ow%2509/78/0701-o!u5/mm
MASS TRANSFER PACKED WITH K Department
of Chenucal
B
IN BUBBLE MOTIONLESS
WANG?
Engmeertng,
Kansas
and L
T
COLUMNS MIXERS
FAN
State Umverslty,
Manhattan,
KS 66506
U S A
(Recewed 16 June 1977, accepted IS September 1977) Abstract-The cocurrent upward mode was employed to absorb pure oxygen mto water tn bubble columns packed with Koch (Sulzer) motionless rmxers The hqtud-side volumetric mass transfer coefficient, Kra, I” the packed bubble column was found to be always larger than that In the unpacked bubble column In the range of lrquld velocattes from 6 7 cmlsec to 39 9 cmlsec, the value of K1.a m the packed bubble column Increased with the increasing hqtnd velocity while that in the unpacked bubble column was almost Independent of the hquld velocity
The equation
of the form
was successfully
adopted
to correlate the &a
data
Koch (Sulzer) motlonless mixers have been found to be effective aeration or oxygenation systems (Fan et al [5]. Hsu et al [6]) Bubble columns are used as absorbers and reactors for gas-hquld flow systems because of their low cost and slmphclty of operation However, bubble columns suffer from considerable hqutd backmlxmg and bubble coalescence To overcome these two shortcommgs, bubble columns are often packed with a variety of packmgs Generally speakmg, the gas holdup m a packed bubble column IS usually higher than that m an unpacked column because of the smaller mean bubble size and the hindering action of the packings The higher gas holdup reduces the static head of hquld I” the column and thus partially compensates the frIctIonal loss between flowmg streams and packings Therefore, the packtngs Increase the total pressure drop only shghtly over that of the unpacked column A bubble column can be operated tn the semibatch, cocurrent flow or countercurrent flow mode A packed column or tower IS usually operated countercurrently (Wen et al [7], Hutton and Leung[B], Shende and Sharma[9]) since the countercurrent mode of operatton
INTRODUCTION
The purpose of this work was to study gas-to-hquld mass transfer tn bubble columns packed with Koch (Sulzer) motlordess mixers Spectficdly, the ObJectives were to determme the values of the hquid-side volumetric mass transfer coefficient of such columns and to correlate tt as a function of operating parameters The effectiveness of motionless mixers m many chemical engineering operations has been well recognized m recent years (Westmore[l], Stevens[Z], Schott et al [3], Rosenzwelg[4]) The Koch motIonless mixer IS one of several different types of motlonless mixers that are available on the market It IS composed of many corrugated sheets of plastic or metal (Fig 1) The corrugation angle of the adJacent sheets in a mixer element IS reversed with respect to the mixer axis so that these sheets intersect each other and form a multitude of mixing cells These mlxmg cells promote generation of small gas bubbles If the mixer element IS Inserted m a gas-hquld flow system Bubble columns packed with tPresent address Monsanto, 800 N Lmdberg Blvd , St Lotus.
MO 63166. U S A
AY whole element
Rg
AY bolt elern&-#
I Koch (Sulzer) motIonless 945
rnlxers
346
K
B
WANG and L
usually produces a larger mean concentration drlvmg force than that of the cocurrent mode of operation However when the concentration of the transferrmg component m one phase remams constant, the mean concentration drtvmg force m the cocurrent operation IS ldentlcal to aat of the countercurrent operation Under this condrtlon, the cocurrent operation IS often preferred because the capacity of the packed column IS not llmlted by the floodmg condltlon The cocurrent upward mode was employed in thts work where pure oxygen was absorbed mto water m bubble columns packed with Koch (Sulzer) motionless mixers
