Escolar Documentos
Profissional Documentos
Cultura Documentos
fPlt
-fa r gcnt p /"
.
e-t
y.JfrGe
$$fi
Doble Mukesh
GE India Technology Centre Pvt. Ltd
7r'{n'
chemical-reaction step is II ;tr*\; flowsheet for a new procls5 i.':* ii Taking that rsaction from th6,
A
,the
ant is
ard:Iti,'
ncc thc
viabilitv
before it reaches the manufacturing plant. Because the capital investment required for tran-qforming a raw material to a finiShed product i,s proportir-rnal t0 ther caltacitv of'tiic planl
Keep in mind that there is usually a knowledge gap between bench chemists and process chemists on the one hanC, and chemical engineers and chemical technologists on the other. If
convcrsion, distribution betri'et'n tirt,' main product and byproducts. lrcat ef fects, florv patterns, backnixing. gasliqtrid-solid interaction and rssucs involving process controi.
success-
thi' i) 6 -0.; I)o\\-('r. it r. rlcsu' ebic trl itrrlrl iis lrrllc ii plnnt as is t:ont
rliistil
srst,cnt rr,ith tht- cxpecttrci futrrrc kct fol the product [1]. -st:aler.rll
tlar-
{ull.r' I'rridge L}ris ga;t. t}rc rcsult-" in the ltrli-scalc plant rnay inclucle a reduccd f ield, longer batch c1'clc Limes, great,cr
during the development of a chen-rical proccss. This article {bcuse-s main}v on st:alcup of thc rcur:tion stcyt, rvhich
n'pir:ailv represer-rts lhr: ireal't r)f lr ncu' J)r'{)c(rss. Thr: articlc as-st:mblcs a ri'idt rangc of practical rcaction-scaleup guidelincs that previousll' havc been dilfuserd throughout the en$neering
anounls of effluents or undesirable byproducts, unneccssarily hazardous operations, or unduly high consumption ofenergy. Regardless of the typer of proces-s involvcrd, the problems of scaleup are highil' sinrilar. Accordingjy, the information presented in this article
should apply to virtually all of the
chernical process industries, including such seemingly diverse activities as biochemical s'r'ntheses, production of paper pulp, and processing of metal orL's.
Iinori'lccigcnirle rcaction scalcup encr()r'n])asses sevr-:rirl ar-ipects o{- cht't-n-
very high speed of agitation might givl entirely different yields in a plant's
litr:raturt'. Included are the key elerntnts in developing :r scaleup -slratr,rr'. and ti'rc unknori.ns tirat thc r:ngilrr,r,r nlust atldrcss carlv in thc scaleult
,
.l',:,tr',i iillclttio jt is iiivi n il,rrt i: ,;i ir, scalrLtps irrvrrlving lgtllrl iolt tincl
lliring.
b:caus{'
thcr rilLt,r-r
p1'cs('nL
istn'.
:r'rr
cial chlrllcngcs [,']1. Intelirgcnt sr:aleup cntails finrling rind coliating all the infornration and rii'tii r^t:rlr.rirld lbr tlrr: dc'sign. cor-rst.r'trc:i;rin. :inrl siarlr.lp ol lhc nttrv full-srzt: irjrrri'. ii tlu.ii rtlso t;tkr'into lir,:('orrnt
46
3;iII",1ICi..i-
hi'drodv-
nanric-. alfect lhe sclection of the herst t1'pe of reactor, it.s intc.rnal design anri
x'ill clictatc
JANUARY
ENGII']iI-RIN'i
V/WVJ CHE
COI/
l
',
TABLE 1. vA[uES OF EXPONENT, FOR THE IMPEILER.SPEED EOUATION, f,, IN SCALEUP BY GEOMETRIC SIMII.ARIW For equol liquid motion or conslont
Froude number, ogiiotor-tip speed, sheor rote or grovitotionolefiects, D
For For equolsolid suspension, equol moss-tronsler rotes or powel
fill a 10-kI vessel via a 5rr-ilia. pipe. Cooling the contents : rr lLil-scaie reactor could require
.1-2 h to
= n=
n=
I 314 213 0
Forequolsurfocomotion, n
Fot For constont Reynolds number or
equolblend lime, n =
= l12
= 2. =3
scale mixing. For instance, for dis persing a gas into a liquid, the, DuE,sug gested ratio of impelier diameter tr ter to tank diameter is 0.25; for bringinl nglng
for
two liquids into contact (as in liq n liquidJiquid extraction), the figure i .rre is 0.4; and for blending of low-viscos iscos'eater ity liquids, it is equal to or greate than 0.6. Similarly, operations that depend on large velocity gradients but low circulation rates, such as gas dispersion, are accomplished at full scale with high-speed small-diameter impeilers. Operations that require high circulation rates are best done with large-diameter, slow-moving impellers.
