterial, w h ile N F is a true filtration w ith the sm aller solutes p assing th ro u g h voids in the m em b ran es {< 10 A particle size). N anofiltration sep aratio n s do not have th e sam e rejectio n cap ab ilities o f reverse o sm o sis, but also do not require th eir h ig h er d riv in g p ressure. T h e y can be used to c o n c en tra te suspended or co llo id al m aterial in aq u eo u s stream s, rem o v e d issolved m ac ro m o le cu le s from the stream , and de-salt stream s. R esp ectiv e ex am p les o f these types o f N F separations include: (1) dew aterin g o f sy rups; (2) p ro tein fractio n atio n /sep aratio n ; a n d (3) rem oving
N aC l from w hey.
In d u stry .Application
A gribusiness D airy - e fflu e n t treatm en t, m ilk dew atering Food - starch, su g ars separation
F erm en tatio n - w ine, dairy
Industrial R ecovery o f v alu ab le products - paints, dyes M etallurgy P recious m etal recovery and rem oval
E nvironm ental R ecovery o f h a z ard o u s by-products Pulp and p ap er - sp en t sulfite liquor M unicipal Sanitary w a stew a te r - tertiary treatm ent
W ater su p p ly - d e salin atio n o f brackish or se aw a te r B io tech n o lo g y P h arm aceu ticals - serum , deionized w ater
G enetic e n g in e e rin g - fractionation, sep aratio n F erm en tatio n - cell h arvesting, en zym e cla rific a tio n
Table l . l Industrial a p p licatio n s o f m em branes
1 .1 .1 M em b ra n e F ou lin g
T h e prim ary su b je c t o f th is d isse rta tio n in v o lv es fouling and h ow the C M D S process can m in im ize its d e le terio u s effects. T h e refo re, it is w orthw hile to ex a m in e fouling m ech an ism s and cu rren t trends used to alle v ia te fouling. F oulants th em selv es can be c h a ra c te riz e d by th eir siz e in su c h a m an n er: ( 1 ) particles - 10"* to 10^ A. a n d include such th in g s as w hite blood cells, red blo o d cells, b acteria, yeast, a n d p latelets; (2) co llo id s - 1 0
to 10’* A. w hich include viru ses. D N A . lip o p ro tein s, album in and a n tib o d ie s; and (3) so lu tes - 1 to 10 A. in clu d in g insulin, a n tib io tics, sugars and salts. T a b le 1.2 illustrates the relative differen ces b e tw e en th e se th ree types o f foulants. E ach ty p e o f foulant
m em b ran es. F o r ex am p le, reverse osm o sis m em b ran es could p o ten tially be fouled by all three o f the abo v e categories o f foulants, w here n an o filtratio n m em b ran e s w ould pro b ab ly o n ly be fouled by particles and co llo id s. It is for this reaso n , that foulant c h a ra c te ristic s a re im p o rtan t con sid eratio n s in the selection o f a m em b ran e m aterial for a giv en ap p lic atio n .
D iffu sivity (cm '/s) O sm o tic P ressu re R elative V iscosity
P a rticles < 1 0"' negligible high
C o llo id s 1 0'” - 1 0"' low m oderate
S o lu tes 1 0'- - 1 0 ° high low
T a b le 1.2 'o u la n t differences [3]
It is th ese foulant sizes that o ften dictate the type o f fouling that w ill occu r on the m em b ran e su rface. Figure 1.1 illustrates the three prim ary fouling m ec h a n ism s that m ay o c c u r in R O an d N F m em branes, and ho w they relate to particle size.
CASE A: PORE MAHROWING/CONSTRICTION
d « dp Adsorption
CASE B: PORE PLUGGING
dp
d - d p BlocKage Ü Ü "
C A SE C ; GEL/CAKE LAYER FORMATION
d » dp Deposition
A d so rp tio n fouling, o r pore narrow ing o ccu rs w h en particles in a feed stream are ad so rb e d onto the m e m b ra n e 's surface, thus red u cin g the av ailab le size o f the pores. T he ch em istry o f the m em brane is such that d isso lv e d m olecules from the feed so lu tio n are quick ly so rb ed onto the m e m b ra n e 's su rface. T his m ay o ccu r by a n u m b er o f m ech an ism s th at include the follow ing: (1) electro static interaction; (2) h y d rophobic
effects; a n d /o r (3) charg e transfer (h ydrogen bonding). In this fouling m echanism , the initial ad so rp tio n occurs q u ite rapidly, and w h en all o f the ad so rption sites are filled, a p seudo ste a d y -sta te is reached. It is this initial adsorption o f so lu te that creates a n u cléatio n site for further build-up on the m em b ran e su rface (m o stly w ithin the m em b ran e structure). A fter this occurs, a m o n o la y e r can form on the m e m b ra n e 's su rfa c e, and. in som e cases, additional layers w ill eventually form , b ecom e com pacted a n d im p ed e b oth the c ro ssflo w and perm eatio n v elo cities [3 |. T h e p h e n o m en o n o f a d so rp tio n is e sp ecially im portant in the area o f nan o filtratio n w here the pore size is large c o m p ared to rev erse osm osis.