APPARATUS
AND
RXPERMRNTAL
PROCEDURRS
Ftgure 2 IS the schematlc diagram of the experlmental set-up Koch (Sulzer) mot!onless mixers made of corrugated stainless steel plates (Ag 1) were packed m the bubble column, 4 m dra (IO 2 cm), with or wlthout spacers Tap water was pumped from a water tank through a rotameter to the bottom of the column Oxygen from an oxygen cylmder flowed through another rotameter to 3/4 m (I 9 cm) sparger at the bottom of the column Dasolved oxygen concentratron at the bottom and that at the top of the packed se&Ion were measured by means of an oxygen analyzer Thts oxygen analyzer had a sensttlvlty of *O 1 mg/l between 5 - 35°C with a repeatablhty of +O 1 mg/l The thermistor used together with the oxygen analyzer had an accuracy of &O 1°C The two oxygen probes were separated by 30m (76 2 cm) The experiments were carried out at five water flow rates ranging from 8 6 to 5 1 3 gpm (0 5-3 2 Ilsec) At each water flow rate, five oxygen flow rates rangmg from l-6 SCFM (472-2832 standard cm3/sec) were bubbhng cocurrently upward through the bubble column The packmg configuration In the column was changed from one experiment to another The five column
T
FAN
configuratlons used m this work (see Fig 3) were the unpacked column, the column contammg 5 AY whole elements with 4 spacers, the column contammg 10 AY half elements with 4 spacers, the column contammg 7 AY whole elements wtthout spacers, and the column contammg 14AY half elements without spacers The AY element has 112-m (1 3 cm) layer height and a hydrauhc dtameter of 0 68 m (1 7 cm) A whole element IS one whose length equals Its diameter whde a half element IS one whose length equals one half of Its diameter The adjacent mlxmg elements were rotated 90 deg to each other to induce three-dtmenstonal mlxmg Durmg each run of the expenment, records were kept of the flow rate and temperature of water, the flow rate of oxygen (at 1 atm and 2O”C), the number and type of motlonless mixers m the column, and the dissolved oxygen concentrations at the mlet and outlet of the column METHODS
ANALYSES
AND
CORRELATIONS
Kla based on the plug flow model The mass flux of oxygen m a dlfferentlal the bubble column can be expressed as dN = KLa(c*
- c)A dz
element
of
(1)
where Kr IS the mass transfer coefficient m the hquld phase, a the effective mterfaclal area per umt volume, c* the equlhbrtum concentratton of oxygen m water, c the concentration of oxygen m water, A the cross sectlonal
I
FIN 2
OF
The relattonshtp between the hquld-side volumetric mass transfer coefficient, KLa, and the concentrations at the Inlet and outlet of the bubble column was derived based on the plug flow model and also on the complete mlxmg flow model An emplrlcal equatton was then proposed to correlate the KLa data
Experunenfal
sefup
I
lo
drain
Mass transfer IIIbubble columns packed with motlonless mixers
947
model leads to
l4=
(KLu).
(8)
Correlatron of K,_a Most of the mdustrlally employed gas-liquid mass transfer systems are operated under turbulent flow condltlons Because of the lack of an understandmg of a purely theoretlcal predlctlon of mass turbulence transfer coeficlents under such condltlons IS not possible at this time It IS necessary, therefore, to correlate the experimental data emplrlcally or semi-emplrlcally Dtmenslonless correlations of the mass transfer coefficients often have the form
So*
11
It)
=-&(=I
(3)
(2)
(4)
(5)
Rg 3 Column configuratlon area of the column, and z the height of the column The mass flux of oxygen can be related to the oxygen balance m the hquld phase dN = d(Lc)
(2)
where L 1s the water flow rate Smce the solublhty of oxygen m water IS small. L can be assumed to be a constant Then, dN=Ldc Combmmg