urface-to-volume ratio
:i,iises as a reactor vessel is scaled r::r For instance, for a spherical ves.,,1. a r.olumetric scaieup factor of
lrt.()00 ri'ould mean a reduction in sur:iicr.-iirea-to-volume ratio by a factor ' r 21.5. Thcrcfore, to input or dissipate the ::i1nro amount of heat, the heat-trans' r i'atc must be increased in propor:r,;n i.r tire. rise in vessel capacity. This ::.er(r{rse can be achieved by providing r'-iriitional heat-transfer area or using : : sher heating-medium temperatures. ll,'rrr'ever, the latter tactic can lead to ,;,,'rcased ski.n temperature and local:
in a small one.
The question arises as to whether the longer mixing period affects the reaction process. A good test for the sensi-
As an approximation relevant for low-viscosity (waterlike) iiquids, it can be said that a mixer-power input of 0.5 to t hp per 1,000 gal of liquid gives "mild" agitation, 2 to 3 hp per
1,000 gal gives "vigorous" agitation, and 4 to 10 hp per 1,000 gal gives "intense' agitation (as might be needed during emulsification, for instance). These figures refer to the power that is actually delivered to the iiquid and do not include friction and gear losses.
elapsed bulk mixing time is to perform the reaction in a geometrically scaled down model of the full-scale vessel, with the
agitation speed the same as in the full-scale one. The results obtained at
these conditions are a good indication of the fuil-scale performance. Here is another example regarding the difference between small-scale and full-scale agitation. In the laboratory, one sometimes agitates the flask (or other vessel) itself, rather than inserting a mixer per se. In fermentation ,\gitation and mixing studies, for instance, most screenings ii aleup of mixing operations involves are done in "shake-flasks" mounted on ,r x"rdr: rang'e of issues. First, rve set rotational, shaking or rocking devices. ,:.,. ..tage ivith some basic points. There is no relationship between the in the first place, agitation must, as "pumping" capacities and shear rates ,; i)nlctlcal matter, differ rvith reactor achieved within these flasks and the , ;rlr, For instance, if we instead in- corresponding capacities and rates ex-ir,,cl on maintaining the same perienced in a fuil-scale fermenter out,.r,1rsr.d crrculation time in the fullfitted with a conventional agitator. , :r jt: \'.:SSl as rve have in a srnaller In short, it is unrealistic or impossir r. \\'i. r','ould liave to have highcr vebie to make all the mixing parameters rr ri,s. because the liquid n'ouid have of the full-scale vessel equal to the in', travei a greater distance. Accord- dividual fluid-mixing and fluid-me; Ii1. f irr\\'e r requ irctnt,nt 1_rel unit .. i chanics variabie in a small-scale tank. :un,,ol.r the fuil scale g'ould have to The eng:neer responsible for scaleup 'rl; in prcportion to the square of must accept this as a fact of life, which ril;rnrcter of the tank. is to be kept in tnind when appiying
Scaleup rules: At least three different kinds of scaleup rules for mixing have been published. Summar)' descriptions of them are as follorvs, rvitir more details available from the references cited:
Scaleup based an similarity concept [8]: In this approach, geometric similarity fixes the ratio of various lengths within the system. These lengths include impeller diameter D, tank diameter T, blade width b and liquid level
scaling up Diameter D 1 to D) under' this concept, the lengths are chosen such that the "before and after" ratio, R, is the same for each pair of lengths:
D2/D
ly';,
of the large vessel and the small vessel are related as foliolvs:
'
ir,rrt -satisfi'ing
that requirenrent
is
' lrnr.rrilr not feasjblt. economically. ,:' r,siilrpi. iithc po\\'er per unit vol':'r
kept COnSlant befrveen ;,-;:ril pilot plant vessel and a 625r,\\'r.rl'{:, tO }tr:
2OO2 47
tional to AFD2; for laminar conditions, it is proportional to N2. I Accordingly, if the power per unit
volume is to be held constant during a scaleup, then N2 equals N/D1lD)a3 under turbulent conditions. And under laminar conditions, N2 equals N7.
where do is the particle diameter, and the power required per unit volume of the liquid is proportional 16 p-{.55. Liquid-liquid emulsion In scaling
p) remains
up Iiquid-liquid-emulsion systems, keep in mind that the shear-rate parameters (viscosity and surface tension)
con-
stant.) The relevant Reynolds number is Duplpt,, where z is the tip velocity (equal to AD) and pr is the iiquid viscositl'. All the scaleup transitions are characterized rvith reference to this
reference rel adionship.