Pore b lo ck ag e o r p lu gging is the only one o f the three discussed fouling m ech an ism s that is c o n sid e red to be reversible. This type o f fouling is caused by large d iam e te r particles p h y sic a lly b locking the pores o f the m em brane, as illustrated in F igure 1.1. T hese c o n ta m in a n ts are often agg reg ates o f a p a rticu la r so lu te in the feed so lution. U pon being b lo cked, the p ores can be cleared o f its plugs by sim ply b ackflushing the m em brane w ith p erm eate, w hich is one o f the m ethods that w ill be d iscu ssed later in this section.
T h e final m ech an ism , gel layer form ation, is a sso c ia te d w ith m ac ro m o le cu la r fouling on the m em b ran e surface. It is a type o f c o n c en tra tio n p o larizatio n asso ciated w ith n a n o filtratio n and ultrafiltration. T he term c o n c en tra tio n p o lariza tio n refers to the c o n c e n tra tio n profile th at has a higher level o f s o lu te nearest to the u p stream m em brane su rface c o m p a re d to the m o re-or-less w e ll-m ix e d bulk fluid far from the m em brane su rface [1]. T his is discu ssed further in the sections dealin g w ith co n c en tra tio n p o lariza tio n an d nan o filtratio n (S ection 2.1 a n d S u b sectio n 3.4.1. resp ectiv ely ).
T here are three im p o rtan t w ays in w hich fouling and c o n c en tra tio n p o larizatio n can be m inim ized: (1) m em brane cleaning; (2) feed pre-treatm en t; and (3) m em b ran e/p ro cess
design. E ach o f these techniques has been w ell research ed , as tlux d eclin e is a serious p roblem in the accep tan ce o f m em brane se p ara tio n s as an eco n o m ically v iable process. T he last tw o areas are considered prev en tiv e, but often it is the tlrst area that is the m ore co m m o n ly practiced o f the three, in c o n v en tio n al m em b ran e processes.
1.1.2.1 C le a n in g
W hen tlu x d eclin e has reached the p oint w here the m em b ran e system is no longer p erfo rm in g at accep tab le levels, c lean in g o f the m em b ran e m ust be undertaken. M em brane m an u factu rers define this situ atio n w ith the follow ing g u idelines: (1) 10% decline in tlux w hile o p eratin g at co n sta n t pressu re and tem p eratu re; (2) 1 0% increase in
d riv in g p ressure required to m aintain co n sta n t tlux; o r (3) 15 to 20% increase in d ifferen tial p ressure betw een feed and co n c en tra te stream s [5]. T he m ethod used to clean the sy stem d ep e n d s on the type o f foulant and the m aterial o f the m em brane. C leaning can be d iv id e d into three categories: ( I ) ch em ical; (2) physical (so m etim es su bdivided into h y draulic a n d m echanical); and (3) ph y sio -ch em ical. C hem ical m eth o d s are ty pically used for irreversible fouling, a n d physical o r p h y sio-chem ical m ethods are g enerally used for reversible fouling situations. C h em ical clean in g m eth o d s are also m ore p rev a len t w here reverse o sm o sis m em brane resto ratio n is co n cern ed , and n a n o filtratio n a n d ultrafiltration m em b ran es rely m ore on the physical and physio- chem ical m ethods.
C h em ical c lean in g acts to d issolve the fouling layer, o r to create a rea c tio n at the m e m b ra n e ’s su rfa c e favourable for foulant rem oval. C h em icals are u sually introduced w ith a lo w pressu re, flushing stream into the m em b ran e m odule. T he type o f ch em icals w hich are used in m em brane cleaning include the follow ing; (1) acid s (H N O3, citric acid
and H3P O4) - used in the rem oval o f c a rb o n a te and su lp h a te scales; (2) b ases (N aO H ) -
used a fte r an a c id w ash to n eutralize th e m em b ran e su rface; (3) c o m p le x in g agents (E D T A ) - n e cessary for the rem oval o f C a p recipitates; (4) en zym es - can c le a n p rotein
build-up on m em branes; (5) d eterg en ts - rem o v e o ily deposits from m em b ran e; (6)
c o n cen trated N aCl so lutions - used to rem ove protein foulants; and (7) o x id an ts/d isin fectan ts (N aO C l and H2O2) - rem o v e biological slim es [6].
A s m en tioned above, physical c le a n in g m ethods are often c h aracterized as being m echanical o r hydrodynam ic. T able 1.3 d e scrib es som e o f the m ore co m m o n m echanical a n d h y d rodynam ic c lean in g m ethods th at are cu rren tly being used for m em brane restoration.
M ethod D escription
forw ard flushing perm eate is p u m p ed into the feed section o f the pro cess to clean foulant from m em b ran e surface
reverse Hushing direction o f p erm eate How is alternated b etw een the forw ard and b ackw ard d ire c tio n
perm eate b ack pressure reversing th e How o f perm eate by ap p ly in g large back pressure, and a t the sam e tim e, allow ing a feed so lu tio n to go to the m em b ran e and w ash aw ay loosened foulant particles
vibration pneum atic h am m ers are used on the pressure vessel to loosen foulant p articles, w hile m aintaining a feed Hush near the m e m b ra n e 's su rface
a ir drain and w ater refill the pressure vessel is evacuated w ith air. then im m ed iately filled w ith w ater w hich creates turbulence at the g a s/w ate r interface (tu rb u le n c e d isp la c es the foulant particles
a ir sparge periodic in je c tio n s o f a ir ahead o f Hush stream (u sefu l for h o llo w fibre m em b ran es)
so nication ultrasonic c le a n in g w ith a w etting agent
sponge ball clean in g po ly u reth an e sp o n g e balls are inserted into the p ressure vessel for a few se co n d s to scrub the su rface o f the m em brane (o n ly w o rk s u n der turbulent How co n d itio n s)
T a b le 1.3 P hysica clean in g m eth o d s [6]
T h ese p h ysical m ethods are o fte n used in c o n ju n c tio n w ith chem ical c lean in g m eth o d s to create the categ o ry o f ph y sio -ch em ical cle a n in g . E xam ples o f these m eth o d s used fo r RO m em brane tre a tm e n t include; (1) u sing rev e rse H ushing w ith a su rfactan t in th e c le a n in g
stream ; (2) a c id w ash stre a m used in c o n ju n c tio n w ith foam ball sc ru b b in g ; a n d (3) peracetic a c id and h y d ro g en p eroxide w o rk in g w ith a reverse flush tech n iq u e [6].