Sh = a(Re)B(Sc)’
(4)
(9)
where 1 IS a characterlstlc length of the system. D..,B the dlffuslon coefficient, u, the superficial velocity of the llquld phase, p the hquld density, and p the hquld vlscoslty For a specific gas-liquid system, I, Dae, p, and p are constant, and thus eqn (9) reduces to
(3)
eqns (1) and (3), we have L dc = K,_a(c* - c)A dr
or
KL=V,
8
(10)
Based on 223 experrmental data pubhshed m the hterature. Gestrlch and Krauss[lO] obtained the followmg correlation for the effective mterfaclal area per umt volume,
a,
Equatton (4) can be Integrated from the bottom to the top of the column as
0!! -03M-0M3r
n=2(j,,
d
the pressure m the column IS nearly constant 1 atm, the solublhty of oxygen m water, c*, can assumed to be a constant Thus Smce
CTdc=ln {e}=yAh
at be
where h IS the column height, d the column diameter, L the gas holdup, and M a modulus for the hquld phase For water at 2o”C, M IS equal to 3 74 x 10”’ Equation (11) mdlcates that the effective mterfaclal area per umt volume, a, IS directly proportional to the gas holdup, l, le ame
(6)
Combmmg
(12)
eqns (10) and (12) leads to (13)
KLa = mv,@c where m U~I)~F
=
-&In{=}
(7)
K,a based on the complete mlxrng j?ow model This model assumes that the mlxmg IS complete, so that the propertles of the mixture in the column are umform m all parts of the v&se1 and are the same as those m the exit stream A material balance based on this
(11)
IS
a proportlonal
constant
RESULTSAND DISCUssKlN The results obtatned are described sectlon
and analyzed
In this
Correlatwns of KLa for packed and unpacked columns The correlations of the gas holdup, 6, that have been obtained by the present authors are included m Table 1
K
%
WANG
and L T
FAN
Table I Emputcal correlatmns for z and Kla
-___ Standard Deviation
5
AY Whole Mixers
(I-3)
a = 0 133 v
9 641clO-~
10 AY Half MXGXS
4
7
2 60x10-*
spacer.5
AY Whole
Mx.zers
(I-7)
(I-6)
14 AY Half Mixers
(I-9) (I-10)
and are plotted III Fig 4 With c: known, only two empnlcal parameters, m and @, need to be estunated m correlating &.a data accordmg to eqn (13) A non&near parameter estlmatlon method (Bard Ill)) was employed to determrne these -two parameters The resultmg expresslons for &_a are summarized m Table 1 for all five column configuratrons rnvestlgated m this work To test the vahdlty of eqn (13). whzch 1s based on eqns {lO) and (12). the values of E and &.a were correlated as
Effective mterfactal area As shown m FM 4, the gas holdup tn the packed bubble columns IS always greater than that in the un-
(161 (17) If the effectwe gas holdup, e,
Dvtdmg
mterfacu4
area. a, 1s proportional
to the
eqn (16) by eqn (18) leads to
If the resultant expresstons for K, depend only on v,, eqn (13) IS vabd From the functlonai dependence of &a, e and KL, on V, and vr summarized m Table 2, it can be seen that KL is strongly dependent on u, but only weakly dependent on u,, mdlcatrng that thts IS the case
v,* CVbf~
FIN 4 Correfattons of gas holdup for unpacked and packed cotumns ----, Unpacked, (I), Packed wtth 7 AY w&ok mtxers, (2). Packed wrth 5 AY whole mtxers, (3), Packed wltb IOAY half mtxers , (4, Packed wcth 14 AY half mixers
Mass transfer
m bubble columns
Table 2
_-
FunctIonal
packed wuh mouonless
dependence
on VI and u,
Functional
Relationship
0 KI_a =
7 81
x
10-3
083
0
(11-l) "s
553
5 AY
Whole
Mxsers
%"
_
_
7 14
"%O
x
10-3
644
" L
KL
949
mixers
"
102 "g-o
0
457
"gO
(11-3)
519
01-4)
e +
E
=
4
93
x
10-2
OZ8
q0 0
4
10
AY
Spacers
Half
r. Mixers
't"
= =
5
20
"e x
10-3
"
=
9
65
x
10
0
-2
b 4
spacers
532
~
-0
*
4
35
x
10
Whole
Mixers
E
=
5
16
x
0
134
666
0
0
623
463 160
-go
631
"
0
589
0
588
" ?"
7 AY
(11-6)
10 -2
(11-7)
(11-8)
"g
"to
-3 %."