Scaleup u:ith gas-liquid mixing: In specifying rnixing equipment, a major principle is that. the impeller blade must be two or three times larger than the largest bubble, particle or fluid "clump" that is of significance to the process. So, when scaling up gas-liquid systems, be aware that a large vessel tends to have a wider bubblesize distribution than a smaller vessel
sion droplet size is proportional to N-2.56D4.17?1.88, and to achieve uni_ form dispersion U3l, N should be proportional
1o
2-2'15.
[,l/]. Another consideration is that the size of the bubbles should not exceed
one-half to one-third of the vertical height of the impeiler blade. In the presence of gas bubbles, the
required blend time for the continuous phase increases. Accordingly, the percentage of gas holdup is an important factor in scaleup. The linear superficial gas velocity increases on scaleup, and there is usually a greater volume
degree of ,nixing), rvith a given impeller geometr.v, the parameters are related as fojiows:
.^y'
The position of the agitator plavs a crucial role in the formation of liquidliquid emulsions. To disperse higherdensity liquid into a iower-density liquid the agitator has to be located in the latter, and vice-versa, Blending: If two or more liquids witir
being
t]'rose contents become turned over. The vessel turnover rate, which is defined as QN rvhere V is the vessel volume. can therefore serve as a criterion
of gas hold-up in big tanks than in small tanks. So the larger-scale system entails longer blend times. The impeller rotational speecl, N, required for completely dispersing a gas in a flat-blade turbine assembly
can be found from the relationship
blended, so that the process is not affected by liquid shear rates, the difference in blend time and circulation between small and large tank is usuallv the only scaleup factor involved. Hou'ever, if there are large differences in density or surface tension between the
for tire intcnsit-"- of agitation. This approach en-rploys an agitation number, rV7. rvhich i-s scaled between 1 to 10 for rlverage btrlk-fluid velocities of 6 to 60 It/min (in other rvords, mild to violent
agitation ), respectively. For s1'stems that are geometrically
= QeTgo.tp.2D4
g is
the acceleration due to gravity. This equation indicates that the rotational
speed required is directly proportional
liquids, then extra shear forces (and. thus, more input power to the mixer-) are required to achieve the ultimat,e levei of uniformity. During blending, there is a relative increase in Reynolds number ri'ith scaleup, regardless of the relative
magnitudes
lhe
scaled-up vessel
equals
to the square root of the gas flowrate and tank diameter, and inversely proportional to the square of the agitator diameter. Of course, the power redecreases with the increased presence of gas. Solid suspension.. In a suspended-
of the incoming-fluid properties. The time needed for blending, l, is proportional Lo TH|ND2. And
in
geometrically similar tanks. proportional to 1/l/.
is
the torque per unil volume is constant if A'iDl = ItzD,:. Accordinglr', torque per unit volume equals 3.27 Nppl'{ fllg,(HlT)1,
3.S:r2lrttl(DlT\1.5, and
quirement
for agitation
Agitation plus heat transfer In the scaiing-up of manv homogeneous, stirred systems, the rate of
heat transfer controls the design of the heat-transfer equipment. In such
systems, h is proportional
16 Jrnrlnr-1.
where Np is the power number and g. is the conventional conversion factor betu'een force uni[s. And the vesse]
solids system in which the solids account for up to about 30Vc of the total
rveigl'rt, scaleup
keeps the solids
of the mixer
that
rvhere
.\'//l fr,'.t, I] {f P.:'.r1. . Pou'er requirements: For a scaledup reactor, tiie po\\'er requirement per tinit volurne, PN, can be determrned from the agitator speed. For
suspension poses no problem 1121. On the other hand, concentrated slurries start behaving likc viscous pseudo-plastic material, rvhich complicates the scaleup task.
