Beruflich Dokumente
Kultur Dokumente
Models a r e defined for various mixing conditions, Les auteurs dCfinissent des modkles pour diffb-
in continuous flow systems. Differential equations a r e rentes conditions d’agitation clans les systgmes ii
derived which take into acrount an effective volume bcoulement continn. 11s dkduisent des Cquations diffC-
of mixing, possillle short-circuiting, holcl-up time of rentielles qui tiennent compte tl‘un volume effirace
the system, partial displacement or piston flow. cl’agitation, cle court-circuits ponsihles, du temps de
The values of the different factors contained in rCtention du systkme et de I’kcoulement par dbplace-
the integrated equations can be determined experi- ment ou “frontal”.
mentally by the particular response of a given system Leu valeurs des diffCrents fiicteurs contenus clans
to a sudden change in composition of the feed. les Cquations intbgrkes peuvent &treolltenues ii partir
A correlation of the effective volume of mixing de la rCponse pariiculi&re cl‘un systcme donnb ii un
and the agitator r.p.m. is presented. cliangement sullit dc la composition cle l’alimentation.
A design procedure is suggested for tank flow Les auteurs prksentent une corrklation entre le
reactors when partial mixing occurs. volume efficace d’sgitation et la vitesse de l’agitateur.
11s proposent une mCthocle de calcul pour les
rCarteurs en rCgime continu lorsque l’agitaiion est
M IXING, which has been referred to as one of the most
intractable of all the unit operations of chemical
engineering ( I ) , has been the object o f niorc and niorc
partielle.
fundamental w o r k over the past years, partly perhaps on geneous, whcrcby sonic of the molcculcs in t h e feed
account of the recognized complexity of the subject (2). streaiii appear aliiiost iiiiiiiediately in the effluent as a
Many of the investigations in this field have dealt with result o f good agitation. T h i s may be confusing bccause,
batch systems and with the power requirements for ‘is pointed out t)y Colburn (‘I), the tern1 “short-circuit”
mixing ( 3 9 4 . 5 - 6 ) . T h e ever increasing conversion to con- should refer rather t o the flow pattern in a tank where
tinuous-flow processes in the industry has introduced some of thc fced is channeled to the outlet, reaching the
various other aspects to the problcm for others to esplore. latter \vithout being mi\ed. I t latcr occurred to Mac-
Among the numerous avenues of approach followed, a \lulliri ( 1 2 ) that short-cut o r channelized flow between
few can be found dealing with the efficiency of niixing. inlet and outlct would indeed alter conditions in a tank.
In spite of all the u o r k done in this particular field, much He suggestcd, rathcr qualitativel!?, that it could perhaps
more remains to lie accomplished as indicated in reccnt
be possible to evaluate the eficicncy of a stirrer b y
chemical engineering works ( 1 , 29 7 ) .
stud!.ing the conversion rate of a reacting system whose
T h e present investigation was undertaken with the kinetics are known,
purpose of providing a theoretical basis for evaluating
the performance of misers. T h e method dcveloped for Statistical methods have been suggestcd to determine
measuring mixing efficiencies applies to continuous-flow blending efficiencies (*I) f o r batches of varying concen-
systems but it is intended to extend it to batch systems as t:-ation fed into a stirred tank.
well. After presenting a quantitative treatment of various Danckacrts ( 1 I ) h a s introduccd the concepts of ‘hold-
factors likely to be encountered in actual mixing, an back’ a n d ‘segregation’ in order to conipare actual to ideal
experimental study relating to part of the theoretical iilhing conditions in continuous flow systems. T h e con-
investigation is exposed. cept of hold-back relates t o the residence time of various
elcmcnts in the system; the author cautions: “the whole
Previous work
F-diagram (and possibly other inforination) is needed if
Machlullin and ’IVeber (8) were among the first to the actual performance of the reactor is to be calculated”.
present an analytical study of the behavior of perfectly T h e notion of real ‘short-circuit’ is mentioncd b u t none
mixed systems, f o r continuous-flow niising vesscls in series.
o f the I; diagrams shoivn take it into account. As to the
T h e y , and others ( 0 . l o ) , used the tcrni “short-circuit” or
“mathematical by-passing” to dcnotc conditions involving concept of ‘xgrcgation’, also obtained from F-diagrams,
it is introduced to give an indication of t h e efficiency
probability considerations, applicable to well agitated
tanks in which the contents are substantially honio- of mixing in a system. LVhile it points t o the degree of
................................................................................. departure o f a system from perfcct mixing, it does not
indicate fully how the syhteni actually behaves. T h e H
1Manuscript received J u n u u q 2, 1959.