In m an y situ atio n s, pre-treatm ent o f the b u lk feed so lution can alter the co n d itio n s that m ay b rin g about fouling or c o n c en tra tio n p olarization at the m e m b ra n e 's surface. E ffectiv e pre-treatm en t requires k n o w le d g e o f the nature o f the fouling m echanism , and th e m em b ran e m aterial (excessive p re-tre a tm e n t m ay be harm ful to the m em brane). E x a m p le s o f pre-treatm ent include: (1) pH adjustm ent; (2) heat treatm ent: (3) change in ionic stre n g th : (4) use o f seq u esterin g ag en ts: (5) ch lo rination: and (7) co a g u la tio n for p re-filtra tio n . T able 1.4 gives e x a m p les o f som e o f the differen t types o f foulants, and w h a t m eth o d o f pre-treatm ent could be used for each.
F o u la n t T rea tm en t D escrip tion
C a salts pH adjustm ent acid ad d itio n replaces su lp h ates and bicarbonates w ith m ore soluble chlorides
silica heat treatm ent raised tem perature increases solubility
co llo id s co agulate and filter co lloids form larger p articles w hich can be filtered out
b a c te ria ch lorination rem oves (k ills) bacteria
F e p recip itate pH adjustm ent acid dose stab ilizes and keep s Fe in solution
T a b le 1.4 P re-treatm ent m eth o d s [7]
1.1.2.3 IVIembrane/Process D esign
T h e a re a o f desig n can be broken d o w n into one o f tw o areas: (1) m em b ran e m aterial p ro p ertie s: and (2) m em brane p rocess m o d ificatio n s. T he form er deals w ith the surface b o n d in g effe c ts o f fouling, and the latte r w ith h y d rodynam ics. T his form o f fouling red u c tio n ty p ically costs less than p o st-p ro ce ss cleaning techniques, and also helps to b e tte r u n d e rsta n d the nature o f tlie feed stream , as w ell as the sep aratio n p ro cess itself.
In term s o f m em brane m aterial, the po re size and the distrib u tio n o fte n go v ern s how a m em b ran e w ill foul, th erefore if the m em b ran e is dense in its pore d istrib u tio n
{i.e.
large n u m b e r o f pores), it is less likely to foul as q uickly. S urface charge o n the m em b ran e can a lso be u sed to m inim ize the effects o f fouling w hen used in c o n ju n c tio n w ith the charge a sso c ia te d w ith the foulant. O ften c o llo id s in the feed so lu tio n are n e g a tiv ely charged.8
a n d th u s a negatively c h a rg e d m em brane w ould act to repel the colloid from its surface. A n o th e r su rface effect is the hydrophilic o r h y d rophobic nature o f the m em b ran e. I f the m em b ran e is very hy d ro p h ilic it w ill a ttract w ater to its su rface, thus ca u sin g the m em b ran e to rem ain " w e t’', an d not a llo w p articles to adhere to its surface as easily .
T h ere are several techniques th at have recen tly been dev elo p ed in b oth the m an u fa ctu rin g and su rfa c e m o d ificatio n areas. Som e o f the m anufacturing adv an ces have c o m e in the in tro d u ctio n o f nitro g en -co n tain in g polym ers, and the use o f new m an u fa ctu rin g tec h n iq u es su ch as dip co ating, plasm a po ly m erizatio n , interfacial p o ly m erizatio n and w ater castin g . T he su rface m o d ificatio n m eth o d s include plasm a treatm ent, fiuorination o f h y d ro ca rb o n polym ers, and grafting o f p h o sp holipids onto the m em brane. P lasm a tre a tm e n t is a n area that includes, d e p en d in g on the d e g re e o f p lasm a d isch arg e, reactions such as su rface p lasm a etching, p lasm a m o d ificatio n o f the chem ical stru ctu re o f the su rface layer, and p lasm a po ly m erizatio n [8 |. T he fiuorination o f the m e m b ra n e 's
p o ly m e r film by using h ydrofiuoric acid o r fiuorosilic acid increases the h y d ro p h ilic ity o f the m em b ran e , and thus red u ces the ability o f solutes to foul [9j. A p h o sp h o lip id co atin g m im ics the m an n e r in w h ich red blood cell plasm a m em branes resist protein fouling. In this so rt o f process, a m icro filtratio n m e m b ra n e 's surface is plasm a etc h e d and then c o a te d w ith a pho sp h o ry lch o lin e so lution. T his process reduces the a m o u n t o f protein fouling at the m e m b ra n e 's surface, and creates a low er fiux d ecline c o m p a red to u n treated m em branes [1 0].