in
/i is a heiit-tlansfer
cor,l'fl
cient, u is the fluid circulation velocit1',.d is the characteristic dintcnsirrn of bhe system (for instance. the agitrrtor diameter) and m is an entpiri-
The minimum impeller rotational speed required for suspending solids, Nrp^, is proportional to d00.2 1D0.85,
scaleup based
on holding the
I
TABLE 2. EXAMPLES OF INNOVATIVE REACTOR DESIGN FOR SCAtE.UPS
Process lssues
. Highly exothermic of benzene lo moleic onhydride . Needs efficienl heol removol . Mold growth smolhers orgonism 2. Pencillin by fermentotion . Need conirolled surfqce growth
i. Cotolylic oxidotion
Methyl lerl-butyl ether from isobulene ond melhonol wilh ion-exchonge cololysl 4. TAX (l -ocetylhexohydro-3,5 dinilro- 1 ,3,5 triozine)
.Exothermic
I '10"C
Stined
vessel Fixed bed Slirred vessel Stined vessel
Chombered multi-tube reoctor with cooling iockei (l9J Submerged fermenlotion lechnology, lor mold lo grow submerged in the liquid food source (l9J Fluidized-bed reoclor to overcome hot spots ond occounl for swelling
ta anl
moleriol
5. Sevofluorone (non-explosive, inholing onoeslhetic) from h e x of luoroisoproponol, HCHO, onhydrous HF, oleum 5 Sullolion ol mixed lollowderived olkonolomides with
chlorosulfonic ocid (a phormoceulicol product)
Exothermic
.Gos releose
Highly viscous
.Connol use onlifooming ogents .Cololysi ottrition due lo ogilotion .Gos leok ol ihe glonds due to
high pressure
Stirred vessel
7. Hydoxyl ommonium
phosphole from
N
CSTR
(see moin
(2t)
HaNOr+H3PO6+Pd+H2
texl)
Gossporged Pocked column {or gqs-liquid
Aulo-oxidotion ol corene with . Exothermic . Low gos-fo-liquid moss tronsfer homogeneous cololyst
'?. Suspension polymerizoiion
. Lumps
formed by feeding inilioled rnonomers (micron-sized polymer dispersion formed !n non-oqueous medium) 3. Suspension polymerizotion
conloct ond exlernol heot exchonger stirred vessel in lhe recirculolion loop. (22'j Semi-bolch Locote monomer feed slirred inlet pipes directly ol the impeller reoclor zone to improve ogilotion (23)
Semi botch
Trcnseslerficotion ol fotty
lmproper ogitolor design leoding lo lormotion of woter droplets in monomer (insteod of monomer in woter), cousing o decreose in heot trqnsfer coefficienl form 50 lo 2 Btu/(h)(fiz)('F) Reduced yield Formotion of oleic ocid droplets in sorbilol, insteod of sorbilol in oleic ocid os desired
siined
reocior
Stirred
botch reoctor
r-t;itt,',, ilt, tt'ltst's ::l()\\'l\' {)\'cf il '. ilr:r 1rr. lirl instlncr'. thc firruta: rri lltrlr lttcrtc or lat'rv dcPosiLs ::.l t.hr heat lran,sler are:r. (Onc - -.i'i, ruLrsc of' snch dcposits is a rii/,r';liulo sr.lnsilir-itl in the mater'l'l.,rt-'. 1i,r lLrt agitatcrl svsl:cnt.
it is olten considered essenrial to assurc that lhe same rabe of hcat transfer is maintained in tl'rc lalgt:-sc:rlo urijt. t,c, avoiil r-nirtct'itrl
material,
dt:gr:rclation. l'rtr Lhrrsc cascs, lr2l/r1 rtrtrsl eqrlal D2lD j, rv[ic]r me:rrts Lhat,
Scaleup can instead be baserd on mainlaining a const,ant imptrller-tip slrt:cd irr thc agitated vessel. In lhet
/r,/h 7 cqual-" (DelD t)"r-t .1'his a1rproach n'ray be rclcvant wl'ren bubble breakagc is an issue.
casr-'.
vessels,
' .
,1.,i
D-i
.. ,j t.,:l) .1 i !
;.,.1\\'t,r
"
"i
a
fiervevcr, this latter relatior.rship indicates thiit the inrpcllcr rotational specd in the largcr r-init should bc ,.qtu'l to {rl {r('.ttcr- tltar tliat rt thr' srniil.I-st:l l': urr rt i clcpr:ncli n g ulton thr: i alrrt' oi lt ). J'huL stipulirtron is conlrarl' lo nornral practice gcnt:rallv, irgita tol s peecl ciecrreases rvi t.h st:aIeup. Furt,irermore, such a scaleup principle is not vr:n' practical (even u'hcrr con-
ibr rnst,ancc. usclirl rr-it lr n()n-\\-\\'Ioniill] flLrid-.. gu.rriitit(,r.: -.inrilur nrappirlg-q (fron small to
scirlcd-u1l vc-sselsi of lht: r'atios rrl local to averagc rate-q olencrg-'- clissiJration. As a rtrstrlL, A-.2/AIi ttquals \,D ti Dri-.
'' ";.; '. ', t'til.'l','rr'''l pt't'rttiil , i,,,r,,,111'. :(.irlr,(l LII) ilt ltrrllti'r 'ii, .,:r{'llttltl l.r,r'.\'r.l' (ll tirt irgi'i:,rt-'r,1,
l
lltc grc:rlcr' 1)l' r'riirtl lr)tirti,)lral spt'r,d irr thc, sclrlt,rl'up vlssr,l, along * itir thr,
sidclt:ri esscntial
). ber:ausr:
2OC2 49
. Botch uersus continuous operafron: Although the laboratory rvork may have taken place on a batch
scale, a continuous reactor (such as a
simple one. And when series or complex reactions are carried out, the product distributions for the two op-
Fluidized-bed reactors
Because the circulation of the catalyst
tions differ.