(11-5)
056
"g
"L
KL
-0
575
"g
a +
VB
485
0
-0
(II-9)
(11-10)
8 102
v
(11-11) "8
P
(II-12) KL 't"
= =
6
54
'ho x
1O-3
0
p
733
%.O
576
O1
0
501
0
528
(11-13)
"g 14
AY
Half
Mixers
=
6
54
x
E
10 -2
-0
v
packed bubble column Since the existence of motlonless mixers prevents bubbles from coalescence and breaks up Iarge bubbles, the mean bubble size IS smaller m a packed bubble column than that m an unpacked column The larger gas holdup and the smaller mean bubble size lead to the higher effective InterfacIal area because (see, e g CalderbankEl2J)
a=& dts
(20)
where dh IS the mean Sauter diameter defined as
(21) Mashelkar and Sharma[13] found that the effective mterfaclal area based on the dlsperslon or effectrve volume was always higher m the packed column. however, the effective mterfaclal area m the packed column based on the total volume (of the empty column) mtght not always be lugher than that m the unpacked column This was due to the large volume occupied by the conventional packmgs, such as Raschlg rmgs, pall rings and Intalox saddles, employed by Mashelkar and Sharma[l31 The gas holdup data for the packed columns shown m Fig 3 are based on the total volume (of the empty column), and are always higher than those for the
=
(11-14)
"g 0
KL
057
L
"IL
633
-0
027
(11-15)
"g
unpacked column Therefore, the effective mterfaclal area, which IS proportlonal to the gas holdup, for bubble columns packed with Koch (Sulzer) motloniess mixers should always be higher than that m the unpacked column The superlonty of Koch (Sulzer) motlonless mixers over conventlonal packmgs IS partly due to the small volume occupied by the former (less than 4%) m contrast to a relatively large volume occupied by the latter KLa for the unpacked column KLa for the unpacked column IS plotted m Fig 5 The correlating curves, based on eqn (I-2). are also mcluded in this figure They fit the data fairly well It can be seen tn the figure that the hqutd-side volumetric mass transfer coefficient for the unpacked column Increases with an mcreasmg gas flow rate but IS almost independent of the liquid flow rate The functlonal dependence of K,a on u, and v,. as shown m Table 2, also leads to the same conclusions The mass transfer coefficient, KL, usually Increases with an mcreasmg hquld flow rate as predicted from eqn (IO) The gas holdup and thus the effective mterfaclal area decrease wtth the mcreastng hqurd flow rate Consequently, KLa for the unpacked column 1s almost Independent of the hquld flow rate, as can be seen m Table 2
K B
WANG
L T
and
048
FAN
T
040
032
399
J 41 024 d kd
ooa-
0
”
I
b
v,, cm&x FIN 5 Llquld-side volurnetrlc mass transfer coetliclent for unV, standard devmtlon = 5 14 x IO-’ packed bubble column (cmlsec) + 67 0 190 A 246 0 325 x 399 for the packed columns KLa for the packed bubble columns are shown m Fogs 6-9 The correlatmg curves, based on the emplrtcal equatrons tabulated m Table 1, are also Included In these figures and dtsplay good fit These figures mdlcate that the KLa for packed columns mcrease with an mcreasmg gas flow rate However, contrary to the unpacked column, the KLa data increase with an mcreasmg hquld flow rate m the packed columns This IS because the gas holdup, and hence the effective mterfactal area, are independent of the hquld flow rate while the mass transfer coefficient, K,, mcreases with an mcreasmg hquld flow rate The functlonal dependence of Kra, l and Kl. on u, and v, IS tabulated m Table 2 Slmllar to the effective mterfaclal area, the hqmd-side volumetnc mass transfer coefficients based on the dlsperslon or effective volume m bubble columns packed with conventional packmgs were found to be higher than column (Mashelkar and unpacked those m the Sharma[l3]) Nevertheless. based on the total volume of the empty column the hqtud-side volumetric mass transfer coeficlents for bubble columns packed with conventional packings may not be higher than those In the unpacked column because the volume occupied by the conventlonal packmgs IS fairly large Therefore, small packings with large void fractions are desirable for better performance For example, the values of a and KLa of the bubble column packed with screen packmgs which occupied about 10% of the total volume of the empty column obtamed by Voyer and Mdler[l4] were higher than those of the unpacked column The results plotted in Figs 6-9 are based on the total volume of the empty column It can, therefore, be said that hqurd-side volumetric mass transfer coefficients m bubble columns packed with Koch (Sulzer) motlonless mixers are always htgher than those m the unpacked column The range of the ltqutd flow rate covered m this work