F o r pro cess m o d ificatio n s, hy d ro d y n am ics refers to the m an n er in w hich the bulk feed so lu tio n reaches the su rface o f the m em b ran e. S om e o f these m o d ificatio n s include: (1) in creasin g c ro ssflo w velocity; (2) im p ro v in g feed sp acers and inserts; (3) use o f pulse
fiow ; (4 ) T a y lo r and D ean vortices; (5) sh o rt path lengths in fiat m em b ran e m o d u le s; (6)
b ac k p u lsin g o f perm eate; and (7) use o f h ig h surface area, low fiux hollow fibers [11].
A m eth o d o f interest th at is capable o f c re a tin g h igh sh ear rates d u e to the p rese n c e o f T a y lo r v o rtices (flo w instab ilities) is the u se o f ro tating m em b ran e m o d u les. In th ese ty p es o f sy ste m , the high sh e a r rate is c a u se d by the h igh m em b ran e speed, as o p p o se d to
m em b ran e d isk used for u ltrafiltration. It u ses the actio n o f a c e n trifu g e to create shear forces a t the surface o f the m em brane that act to reduce fouling on the surface. T his is illu strated in Figure 1.2: w here Fd is the force asso c ia te d w ith the m em b ran e flux. Fc is
the cen trifu g al force' w hich is a function o f p a rticle m ass and ro tatio n al speed, D is the force d u e to back diffu sio n . Tr is the radial sh e a r force, and To is the tangential shear force. A system such as this also reduces the so lute build-up due the density differen ce betw e en the p olarized layer and the b u lk so lu tio n (co n cen tratio n p o larizatio n ). A nother m em b ran e m odule that will g ive the high sh e a r rate is the ro tatin g a n n u lar m odule (F ig u re 1.3). T his type o f system relies o n a rotating inner m em b ran e m odule and a sta tio n a ry o u te r c y lin d e r (p ressu re v essel) to create the T ay lo r v o rtices on the m em brane su rface. T h is type o f process is c a p ab le o f co n tin u o u s m icro filtratio n in the turbulent T a y lo r-v o rte x regim e [12]. T h e b asic theory b ehind these e n h a n ce d sh e ar force m em b ran e sep aratio n p rocesses is sim ila r to th at o f the researc h to be d iscu ssed in this dissertatio n .
F ig u re 1.2 Forces ac tin g o n a p article at th e m em b ran e su rface (S p in tek
a pparatus) [1 1]
T he term "centrifugal force” Is a postulated one. H ow ever, because the term has co m e into pop u lar use, it w ill be used through out this dissertation to refer to th e force d irected outw ard from th e axis o f rotation.
10
n u rc r c y lin d e r
F igure 1.3 T ay lo r v ortices in a rotatin g a n n u la r m odule (see F ig .l o f [1 2 |
1.1.3 E volution o f C en trifu g a l M em b ra n e P rocesses
A s d iscu ssed above, advantages can be o b tain ed by using rotating m em b ran e p ro ce sse s to c reate tu rbulent, o r at least unstable. How at the m em b ran e surface. B efore d e scrib in g in detail the process used for the e x p e rim e n ts con tain ed in this dissertatio n , it is w orth lo o k in g at som e at som e o f the e a rlie r process d esig n w ork done in this area. W ork d o n e in this area by G renci [13] in tro d u ced a system w here a rotating cy lindrical m em b ran e uses the cen trifu g al force g e n erated to create the pressure necessary to driv e the reverse o sm o sis process. In this system , the m em brane is placed on the inside o f a ro tating c y lin d e r w all, and the salt w ater feed is introduced in the c e n te r o f the cylinder. T he actio n o f the cen trifu g e causes the m ig ra tio n o f the salt w ater to the w all, and thus to the m em brane. A n in v en tio n by K eefer [14] ad d resses the problem asso c ia te d w ith kinetic en erg y losses due to rotor w indage in a rotary reverse osm o sis process. A n im p e lle r feed pum p, w hich is integral to the ro tatin g m em brane p ressure vessel, creates the req u ired feed p ressure for the reverse o sm o sis process. T he ad v antages o f the ce n trifu g a l e n v iro n m en t on the m e m b ra n e 's su rface can still be ach ieved in th is c o n fig u ratio n . As the w o rk in g p ressure is due to the p u m p and not the centrifuge, lo w er ro ta tin g sp eed s are p o ssib le, and th u s a reduction in w in d a g e losses.
T he theory b ehind the process in vestigated in this dissertatio n w as in itially developed from w ork do n e by W ild and V ickers in v o lv in g a centrifugal rev erse o sm o sis (C R O ) d esalin atio n system . In C R O , low pressure feed w a te r enters the d ev ice a lo n g a rotational axis and m o v es to the p eriphery o f a sp in n in g ro to r w here the p ressu re d ev elo p ed is sufficient to d riv e the reverse osm osis process. P erm eate is released at the perip h ery , and the retentate is returned to the rotational axis b efore leaving the dev ice at low pressure. T his critical asp ect o f C R O results in energy sa v in g s o f 40-60% that o f c o n v en tio n al high pressure reverse o sm o sis [15]. The p atented d e sig n uses an e v a c u a te d en clo su re to further reduce en erg y losses due to the ro to r w indage [16]. T h e d ev elo p ed C R O p rototype uses six teen co n v en tio n al spiral w ound m em brane ca rtrid g e s on the rotor. In using the c irc u la r p attern (F igure 1.4a) o f spiral w o u n d cartridges (F ig u re 1.4b) in the C R O process, an infinite nu m b er o f m em brane orien tatio n s w ere p o ssib le, w ith respect to the ce n te r o f rotation. In this type o f a p p aratu s, there is no w ay o f kn o w in g i f one p articular m em b ran e o rie n tatio n is m ore a d v a n ta g e o u s than any o th er, w ith regards to ex p loiting cen trifugal forces in the reduction o f fouling.