If a tubular
reactor is
chosen, its diameter affects the radial and axial temperature gradients, which can introduce heat-
the
be
transfer and
challenges
flow-distribution
in
cases
invoiving a single reaction step and requiring only a short residence time
o
Temperatures, concentrations,
residence tim-es: Temperaturc and conccntration gradients, as rveli as residence-time distributions, will not necessarily be the same on a larger scale. The differences may adversely affect the yield and product purity. Such considerations may
pansion and catalyst attrition must first be studied thoroughly. When one scales up a fluidized-bed reactor, it is not suf{icient solely to understand the effect of increasing the vessel size _- the interaction of the
flow conditions (ust as stirred-tank reactors operate under weil-mixed isothermai conditions). But it will not be so in the full-scale plant. The temperature and concentration gradients can never be made to be uniform, which leads to a difference in effective heat conductivity. For constant fluid velocity; the reactor throughput increases as the square of diameter of the tube. But with the increase in diameter during scaleup, temperature-distribution effects wiil be more pronounced. With exothermic reactions, the center of the tube will experience the highest temperatures;
tion or removal. Coils are attractive rvhen a high heat-transfer area or coefficient is important; conversely, coils should be avoided and jackets used rvhen, for instance, the process liquid is highl"v viscous or prone to cause fouling, or contains a high
terstitial gas dynamics []41. In addition to the maintaining of geometric similanty, scaleup of a fluidized bed requires maintaining h1,drodynamic similarity. This latter similarity rule requires that the foiiowing trvo conditions be satisfied:
(U o
- U *l
= mu2(tJ u - U rrii
and
percentage of solids
U^f = mu2(U
^)i
with endothermic reactions, the temperature there will be the lowest. The resulting radial temperature gradients, like hot spots, can cause loss of selectivity, coking or fouling in the catalyst bed, and, sometimes, mnaway conditions that may lead to unsafe situations or destroy the bed. In pilot plants, small catalyst parti-
is the scaleup factor, and Subscript i represents the base-case (before scaleup)
condition. The first condition assures
geometric similarity in bubble coalescence; the second assures the geontet-
n'ide range of catalyst-handling issue-s can arisc that are not necessarolrvrous at Lirc laboraton' ,sc:tle bnt can afferct the catalv-st activity. lif'e and regenerability. Exarnples include the catal5ust shape and dirnension-s (for heterogeneous catalysis),
of the necessary and sufficient, conditions for hydrodynamic similaritv. F'or scaleup and development probk'nis. it is suggested that both the hydrodr'namic-similarity rule and the identitr' rule, described in the next section, bc applied. Identity rule: The identitl, r'ule dcals only with increasing the diameter of a fluidized bed. It is especiailv useftrl f'or
ilv
cics are generally preferred; for one thing, they minimize mass-transfer
effects when developing the rate equa-
Also to be considered i-s horv thc catalvst is charged and di-"chargecl Stirred-tanh ler.su.s tubul.ar re-
actor: ])er unit of volrrmc, a fullscale strrred tank is less expensive than a tubular (nominally plug-flor,r') reactor. Horvever, the choice is not a
sludying chemical reactions; that is, for cases in rvhich the fluidize'd particles consist of catalyst.