2Dspomnent of Chemical Engineering h v a l University Quebec, Quc. and S values thus defined involve rather tedious deter-
Cmt&ution from the Departmsnt 01
Chemkd Engftddng. L a d Uni-
minations besides being questionable as to their real
advantage.
_-9.c, 0
0 05 10 15 0 05 I 0 I5 20 i
Y
Figure 1-Perfect mixing : the feed is dispcrsed instan- Figure 2-Partial mixing and short-circuit: a fraction of
taneously. The composition of the outlet stream is the the total volume is perfectly mixed; part of the feed goes
same as that of the tank contents. directly to the outlet.
Factors involved feed, or all of it, may give rise to piston or plug flow.
W i t h t w o niiscible liquids, in a batch process, a lnasniuch as the liquids in the stagnant and well agitated
perfectly homogeneous mixture can be obtained when zones are not separated by a membrane, fluctuations or
agitation, with the help of diffusion if necessary, is main- circulation may arise. Liquid in the stagnant zone may
tained for a long enough period of time. In a continuous be replaced b y some from the well agitated zone or f r o m
flow systcni, however, it may be that perfect mixing can part of the feed. Such a circulation may go on steadily
never l)e attained. T h e r e can be stagnant regions in the in the form of piston or plug flow.
reservoir, part of the feed may go directly t o the outlet, Nagata and others ( 1 7 - 1 8 , 1 9 ) have studied mixing in a
piston o r plug flow may take place. In order to study the tubular reactor where piston flow exists, but in t h e
behavior of such a systcm and the efficiency of mixing, absence of any short-circuit.
one can consider the tank or reservoir to act as a damper R4ixing prot)lems met in actual practice may involve
to variations in feed composition. T h e mathematical different combinations of these three factors: effective
treatment of this problem thus becomes similar to one volume, short-circuit, piston flow. Some particular com-
performed on a capacity in a servo-mechanism circuit, binations, more likely perhaps to be encountered than
where a transfer function exists, from which can be others, will now be considered.
obtained transient o r permanent responses.
T h r e e factors that may be met in actual mixing con- Theory
ditions are taken into account. As for a servomechanisni element, the behavior of a
( a ) Effective volume mixing s)rstcni can be studied b y means of the answer to
For a given agitator, part only of the total tank either a step, a linear or a harmonic demand. O n e only
contents may be well mixed the rest being taken u p by of the answers being necessary to determine a transfer
stationary or stagnant zones. T h a t part of the total function, thc first or step demand has been chosen o n
volume where mixing is complete will be referred t o technical grounds suggested b v the experimental deter-
as the ‘Effective Volume of Mixing’: it is the part which minations to be niade. For the different systems con-
acts as a damper to any variation in feed composition sidered, the following are the answers to a negative step
and which is considered in determining the transfer deniand.
function. A sorneivhat similar concept has been used b y
I . Perfect mixing
Gerniain (1:) for evaluating contact time in a tubular
reactor: the author replaces the reactor volume, in which T h e contents of a reservoir, Figure 1, occupy a volume
the temperature is variable and passes through a maximum of V cubic feet; the initial concentration is c,, Ibs./ft.s.
of T,,, by a smaller ‘equivalent volume’ in which the q cu. ft./hr., of composition cb., are assumed to be fed
teniperature is uniform, at T,”. continuously into the reservoir. T h e feed composition cB.
is changed suddenly at time t = 0 from c, to 0, creating
Macllonald and Piret ( l o ) have observed qualitatively a step demand. A t any given time t, mixing being perfect,
that dead volume can retain solution f o r indeterminate the tank contents are uniform throughout and of com-
periods of time within a vessel. Danckwerts (14) also position c, which is also that of the effluent.
considered dead space in a mixing tank when introducing
the concept of hold-back. A material balance f o r a time element d t gives:
( b ) Short-circuit q c d t = qcdt + d(C‘c). . . . , ., , ..,...(I)
Depending on the relative position of t h e inlet with T h e feed composition, cF, being then zero, and t h e
rcspect to the outlet, part of the feed may go directly volume of the liquid in the reservoir being constant,
to the latter without being mixed, causing channeling or Equation (1) can be written:
a short-circuit. O n l y the remaining portion is thus avail-
able for mixing. T h i s situation is unlike that arising from
the definitions of short-circuit introduced b y some
authors as mentioned already.
U p o n integration one obtains:
Macl)onald and Pirct (In) report that they have ob-
served channeling, in the course of experimental deter- - _9
minations, especially at lour levels of agitation. V t
c/c. = e
( c ) Piston flow When plotted o n semi-log paper, as in Figure 1, this
In many systems, depending on conditions, part of t h e equation gives a straight line.
t_!