T he process referred to in this d issertatio n is the C entrifugal M em b ran e D ensity Separation (C M O S ) p rocess. C M O S applies the sam e process as C R O (feed along axis o f rotation an d p erm eate release at the perip h ery ) to develop the p ro ce ss pressure, but unlike C R O , uses fixed o rien tatio n s o f the m em b ran e s w ith respect to the ro tational axis. In a process su ch as the C M O S system , the m em b ran e can be o rien ted in one o f several w ays relative to the axis o f rotation, each o f w hich is expected to cre a te a specific type o f reduction in fouling. T h is system utilizes c en trip etal acceleration an d C o rio lis forces, generated by a ro tatin g cen trifu g e, to create the unstable flow n ecessary to reduce the fouling (in clu d in g c o n c en tra tio n polarization) at the m em brane surface. T h e follow ing sections w ill d e scrib e h o w the process w o rks, m em brane o rie n tatio n , previous and ongoing academ ic w o rk in v o lv in g C M O S , an d p relim inary calib ratio n ex p erim en ts.
1 2 Conceniraicd Scawaicr Out Fresh Water Out Feed Seawater In
F igure 1.4(a) C R O a p p a ra tu s (see F ig .I o f [1 5 |): and
CON' ,centbat£OUT m e m b r a n e PERMEATE _ SPACER FEED SPACER MEMBRANE
1.2
CMDS Process Description
The ex p e rim e n ts for this d issertation involve the use o f an a p p a ra tu s d esigned and co n stru cted at the U niversity o f V icto ria (F ig u re 1.5). A b rie f ap p a ra tu s description follow s, but a m ore com plete d e scrip tio n is av ailab le in [18]. T he ap p a ra tu s has been d esigned to a c co m m o d a te rotors o f vary in g g eom etries and o p e ra tin g speeds. T he principal c o m p o n e n t o f the d esign is a 5 -foot diam eter rotor h o u sin g w hich can be ev acu ated to m in im ize (fictional p o w er loss and heating due to w in d ag e. A 15 hp drive m otor, d isc b rake, and oil lubricated b earin g s are m ounted on the rear o f the rotor housing, a n d the en tire housing is m o u n ted on vibration iso latio n feet to m inim ize the effects o f m in o r im balances in the test rotors. A circulation pum p, tanks, heat exchanger, and p lu m b in g tak e fluid to a rotary c o u p lin g w hich co nnects the sp in n in g rotor to the stationary p lum bing. T he rotor system co n sists o f an arm w ith a m em b ran e head at one end an d a c o u n te r balance at the other. T he shaft and m em b ran e head have been co n stru cted o u t o f titanium in o rd er to red u ce the w eight a n d the stress problem s asso ciated w ith high g forces. P lum bing fixed to the arm c arries the feed fluid to the m em brane head an d returns the retentate to the rotor axis (F ig u re 1.6). T he rotor is capable o f a c h ie v in g pressures up to 8300 kP a, at rotational speeds up to 2200 rpm.
14
S i
!
F igure 1.6 C M D S A pparatus - d o o r o p en
M em brane m odules used in this app aratu s are stack ed in a m em brane h o ld er, w hich is in turn placed in the m em b ran e head that is a tta c h e d to the rotor. F igure 1.7 sh o w s how the individual m em b ran e m odules w ould be sta c k e d in the m em brane holder, and F igure 1.8 rep resen ts the m em brane head c o n fig u ratio n and its asso ciated n o m en clatu re. T he m em brane m odules each consist o f a p e rm e ate sp a ce r fixed betw een an im perm eable layer and a lay er o f m em brane. T he m em b ran e m odules, up to 9 in the m em brane cell, all face in the sam e d ire c tio n and are su b je c t to the sam e relative d ire c tio n o f centripetal acceleration. Each m odule has an av erage m e m b ra n e area o f 50 cm ". T h e perm eate that flow s through the m em brane head is e v e n tu a lly released into the v acuum housing. Prior to this release, the p erm eate flow rate, c o n d u c tiv ity and tem perature a re m easured in a custom m easu rem en t cell (d iscussed fu rth e r in S ection 1.3.1). T his instrum entation, alo n g w ith o th er ap p a ra tu s system s, rep o rt to a L abV iew " interface w hich m o n ito rs and tracks perfo rm an ce, and in the event o f failure, sh u ts d ow n the apparatus.
peT itie^te
?ott
7= r= F= t
Active
Surface
Permeate
p l o w S p a c e rimpermeable
B a c k m SStacked
M e m b r a n e M o d u le sFigure
^ 7 Membrane m« odule stack [191Figure
1.8 M em brane^dconUgutauoo
16
B efore the process is initiated, the m em brane h o ld er (co n tain in g the m em b ran e m odules) is secured into the m em b ran e head w ith eig h t bolts. T he vacuum housing d o o r is sealed and the vacuum pu m p is tu rn ed on to bring the interior vacuum to ap p ro x im ately 28 in Hg. W hen the p ro ce ss begins, feed enters alo n g the rotor axis and is p um ped at low- pressure to the m em b ran e head at the ro to r periphery. As the ro to r sp in s the pressure increases w ith th e ro tatio n al speed. It is this p ressure that provides the d riv in g force for the perm eation a c ro ss the m em b ran e, a n d the p erm eate released at the p erip h ery produces the tran sm em b ran e pressure. T he c o n cen trated feed returns to th e rotary coupling and then exits the v acu u m housing at low p ressure to be returned to the process feed tank.