According to this rule, the catall'stparticle size, bed height and gas ve-
50
Process
i. Polybutylene
terephtholote from bulonediol ond cotolyst
polymerizotion of
styrene with butyl
methocrylole
TABLE 3. SOME ISSUES ARISING DURING REACTOR SCAIE.UP Observqlions during scole-up Setup lssues . High heoting role ochlevoble ln smoll . High eslerificotion Stined (polymeizotion) rqte will scole, not in lorge scole, due less heot-lronsfer oreo vessel . Heol buildup in loler stoges per unit of reoctor volume . High ogilolor tip speed in lorge scole leods lo polymer gives more of degroded producl (24) degrodotion . Mixing k1o does not depend on reoctor size but is proportionol Airlifl to U,ro. no* 0.86 (25) reoclor . O2-lronsfer tote lo liquid .lncreosing reoctor size leods to Slined Moleculor weight ond - increose in moleculor weighl ond polydispersion polydispersion vessel - increose in initiotor quontity to mqinloin some Porticle size moleculor weight Auto-occelerqlion - longer time needed to ochieve oulo-occelerotion behqvior
(26)
4. Suspension Stirred polymerizolion tonk of vinyl chloride 5 Finol slep to vitomin 86 Stirred (diozotizotion of nitro lonk to omino compound) ,5 I'Jeutrolizing on ester Slirred prepored in dilule, lonk olkoline, olcoholic
solution using ocid
. Size vories with rpm -0.e, height-0'I3, ogitotor dio. -0.e . Minimum rpm vories with ogitotor dio. -l.l
(27)
. Lorger liquid surfoce leods lo higher liquid evoporotion rote ond chonge in concentrotion
(28)
side product
. Longel oddition time of ocid leods to reoction between woler ond ester
/ lieutrolizotion
o{ lobile
Slirred
lonk
posed product, due io high concenlrolion grodienls Use of No2CO3 os neulrolizing ogent ond increose in ogilolor rpm, to decreose pH grodienls Addition of olkoli to oqueous-rich porfion leods lo
dilution [28)
3. Biochemicol reoction
using onimolcells Slirred
Teoring of bio-orgonism
. Use of shoker
lonk
9. Production of
surfoclonl using oir for oxidotion
Stirred
botch
with oir
bubbting
l0 Gos-phose oxidolion
I
CSTR
I Three-phose
equilibrium reoction
rvith 45% solids
CSTR
12 Fine-chemicol
Stirred-
tonk reoclor
Stirred-
tonk
reocior
High tip speed in ogitoted vessels leod to teoring ond breokup of cells a lncreose in viscosity during reoction prevented N2 in Foom oir frorir escoping; use of puie 02 obvioted this prob-lem Excessive heot generotion a Decreose in heol-tronsfer coefficient due to foom lncreosed botch time increosed the bolch lime (5) Use ol diluents con reduce reoction role Heol removol Decreose the process volume Moinloin the low conversion (6) Use of N2 sporging lo drive the gos (2JJ to inodeLow conversion due quole removol of second goseous product, NH3 . Use of second impeller lo oller the regime from lnsutficienl liquid droplet coolescence lo dispersion-controllin g (23) sheoring, leoding lo coolescence ond formolion of lorge droplets . Locote diluent oddition neqr the intoke of Non-uniform, brooder size impeller to ollow crystol growth rother distribution in lorge scole thon nucleoiion to dominote f23l
scole
rr'. nrrrst all be hcld constant dur.';1lr,i1r'r Srr ntrrst the clintensions r l i. ii,rv, -rlistriltLrtion cquipnrent.
r; ;r: ljrr,,rrifjce rlirtnrr;tcrs. and thc
bot.h the small and the largc scale. The ir-rcrease in the capacity of the svstem is purely through the increase in t,he cross-scctional area of t,he {luirliz.r'cl hrrrl
.
Ol
H,/II, = 7.,/7, = prlp, = d rlcl, ri'ltt'rt, tiit' Srrlrst'r'rpts s ;rntl i t'r,i)r'('s('n1 1ltr, slrr;ril- rrrrrl lru'rlr'-srlrlr, ','r,:-st'ls. r-cs1u:ct ivclr i''or lttirining 1hc gconrt.tricaIlv sirniiar {luiciiz.ing conditron bct,n'r'on ll,-,ds of different scaies, the similaritv nrle also reqrrircs t.hat both the fluidizing.
Sirrri/rrrilr' r'ttlc:'l'lris rrrlr' nt'r'liitt'ts tcr itltilllurlq gt,onrctlicitilv stntilat' flLridiz.ing cortclitions betq'ccn lteds of dif:
:' :r',:
'v' jil iltrrittr' 1()r olh('r qtar:ttttetr:r i,, ilrr', loncituclinnl clistribtr',1' :
i:
g an rl tltt-
il
\'(\rti
g,-(l
'
rr)rrir.lr- cortstirnt. ;tnrl llrcrt:'::,, lri.rrtr,:rl rr';rc'tion lrt'hltvirtr 1,, ,,r'i;,.,'ii,tj 1l lrr. irl,'ltiir:;rl ()n
gils vr:locitv ilncl thc minirnunr fluicliz.ation vt locitr, lte variccl prclpor'tionatclv rl'ith thc sq\lelc root of thc
2OC2 51
scale ratio.
A consequence of this is that the sintilaritv rule does not reqtrire geometrical similarity for the
particies themselves.
resistance. The modifying effect depends on reactor type and size. There-
solids. Based on
N Dru
12
in the scaleup.
Three-phase contact is achieved in either a stirred slurry reactor, where mixing and solids suspension are
d pu
6) =constan r
Stirred-tank reactors For scaleup of a continuous stirredtank reactor (CSTR) with multiple impeilers and a large height-to-diameter ratio. studies have shown that the ratio r of the internal circuiation rate to the vessel throughput q should be
or
trickle-bed reactor, where the catalyst is held firm in a packed bed while the liquid trickles down the column slowly and the gas rises from the bottom. The slurry option is the more popular.