Cl
Rg 6 Llqutd-stde
compared to those reported tn the and Sharma[l3] covered the range
i2
40
9
cm/set
mass transfer coefficient for bubble column paced wuh 5 AY whole rnlxers and 4 spacers standard devlatlon = 9 64X 10e7 V, (cmlsec) + 6 7 0 I9 0 A 24 6 0 32 5 x 39 9 Packed -----Unpacked VI (cmlsec),
volumetrtc
parameter
&a
was relattvely large literature Mashelkar
k
16
of hquld velocltles from 0 1 to 0 5 cmlsec and found that the superficial hquld velocity had no effect on the values of K,_a m the bubble columns packed wtth conventional packings Voyer and Mtller [ 141 covered the range of lrquld velocities from 0 5 to 3 cmlsec and found that there was a considerable increase m the value of K,_a with an Increase In the hquld flow rate In the bubble columns packed with screen packmgs In this work, we covered the range of liquid velocltles from 6 7 to 39 9 cm/set, and found that there was an Increase in the value of &a with an tncrease m the hquld flow rate m the bubble columns packed with Koch motlonless mixers 048
9
032
j
024
016
------
008
67
0
58
cm/sex
Rg 7 Liquid-stde volumetric mass transfer coefficient for bubble column paced with 10 AY half mixers and 4 spacers standard devlatton = 2 37 x IO-’ V, tcmlsec) + 6 7 0 190 A 246 Cl 32 5 x 399 Packed ----unpacked V, (cmlsec) parameter
Mass transfer
III bubble columns packed
with motIonless
951
mixers
OPB
w
399 Cl24
032
-
SAY
_____
unpockd
0
Plug
A
[email protected] -
Flow
Mixas
whole
4
Spacsrr
Modal Muung
Model
190
crzo !
024
O.l6
006
0
0
a
16
24
ys*
cm
32
40
46
/set
Rg 8
Llqtud-side volumetnc mass transfer coefficient for bubble column packed with 7 AY whole mixers standard devlatlon = 9 23 X IO-’ V, (cmlsec) 0 190 A 24 0 0 32 5 X 39 9 Packed --unpacked VI tcmlsec) parameter 046
Q40
J
8
“0 Fig
0.32
L
T Y d k?
024
al6
24
16
%*
cm
32
40
-3
4.9
/set
FIN 9 Llquld-side volumetnc mass transfer coefficient for bub ble column packed with 14 A?’ half mixers standard devlatlon = 158x10-’ V,(cm/sec) + 67,0 190 q 325 x 399 Packed -----, unpacked V, (cmlsec) parameter Effects of axrai mrxrng on &_a The values of &a m a bubble column depend on the residence time dlstrlbutlon or the degree of axial mlxmg m the hquld phase The plug flow model yields the lower bound of K=a while the complete mlxmg flow model yields the upper bound of Z&a (Deckwer et al [IS]) This IS demonstrated by Fig 10 where the values of KLa substltutmg experImentally measured obtamed by concentrations Into
at eqns
the
Inlet
and
outlet
of
the
10
’
32
cm/se2
Liquid side volume(ric mass transfer coefficients on plug flow modef and completely rnlxmg model
based
on the complete mlxmg flow model and obtained quite high values of KLa Mashelkar and Sharma [13] made use of the axial dlsperston coeficlent m the hqurd phase reported by Relth et al [18] and the experimental data of Towell et al 1171 to recalcuIate K,_a values based on the axtal dlsperston model They found that the Z&a values based on the axial dtsperslon model were very close to those based on the plug flow model In order to assure that the flow patterns m the systems under mvesttgatlon were mdeed of the plug Row, the extent of axial dispersion was measured m the bubble column packed with 5 AY whole mixers without spacers at two low hquld velocltles, uI = 2 0 cm/set and vI = 2 9cm/sec, where the axial dlsperston could be appreclable The modified Peclet number, Per, based on the column diameter, d, and the relative velocity between the gas and hqutd phases, v,, was found to be nearly constant Pe'f=g=207*40