1.2.1 M em b ra n e O rien ta tio n
As m entioned in S ec tio n 1.1.3. the m em brane m odules in the C M D S process can be fi.xed into one o f several o rien tatio n s. M em brane orien tatio n plays an im p o rtan t role in the d esign o f the ex p e rim e n ts. G iven o rien tatio n s are available w hich w ill b est exploit the benefits o b tain ed from the C orio lis and cen trifu g al forces. In later ch a p te rs, this will com e to light w h en d e scrib in g the role each o f these tw o forces play in the reduction o f fouling and c o n c en tra tio n p o larizatio n at the m em b ran e surface.
In an effort to k eep the nom en clatu re for th ese o rien tatio n s straig h tfo rw ard , analogies have been d e v e lo p e d w hich utilize the term s "p itc h , roll and yaw ". T h e se three term s refer to the ro ta tio n s ab out the z. x and y axes, respectively (refer to F igure 1.9). T he pitch angle is a c h ie v e d by u sing one o f the tw o m em brane heads av a ilab le for the C M D S apparatus. H ead #1 rep resen ts the 0° p itch an g le, w hile head #2 rep resen ts the 90° pitch angle. G iven th e c o n stra in ts o f the hard w are, these tw o pitch an g les are the only ones currently av a ilab le in the C M D S apparatus. T he roll an g les av ailab le d e p e n d on w hich o f the tw o m em brane h ead s is being used. For head #1. o nly tw o roll an g le s are possible. 0° and 180°, and th ese d e p e n d on h ow the m em b ran e m odules are placed into the m em brane h older. I f the m o d u le s are placed in the h o ld e r w ith th e active m em b ran e su rface facing dow n, the roll a n g le is c o n sid e red to be 0°. W h en the m odules are p laced in the holder w ith the active m em b ran e su rface facing up, th e roll angle is c o n sid e red to be 180°. W hen using m e m b ra n e head #2. any roll an g le, from 0° to 360°. is po ssib le to obtain. A s
is the case w ith h ead #1. m em branes c a n be placed into the m em brane h o lder w ith e ith e r the a ctiv e m em b ran e surface facing up o r dow n. In th ese cases, the roll an gles d ifferin g by 180° c a n be a c h ie v e d w ithout h av in g to ph y sically tu rn the ro to r 180°. Y aw an g les alw ays re fe r to the w ay in w hich the m em b ran e h o ld er can be m ade to rotate w ith in the m em b ran e head. G iv en that there a re e ig h t bolts co n n e ctin g the h o ld er to the head, there are e ig h t d iffe re n t y aw an gles possible: 0°. 45°, 90°, 135°, 180°. 225°, 270° and 315°. F igure 1.9 illu strates ex am ples o f ho w the m em brane is o rien ted w ith respect to the axis o f rotation.
W ith th ese three descrip to rs, su b seq u en t o rien tatio n s w ill be given by 3 num bers, in the o rd er o f pitch, roll a n d yaw . For ex am p le, the o rie n tatio n 0,180,90 refers to a pitch o f 0°, a roll o f 180° and a y a w o f 90°. O cc a sio n a lly , the letters p.r.y will be used if a g eneric d e scrip tio n o f the o rie n tatio n is d esired for e x p erim en t d escrip tio n . An ex am p le o f this w ould o c c u r w h e n d e scrib in g m em brane head #2 ex p e rim e n ts as 9 0.r,y ex p erim ents.
O ne o th e r rela te d co m p o n e n t o f m em b ran e orien tatio n is the d irection o f spin o f the cen trifu g e rotor. T h e C M D S ap p a ra tu s is cap ab le o f rotating in eith er a clo c k w ise or co u n te r-c lo ck w ise directio n . H ow ever, for all o f the ex p erim en ts describ ed in this, and the next tw o c h a p te rs, the directio n o f sp in is alw ays co u n ter-clo ck w ise as illu strated in Figure 1.9.
1 8
Active face
0
,
0,0
90,180,0
t
Roll PitchFeed flow
Permeate face
^ 0,180,180
Rotation about
axis
90,180,90
90,90,0
90,90,90
1.3
Previous and Ongoing Academic Work on CMDS Process
O v er the past few y e a rs, w ork inv o lv in g the C M D S process h a s g iv en rise to several academ ic w o rk s by o th e r grad u ate stu d en ts, asid e from the w o rk b eing describ ed in this d issertation. T hese have included d e v e lo p m en t o f a c u sto m m easu rem en t device, com putational tlu id d y n am ic m odels, and a d e te rm in a tio n o f a n g u la r influences on the C M D S process. T h is section w ill p ro v id e b rie f d e scrip tio n s o f th ese w orks, and how they relate to the e x p e rim e n ts to be d escrib ed in this and later ch apters.