In general, when geometric similarity is maintained, lower stirrer speeds are needed for suspending solids in larger tanks. However, small difference in the exponent on b lead to large
in
r=Qlq
Q is proportional to l/D3. The ratio of pou'er in sniail and large scales for geometricaily similar systems (that is, D is proportional to T).and equal den-
slurry reactor is that the solid is homogeneously dispersed in the liquid. In developing and scaling up this type of ieactor, the most critical elements are understanding the hydrodynamic issues, the liquid circulation pattern and the slip between the
solid and the liquid. The size of the gas bubbles depends on the dynamic equilibrium between
differences in the power consumption per unit volume. For given impellersize ratio and clearance ratio, the cntical speed is almost the same for turbine and flat-paddle impellers but
as
Biotechnology-process scaleup Scaleup of agitated-vessel processes that use animal cells poses special problems. For one thing, the tips of larger agitator blades tear at increased shear rates. Conditioning the cells to withstand the required sirear,
rically similar systems and equal mean residence times in both the
scales, the vessels throughput
will
be
related as,
i'q1q)r/3 = TJT,
These trvo equations lead to P,JP. = (T,J'l r\L =
(q
or encapsulating the cell organism in or on micro-scale particles, are trvo of many strategies being tried to sidestep that problem. Both the shear rate and the circula-
Accordingly, a tenfold scaleup of the tank diameter means an increase in power requirement of about 105 for
bubble breakup is promoted. The holdup of solids-free gas increases with increasing reactor diameter, and liquid holdup correspondingly decreases. The liquid holdup governs the
Successful scaleup
has
liquid residence time, which is the principal scaleup parameter for predicting product yield. The bubble
shape and prevalence, and the resultant rise velocities, are also important fbr reactor performance.
been
achievr.d for the continuous biending of loiv-viscrisitv fluids using the criferir.rn t,hat q/Q is held consLanL 1171. This crirerion rs similar to the one discussed earlier in this section. The relationship PJPr= (?"/?r)5 has Lreen found to be valid for a continuous
stirred-tank reactor by other authors, for rrierrtical reaction yields and mixlr){ lr lnts iilld t'ortstitnt, turrtovcr tintt,
lt
G a s/Ii q ui d/s o I i d:re a cto rs 'llhe conver-sion, selectivity and stabilrt-v of catall'zed reactions in fluid-solid reactors can be significantly modified l.r',' the iniluence of physical transport
speed, N, required for lifting the solids in a siurry rector is inversely proportional to the impeller diameter, D, raised to the 0.67-0.85 power. But to obtain a satisfactory degree of hornogcneity, the particles must not only be lifted {r'orn the bottom buL also carried throughou[ the volume of'the vcssel in sufficient quantities. A relevant relationship is:
NDzlde = constant
The stirrer-blade thickness, b, also
has an effect on the suspension of the
2OO2
52
COM JANIJARY
,rrr,ilt,rl into. ibr instancc, tubtrliir, :,, r j-r'olLrntn, -qtatic-mixer, bLrbble, rr n or {luidized-bcd reactors.
require-
r-(.cvc'le loop-", cooling jackets r':l i hr: ir:ngth of tl"rc reactor tube, ,,i l'ir.r nlc,a.ns. And 'n'arious impellcr -:rn-( irave been used to improve
\.
tie-q, gas, liquid and solid holdups and h1'clroclynamic regimes, have to bc rneasurcd at the larger scale. Arriving at a mathernatical modcl that describes the process reasonably well in the large scale is important for designing large commercial pro-
'.,lirrn vfficiencr,. -,, iltstructive, by supplemcnting , r-1t(,1r,'-r g(It(lral guidclincs, is a -. :1 |;irl11'rrlar issues tliat can arise ' :.,, :ciilc,ults of spccific proccsses =.',tiii;0ns. F-or i:itl arraV of thenr
cessing units. This is feasible only rvith a correct model and good estirnates of the constants. I Edited by Nicholas p. Chopey
Author
f)oblc Illukesh is a proggam lcadcr in the Chcmical Engat the John F. Wclch Technoi og'y Ccntre' ol GFI India 'l'ech-
, :l,l,.li
in
. ,: r- scrrlc.lrlt r,,ffort.