1
(22)
where (23)
bubble
(7) and (8) are plotted Although the actual values of K,_a he between the upper and lower bounds, the assumption of the plug flow has been found to be reasonable for packed bubble columns (Carleton et al [16]) Towel1 et al [17] reported their K,a data based column
24
la
Lld was fixed at 7 5 and at the lowest 6 7 cmlsec, of the present gas absorption Ur
(I - 6)Ur
hquld velocity. experiments,
952
K
B
WANG
ranged from 0 09 to 0 15 Assuming that Pet remamed constant at the value given by eqn (22), we can estimate that Pe, ranged approximately from I5 to 26 This estlmatlon of Per should provide a further support that the plug flow condltlon was essentially approximated in all absorption experiments CONCLUSIONS
The slgmficant conclusions of this work are hsted below 1 Kro m a bubble column packed with Koch motlonless mixers IS higher than that m an unpacked bubble column of the same size 2 KLa III a bubble column packed with Koch motronless mixers Increases with an mcreasmg hquld flow rate while that m an unpacked column IS almost Independent of the hquld flow rate 3 The KLa data for the packed and unpacked columns can be correlated accordmg to the form KLa = rn$e Acknowledgements-This Engmcermg Expertment
work was partially supported by the StatIon, Kansas State Unlverslty The equipment and packmgs were supphed by the Koch Englneermg Co Inc NOFATION
effective interfaclal area per unit volume, cm-’ A cross sectlonal area of the column, cm’ concentration of pure oxygen in water, mg/l .z equlhbnum concentration of pure oxygen tn water, mg/l at the bottom of the Cl3 oxygen concentration column, mg/l oxygen concentration at the top of the column, mgll d column diameter, cm D AB dlffuslon coefficient, cm2/sec d,, mean Sauter diameter, cm h total column height. cm KL mass transfer coefficient m the hquld phase, cmlsec hqmd-side volumetric mass transfer &a coefficient, see-’ Kra based on the plug flow model, set-’ (Kra bF mlxmg flow &_a )C KLa based on the complete model, set-’ 1 characterlstlc length of the system, cm L water flow rate, I/set a
and L T
FAN
parameter, defined In eqn (13) modulus for the hquld phase, detined m eqn (11) mass flux of oxygen, glsec Peclet number m the liquid phase modified Peclet number, defined m eqn (22) supertIcla1 gas velocity, cmlsec superficial hquld velocity, cmlsec relative velocity, defined in eqn (23). cmlsec he@ from the bottom of the column, cm Greek symbols a parameter, defined m eqn (9) @. y exponents, defined In eqn (1) c gas holdup p hquld density, g/cm3 fi hquld vlscoslty, g/cm-set RBFBRENCE
L E . “Mixmng with no movmg parts to make btg Impact m Europe.” Proc Engng 1970 September 87 I21 Stevens B , Proc Engng 1973 Aprll 76 B and LaBombard D, Chem I31 Schott N R Wemstem Ennna Proar I975 71(l) 54 M D Chem Engng 1977 84(10) 95 141 Ro&&welg ISI Fan L T .Hsu H H and Wang K B . J Chem Enana Data
[II Westmore
1975 20(l) 26 Fl Hsu H H , Wang K B and Fan L T , Water Sewage Works L975 122(2) 34 [71 Wen C Y , 0 Brlen W S and Fan L T ,J Chem Engng Data 1%3 8 42 I81 Hutton B E T and Leung L S . Chem Engng Scr 1974 29
1681 [91 Shende B W and Sharma M M , Chem Engng Scr 1974 29 1%3 Area Bubble [lOI Gestrlch W and Krauss W , Spec Intetfac Layers 1976 16(i) IO [ill B&d Y Nonlinear Parameter Esfrmatdon and Programmrna Program Informatron Department, IBM I%7 1121 Caldkrbank P H , Mccmg, Theory and Fractlce (EdIted by Uhl V W and Gray J B ), Vol II Academic Press. New York 1967 [131 Mashelkar R A and Sharma M M , Trans Instn Chem Engrs 1970 48 7162 I141 Voyer R D and Mdler A I, Can I Chem Engng I%8 46 335 [ISI Deckwer W D , Burckhart R and 2011 G , Chem Engng SCI 1974 29 2177 [l61 Carleton A J , Flam R J , Renme J and Valentm F H H , Chem Engng Scl I%7 22 1839 [I71 Towell G D, Strand C P and Ackerman G H , Mdxrng Theorv Related to Practrce (EdIted by Rottenburg P A ) The Instdution of Chemical Engmeers, London I%5 WI Relth T , Renken S and Israel B A , Chem Engng SC1 1%8 23 619 rt91 Sherwood T K PIgford R L and Wdke C R, Mass Transfer McGraw-Hffl, New York 1975