1.3.1 F R A C T
T he flow rates p ro d u ced by the ex p e rim e n ts u sing the C M D S p ro to ty p e w ere envisioned to be fairly low (< 20 m L /m in). and thus it b ecam e n ecessary to m easure the flow w hile the process w as running. D ue to this factor, and also the h arsh en v iro n m en t present in the C M D S process (2000 rpm and 3200 G 's ) . it w as d eterm in ed th at th ere w as a need for a custom m ade m ea su rem e n t device. T his device, referred to as the C M D S Flow Rate and C o nductivity T ra n sd u c e r (F R A C T ). w as d ev e lo p e d by Peter B yrnes in ful Ailm ent o f the thesis portion o f his M .A .S c. degree [20]. T his w as co n ceiv ed , d e sig n e d and built for use onboard the m em b ran e head at the end o f the rotor o f the C M D S apparatus. In the concept p h ase o f this w ork, it w as determ in ed that flow rate an d co n d u ctiv ity w ere the target q u a n titie s to be m easured by this device. T hese tw o q u a n titie s w ould help to determ ine h ow m uch perm eate w as p assin g th rough the m em b ran e , and at w hat quality (% rejection = ^ c o n d u c tiv ity )). A s the p erm eate w as to be rele ase d into the vacuum housing, this d ev ice had to be m ade to capture, m easure, an d rele ase the perm eate. T he challenge asso c ia te d w ith this d e sig n w as to b u ild a d ev ice that co u ld w ithstand high g ravitational forces a n d v ibrations, w hile being able to d e liv e r sig n als from w ithin the device to a user in terface located ou tsid e.
From initial ex p e rim e n ts, it w as determ in ed that the ranges fo r flo w rate and con d u ctiv ity should be 0 .5-20 m L /m in and 0-5 m S /cm . respectively. F o r th e flow m eter p art o f the device, a "All and e m p ty " c h am b er w as en v isio n e d for use. T h is c h a m b e r w ould have an electro m ag n etic a c tu a te d valve th at w o u ld a llo w for the co n tro l o f the filling an d d raining o f the d ev ice. Inside o f the fill ch a m b e r, fluid level-sen sin g e le c trica l c o n tacts w ould
2 0
create the signals n e c essa ry for the flow m ea su rem e n t determ ination. T h e co n d u ctiv ity m easurem ent w as to be d e term in ed by using tw o parallel platinum p lates p laced in side o f the fill cham ber. T h is m easu rem en t w as tem p eratu re co m p e n sa te d b ased on a tem p eratu re-sen sin g chan n el w ithin the F R A C T . T h e conductivity v o ltag e sig n al w as converted to m S /cm (X ), w h ich in turn co u ld be converted to ppm N aC l (Y ) usin g the follow ing 3‘^‘* o rd er po ly n o m ial:
Y = -3 .1 5 0 0 + 4 8 6 .5 4 X + 10.176X " -0 .7 4 5 3 .\1 0 '-X ’ ( l . l )
O nce initial testin g w as co m p lete, the F R A C T w as attached to the perm eate port o f the m em brane head for d y n a m ic testing. The p e rfo rm a n c e o f the FR A C T d u rin g this testing is sum m arized in T a b le 1.5. M ea su rem en t R an ge P recisio n Flow R ate 0.5-20 m L /m in ±2.5% (w lO m L /m in ) C onductivity H igh R ange 0.44-5 m S /cm ± 0.3% L o w R ange 0-0.34 m S /cm ± 0.1% T e m p era tu re 15-45 X ± 0 .5 % T a b ic 1.5 F R A C T o p eratin g ranges
1.3.2 C o m p u ta tio n a l Fluid D yn am ic M odels
N um erical m o d elin g o f the C M D S process co m p lem en ts the e x p e rim e n tal w ork co n d ucted using the process. T he C FD m o d els w ere developed by Jon P haro ah in fulfillm ent o f the th esis po rtio n o f his M .A .Sc. d e g re e [21 ], as well as in his o n g o in g PhD d issertatio n research. T h e m odels consist o f a th ree-dim ensional flow ch an n el w ith a perm eable m em b ran e su rface, w here the m em b ran e itse lf is m odeled using a b o undary co n d itio n rep resen tin g the preferential rem oval o f one co m ponent o f the so lu tio n [22]. S im ulations in v o lv in g both a conventional static m em brane process and the rotating C M D S process w ere e x a m in ed , and com pared.
T h e fluid flow in the C M D S process is d e te rm in e d by the c o n serv a tio n o f m ass, the N av ier-S to k es e q u a tio n s and a sc a la r transport e q u atio n . T hese g o v e rn in g e q u a tio n s are
o u tlin ed in both o f the above noted references. T h e boundary c o n d itio n s sp ecified for these equations include inlet, o u tle t and n o-slip co n d itio n s, a s w ell as a selective m em brane b o undary co n d itio n . T his m em brane boundary c o n d itio n is b ased on both the m ass ( J s ) and vo lu m e ( j v ) tlu.\ across the m em b ran e, rep re sen te d by the follow ing
respective equations:
y , = I ( A f - A H ) (1.2)
cy ) (1-3)
K
w here R is the m em b ran e rejectio n , cy is the feed co n c en tia tio n .