................
-.iir----
outlet would remain constant a t c, for a period of time
t = (1 - ni) V/q, which represents the time required
07
04
~ \
1 to displace all the material out of the stagnant zone at the
prevalent rate of flow. T h e variation of the outlet com-
position with time, in either case, is obtained in the
following way:
q c d t = qc'dt + mVdc'
dc' q
dt
- + -rnV c'=O
I I
.....................
10 15 20 _.25
-
"1
V
(1 - m ) V
Figure 3-Partial mixing and piston flow: a fraction of Rut, c = C' a t time: t -
(I
the total volume is perfectly agitated; piston flow takes
place before or after mixing, as shown.
l'cl
I n c/c, = I n n - - t
mV
Thus, plotting c/c, vs qt/V on semi-log paper as in
Figure 2, the value of n can be read directly on the con-
centration ratio axis, a t t = 0. T h e straight line has a
dope of - n/m, from which can be derived the value of
rn, once n has been determined.
3. Partial mixing and piston flow
W i t h a fraction m of the total volume well agitated,
the feed can be considered to give piston flow in two ^,l . . . . . . .* . .
different ways,' as shown in Figure 3:
(a) T h e feed is introduced into the zone of perfect
mixing, displacing an equal volume into the stagnant
I h r i n g this period, niatcrial o f composition c,.. = c,, Plotting c / c , vs q t / V on scmi-log paper, for the two
cntcrs the m i i c of pcrfcct mixing froni thc stagnant zone. rcgions of casc A, givcs curvcs such as shown in Figure 4.
T h e folio\\ ing rcl:itions are o1)taincd: ( R ) P a r t of the fced goes to the stagnant zone and
+
ll‘{(.l.(ll p < l ( ~ , , l i l= (1) + 11) q(.’tlt + lIl\’(iC’ displaces an equal volunie 1)). piston flow around the zonc
(It.’ (1) + II) ‘1 c., = 1”1”<>
o f pcrfcct niidng, directly to the outlct, as indicated in
Figure 5. T h c outgoing material is madc up in part by
(It Ill\. Ill\‘
this strcani, in part by liquid forced o u t of the zone of
+
(1 - ni)V
4- , a) 0 <t <
- (1,
~~
11) II I’q
P GV
c’/c,, = .~
1’ + I1 + 1’ 1; II
c = (1) + II) c‘
-~
(11 i n) ti
Ill\’
c‘/c, = e
Now,
c = IlC’
--In\’
nq
nsq
&& c‘ = c, at time t = O
1 hTow, taking account of the reactant in the feed and
K2 = ~
(s + I) (1 - Ill)
.-
1 1 - 111
srn
( ‘ L +
1 - 111
c/c, = _ _nsz
l l [ ( s IMll
___ + _- 1 ~__ +
(1 s)2+ 1 + s
f
I
~--
+s 111
~ ~~
Ill\;
n
&
INLET
i
VARIABLE
SPEED
REDUCER
3
I
-
Figure %Experimental apparatus.
t MIN
01
0 50 100 150 200 250 300
RPM
100 150 200 300 400 500 700 900
R PY
Figure 1LShort-eircnh effect vr. agitator rpeed. Figure 13-Variation of m and n with agitator rpeed.
~OIY produced by thc introduction o f the feed. Such an \vhcLii a chemical reaction is prcsciit.
c , : initi;il coiiccntr,ition.
effect appears to have k)ccn nicntioned first b!~ VacDonaId c' : conceiitr;i(ioii i i i the LOW of pcrfvct mi\iiig
Md Piret (I"), \\.hen measuring niixing times in a vessel. I<, K 1 ,I<;, I<?, I G : Colistants of integration.
. AS the value of ni reaches 1, when the agitator speed 111 = Fractioii of t h e t o t ; i I voliirni~which is prrfrctl!. m i u d
b increascd, a niaxinium value of (111 - tiil,) is reached. i n , : value of 111 whcii thc ;igitator is a t rvst.
This vduc remains constant thereafter even if the agitator II = Frartioii o f thc fcccl entering t h c zoiic o f p ,rf-ct iiii\iiig.
: value of II whcii the agitator i \ at rest.
peed is increased tjccausc the system behaves as one 11,
N = .Agitator spcrtl, l<.l',hl.
g~vingperfect niixing.
p = Fraction o f the feed going t o the stagiiaiit z o i i ~ * .
Procwding in a similar \vay, the values o f (n - n o )
I
(1 = Iiatc of flow,f ~ . ~ ' h r .