.'l ir
rrl thoughts . rr('c('ss of any' commcrcial opera' i, ;,,'r'rcl. ,rn rrnderstanding the ef' :i.ri,,irp orr thc process. Unlil re: -. .,'r,t'\' litt.lt attention was paid ,,, ,:ir is an intcrclisciltlinan.sci-
Pvt Ltd. (Sv l52, F,xpnrt I,romoliolr Irttlrrstri;rl I'rrrk I'lr:rsr' 2, Iloodi ViIlrrgr,. \\,hitcfi('ld Roird. Ilangllorc 560066, India; Phonc: +91-80-U412050; cmail: Mukesh.doblc @gcind.ge.com) Previously, he u,orked for l6 in the ICI India R&1'Centrc. His areas of' .r,ears cxpt-.rtisc afe in modeling and simulation,
procc-ss dr:vclopntent,
rt'i'r rnconlpilssos
ginecring, N{ukesh has prrbiisherl ncarlr,g0 tcchrtical ltapcrs, as rvlll as six jourp:rl liticl,-s 61 1l&D rlnnag(-'nl{'lit. irnrl is co-iirrl}rot. ol't.hc lrook. "lIorlr'g,,rr,_'orrs ('atal-r.sis: Nlcclrrrrrisln irn<l In-
scveral
drrstrral Applicntions" (\\'iler' IrrtetrScicncc, !000i. In 1{){.16, lic rcct,ir.od t}tr. Inclian Ir.rstitut,c ol'Chcnrical lingrnccrs' ilcrdillia An,3;11, frrr cxcclL-.ncc
"
in basic r.csc:rrch. [ie holds i].J'cch. and Nl. Tech degrces in chenrical enrinccrins from IIT, Madras, and a Ph.D. in that"ficld fr6m thc l-lniversity of A-qton in Birmingham, U.K. He has carricd out post-d-octoral research itt Cambridge Univcrsit.l' in thc U.K., a-s '"vell as aL Tcxas A&-li{ l]nivcrsitr'. Ilc is a ntcnrbcr of AIChE and of the Indirrn Insl ittrtc of Chcnrical Engincr.rs.
For over 8 years, Liqui-Cel@ Membrane Contactors have been used in a variety of installations for COz and Or removal. They have a track record for removing other gasses as well. To date, membrane contactors contintJe to degas more than 300,000 gpm (68,180 mr/hr) of water. They are proven in the field and offer many advantages over conventional technologies. Liqui-Cel Membrane Contactors
are:
'I
t.l't'Ilc('s
'' :;. .J J' . oiirr'r-s. -Scelc-irp \{t:thoiirilogr. :. r:tr,:,. J'r,,cr.s.,,". (irrll' I'rrl'lislrirrg i 1,, r:t r,:t. l {19, j ., ..r: it . r'l:,','ri. Ir,ru'. /Lrg.. \rtll. [)2. No. 2. p.
l"it' '1',"rr.
l4 llrrrio,
1973.
9, p.
'
1ir. (kilrLstein, A.
. Compactp. 1021,
vacuLtm towers
,. i r ); q.L;r,.. I )., ]-l,lrp F rrrnt i.i|tt tri I.'rrll-sr.:tlr, -: il. t'i ,.r,t !..'tg.. ir i i(). \rl'i'ntlx.r ll{ljl ' r- ' " .l I j i:ircr,,.,.rj lri li,'lcirrr Sr,lrlr, 1lir. i-r.g !',,1g. rr ii..lur"ri, l(X)r) . .,..,::,i. Il.. i'rrtrinll \rrnrllcrs orr inirt'r.t'nt , . . r 'l;, ,' /'-'rg . p s2. \larch 19ti{j 'i \' ('!:' rttti'clt \''rl. 11 No. 9. p. . .
.lulr lll1ll
20. (lonvcrs. r\
':,. l) . ('ltt'trtl,'r'lr
. Modular'
sive
95. 1981.
,,:
21. I\lan'n^o^rvski. C. '{I., (l/rcnrlcr'lr. \/ol. 17, p 560.1987. 22. Xlrrkcsh, [J.. othcrs, (.thcnt. Ettg. J \/ol. i18, p
. Warranted to I ppb of O,
and lppm of CO,
To learn more about Membrane Contactors, call us or visit orir website at www.liqui-cel.com.
15:t I9S8
\.ol
,q't.
\,r
7.
1;
Lr:rrg.
I 1|{r
I)
1,.. ('lt,:rrt.
l)rtg Prog, yl
2i3. ,lrrne
'rl
",'a I .\ . i'
,
I,
1
. ,,llr,'r':. ( ,i.i'r:t l-rrg. S,'r , \'ol -.ti:,. i1)e{j. t'..'i"]Ln,l )lrxtng'fcrlhn,'logr'." .\',i\ )r'ti.. L(t\,)
/lr'.s
&
1)('r,
(ELGARD
.
lllohrontol
; l. I ii!,()
\/ol.
20. No. 4. p
che.com/adlinks or circle 36
Cllt-l'/lCAt [:NGINEERING WWW CHE.COM JANUARY 2002 53