Cp
is the perm eate co n cen tratio n .Lp
is the m em b ran e p erm eability. AP is the h y d ro sta tic p ressure d ifference and AIT is the o sm o tic p ressu re difference.T h e m odel and b o u n d ary co n d itio n s are solved w ith a finite e le m e n t a n a ly sis using the C F D code T A S C tlo w S d . S om e o f the results for this a n a ly sis are related to the co n cen tratio n p o la riz a tio n e x p e rim e n ts co n d u cted on the C M D S ap p aratu s, and are d iscu ssed in S ectio n 2.3 o f th is dissertation. In g en eral, the C F D ex p erim en ts help to h ig h lig h t p h en o m e n a th at m ay be occurring at the su rface o f the m em b ran e , w hich in turn m ay be reducing the effects o f concen tratio n po larizatio n a n d /o r fouling. T h is type o f w ork helps to sh o w h o w the th eoretical w ork can g reatly e n h a n ce the o verall q u ality o f the ex perim ental w ork. It can pro v id e directio n for the ex p e rim e n ts, thus n eg ating the need for a lot o f "tria l and e rro r" w ork. T his factor is im portant b ecau se o f the sig n ifican t am o u n t o f tim e n e c essa ry to c o n d u c t a single set o f ex p e rim e n ts. A n e x a m p le o f this en h an cem en t w as h ig h lig h te d by the C F D d e te rm in a tio n that sh o w e d th at C o rio lis forces w ere playing a sig n ific a n t role in the creatio n o f the a fo re m e n tio n e d flow in stab ilities on the m em brane su rface. It w as this w ork w hich d ire c te d the e x p e rim e n tal focus to w ards ex am in in g specific o rie n ta tio n s (to be d escribed in the next tw o ch ap ters).
1.3.3 A n g u la r In flu en ces in C M D S P rocess
A s the current C M D S apparatus is a p ro to ty p e , a v iew tow ards eventual sc ale -u p has been exam ined. T h is research w as co m pleted by A lv in B ergen in fulfillm ent o f the thesis portion o f his M .A .S c. degree [18]. T h is w o rk ex am ines angles th at need to be considered for an eventual m odule design th a t c o n ta in s m ore m em brane m aterial. T his thesis goes into g rea t detail on the w ork th at w en t into the m echanical d esig n o f the C M D S pro to ty p e ap p aratu s. B ergen also e x a m in e s su ch areas as tem p eratu re d ep en d en ce and tem perature co n tro l. T h is becam e an im p o rta n t issue, as it w as d ete rm in e d that tlux increased by 3 .6 % for every
°C
increase in tem p eratu re. He also ex p an d s upon the co n centration p o lariza tio n w ork (to be d isc u sse d in the next chapter) by tak in g a c lo se r look at the rele v an c e o f the yaw angles to tlux e n h an cem en t.B ased on his ex p e rim e n tal w ork. Bergen d e te rm in e d that the m em brane o rie n tatio n in future C M D S m em b ran e m odules should em p lo y radial feed How w ith m axim um C oriolis a cceleratio n . T his lim its the m odule d e sig n to a co nfiguration w ith an n u la r rings o r axial vanes. H e determ in ed that d e v e lo p in g a cartridge design w h ere the en tire c ircum ference o f the rotor is filled w ith m e m b ra n e m aterial w ould m ax im ize pack in g density but w o u ld be difficu lt to im plem ent. A ro to r d esign w ith d iscrete cylindrical pressure vessels a n d m o d u lar m em brane ca rtrid g e s w as found to be a practical alternative. A c a rtrid g e d esig n em ploying a se rie s o f double sided m em b ran e disks aligned in the p lan a r orien tatio n w ould c o n ta in p ro p er geo m etry to m axim ize p erform ance a n d w o u ld sim p lify m anufacturing. B ergen proposes that by in co rp o ratin g the feed d istrib u tio n c h a n n els into a split m em b ran e d isk support shell, a ro b u st and easy to assem ble c a rtrid g e d esig n w ould result. T h e m o d u le w o u ld be functio n ally sim ila r to the spiral w o u n d m o d u les used in the C R O ro to r d e sig n [16], but o p tim iz e d for the dynam ic en v iro n m en t.
1.4
Preliminary Calibration Experiments
Initial e x p erim en ts w ere c o n d u cted on a c o n v e n tio n a l p ressu re-d riv en m em b ran e (static) apparatus (F ig u re 1.10). T h is apparatus has an iden tical m em brane head to th a t used for the C M D S a p p a ra tu s. T his allow ed for the sa m e m em b ran e holder, c o n ta in in g a g iven
also consists o f a h ig h -p ressu re rec ip ro c a tin g pum p, ca p ab le o f p ro d u cin g ap p lied pressu res o f up to 690 0 kPa. A dditional e q u ip m en t in clu d es a blad d er a c cu m u la to r to d a m p e n the p u m p 's p u lses, tw o feed tanks, a back p ressu re regulator, p ressu re reg ulator, an d associated p lastic a n d stain less steel p lum bing. Flow and c o n d u c tiv ity are m easu red in the static process u sin g a tim ed fill volum e and a bench top co n d u c tiv ity m eter, respectively.
I
wleas
Figure 1.10 S tatic m em b ran e ap p a ra tu s
T h e initial calib ratio n e x p e rim e n ts w ere p e rfo rm ed to see i f th ere w ere an y factors that co n trib u te d to im p ro v ed flux perfo rm an ce, reg ard less o f w h e th e r the p ro ce ss w as d y n am ic o r static. T he ex p e rim e n ts w ere co n d u c te d to e x a m in e m em b ran e o rie n tatio n w ith in the m em brane h o ld er, flo w inlet an d o u tle t angles