, 01 N were plotted on log-log paper as sho\vn in Figure
r = lieaction rate, llis.,/(ft.3) ( h r . )
12, 'giVin$ a straight line o f slope 3. T h e values of n arc s = Fraction of part nq of the f c w l eiiteriiig t h e z o ! ~o f pvrfcrt
Gim d a a d Lo the agitator r.p.ni. by the equation: inixiiig which sul)~eqiieiitlypasses o u t into th,. s t : ~ g n a i i t
-
n n,,= I) N:' zone.
t = 'Line, hrs.
wbere, for the p:irticular s>,steiii studied, no = 0.77 and (1 - 1n)V
. b5 2 x lo-". ti : ~~ ~
nsq
'',
When plotting, on a linear scale, the values of 111 and v = Volumr, I t 1
9 PI 8 function o f S, as in Figure 13, both variables arc
' , f m d to reach 3 niasinium value o f 1 at an r.p.m. o f Re jerences
proximately 220, indicating that the system behaves ( 1 ) Coulson and Rirh:irdson, Chemical Engineering. Vol. 11, p. 917.
M as one giving perfect mixing. McGraw Hill Co. (19,55).
( 2 ) Perry, J. H., Chemical Enginerrs' Handbook, 3rd Ed.. p. 1195,
hfcCrew Hill Co. (19.50).
Co~lurbns ( 3 ) Rushtiin. J. H., Ind. Eng. Chem. 45, 9 3 ( 19.53 ) .
"
The method c1cscrit)cd to determine mixing efficiencies ( 4 ) liuahtrm, J. IT., Ind. Enp. Chem. 46, 133 ( 1 9 5 4 ) .
h COntinrious flow s y s t e m should prove to be a very ( 5 ) Ru\htoii. J . H., Ind. Eng. Chem. 47, 582 ( 1 9 5 5 ) .
( 6 ) Riishton, J H., Ind. Eng. Chem. 48. 552 ( 19.56).
:useful ta)l in evaluating the actual performance of ( 7 ) X.lcC;ihv .ind Smith, Unit Operations of Chemical Engineering,
ea Not only is it possible to obtain optimum results p. 292, MrGmw Hill Co. (19.56).
(8) hlacMullin, H.B., and Weher, H.. Trans. A.1.Ch.E. 31, 409 ( 1 9 3 5 ) .
,:withigiven mixer h i t the relative performance of mixers ( 9 ) Drnbigli. K . G., Trans. Faraday SOC.40, 3.52 ( 1 9 4 4 ) .
id m e r e n t design can be estaMished since efficiency 1 0 ) Weher, A. P.. Chem. Eng. Prop. 49, 26 ( 1 9 5 3 ) .
imcom,remcnts arc possible. 1 1 ) Colbiirn. A . P , Trans. A.1 Ch E. 31, 457 ( 1 9 3 5 ) .
4.. The COntribution t o the design o f tank flow reactors 12) M,ichlnllin, R . B., Chrm. Eng. Prng. 49, 33 (19.53).
1 3 ) Beaiidry, J . P.. Chi-rn. & Mrt. Enpr 55 No. 7, 112 ( 1 9 4 8 ) .
Add &o be o f iniportance because a method is now 14) I>anckwc~rt\.P. V., Chrm. Eng. Sci. 2. 1-13 ( 1 9 5 3 ) .
+lablt to take into account the effects o f partial mixing. 15) Germ,iin. J E., Bull. Soc. Chini. France ( 19.57) 484-5.
l $ n p v e d designs, n i t h the Ilcst possible use of agitators, 1 6 ) hlacDonald, R. W., and Piret, E. L., Chem. Eng. Prop. 47, 363
( 1% 1 ).
i h d tCguIt.
1 7 ) N.igat.i, S., Epuchi, W., Inainiira, R . . Tanigawa, K., and Tan&q ,
)
i+
I
The
cxpcriniental results obtained to date arc very T., Chcm. Eng. (J.ipan) 17, 3x7 (19.53). . ..
, l\'hilc they do not draw a coniplcte picture 18) N ~ g a t a .S.. Eguchi, W., nnd Tanaka, T..Chem. Eng. (Japan)
163 ( 19%).
la,
'
to the liinitcd extent o f experimental work 1 9 ) Suzuki, E.. and Saito. J., Chem. Enp. (Japan) 21, 91 ( 1957).
t)c perfornicd, they point to a number of 2 0 ) Smith, J. M.. Chemical Enginrrring Kinetics, p. 172. 17.5, M c d
Hi11 Co. (19.56).
h t s t i g a & ) n s v.hich should lead to a coniprchensivc
Eaowledgc of niixing. * * *