ABSORPTION PROCESS UTILIZING PACKED TOWERS

V O L 5 9 NO. 2 F E B R U A R Y 1 9 6 7 41
 0
Figure 1. Overall &ne in absorption
TIE absorption process is shown to be more com-
plex than the dfisional resistance encountered
at the interface between the gas and the liquid
in this review of the classical theories of this
process
he absorption process is classically regarded as
Tbg one of diffusion. However, in the last fifteen
years some investigators have begun to recognize that
these classical theories do not explain all of the facts.
Whitman [laterWiandWhitman (74,79)lisgenerally
regarded as the originator of the classical two-film theory
of mass transfer on which practically all work has been
based, especially that related to packed towers. But
subsequent experience through the yeara indicates that
the two-film theory leaves a bit to be desired as a useful
design concept.
Higbie (72) reported further study of the two-film
concept in 1935 with the added modification of a con-
centration gradient at each lilm, rather than the very
thin film of Lewis and Whitman. This has become
known as the penetration theory; and though it was an
improvement on the original two-film theory, it still did
not explain all the observed phenomena of gas liquid
magll transfer. In 1951 Danckwerts (6),recognizing the
limitations of the stagnant film concept of both the two-
6lm and penetration theories, postulated a rapid surface
renewal theory. However, after offering a speculative
philosophy, he did not attempt any experimental proof.
Lewis and Whitman (74) anticipated the development of
Higbie and D a n b e r t s by discovering that an agitated
vessel the thicknes ofthe film will vary as the 0.8 power
of the velocity of rotation.
Cumnt theory regarding absorption maintains that a
soluble gas moves from the gas to the liquid phase under a
driving p m which is the difference in p m u r e
baween the mole fraction in the gas phase and the vapor
AUrnoRS John S. Ecknt is Cfief Engineer and Earl H.
Fm&,Lcrmmd R. Rollison, and h i s F. W a k me EngincCrs
for U.S. Stoneware, Inc.
42 INDUSTRIAL A N D ENGINEERING CHEMISTRY
~ ~ ~ ~ ~ _ _ _
2 3 4 5 4 7 8
615 UIE fH./sd
Figure 2. Perfmame ckaracfm’stics of packed be&
pressure of the same gas in equilibrium with the liquid -
~
phase:
F = A, - Ai
The rate of transfer of this gas is directly proportional to
the driving force but is resisted by its diffusion rate in
both the gas and liquid phases. These restrictions have
come to be known as the liquid and gas film resistances.
I t is purely a philosophical point whether these two films
are very thin or whether they vary in concentration
frum one side of the interface to the other. Several
things are known about these films. Their resistances
are best determined by experimental operation of the
system, and they are seldom, if ever, separated for use in
industrial design practice. In other words, the film
mistance used in industrial design is used as a combined
constant:
-1= - +l - m
Koa koa ~ L R
or:
_1 = _ 1
KIP kLa
+-d o1a
Here koa and k,a are the individual film resistances and
m is the slope of the operating line (Figure 1). If the
operating line lies above the equilibrium line, absorption
or m a s transfer from the gas to the liquid phase will take
place; while if the operating l i e lies below the equilib-
rium line, then maw transfer from the liquid to the gas
phase, or “stripping” will take place.
A more useful way of expressing the 6lm constant is by
- r
mass transport rather than by resistance:
N = Koa VAy
and
N=KLRVAX
Koa and KLRare usually referred to as the maw transfer
coefficients. Koa is most useful in designs when com-
position of the gas phase is of primary interest and K M
where composition of the liquid phase is of primary
intereat. A b s o r b are generally designed using the
_ ~ ~
K d coefficient. Colbum (5) expresses this transport
in the form of transfer units where the above expressions
take this form:
The useful form of these equations, as used in designing
packed absorbers, is:
Adequate methods for calculating the required area of a
tower (8) are already in general use so that it will not
be necessary to review these techniques here.
The chief problem in designing an absorber is deter-
mining the m a s transfer coefficient (&) for a packed
tower and the plate efficiencyfor a plate tower. Much
more is known about packed absorbers than the plate
type, since most tower research has concerned distilla-
tion. A great amount of work has been done on the
use of packed beds for absorbers, both on a laboratory
and plant scale. Much of this work has been published,
along with a great amount of theory in the realm of
mathematical models.
The absorption process is complex and, therefore, has
not yielded to accurate mathematical analysis. h i s
(74) was dose to the truth when he found that agitation
of the liquid reduced the film resistance. The further
improvements of Higbie (72) and Danckwerts (6) result
from their perception that the film resistance would be
sensitive to agitation as first determined by Lewis. The
degree of this effect is best illustrated in the distillation
process, where higher pressure drop operation is more
commonly encountered than in absorption (Figure 2).
Work done in distillation (9, 70) holds a major key to a
few of the problems.
Figure 2 illustrates the controlling domains of masa
transfer. I t is rather evident that the diffusional process
Figure 3.
Koa us. %
' conversion
C02-NnOH-H~0
E. 0.9
3 0.8
only controls at low pressure drop where gas friction is
not sufficient to induce much surface turbulence in the
liquid. When the pressure drop in the bed becomes as
high as 0.6 inch H20/foot packed depth, it is further evi-
dent (Table I) that the mass transfer, resulting from in-
tense mixing of the gas and liquid at the interface, is so
much greater than the slower diffusional process that the
effect of variables generally thought to control the slower
diffusional process is almost completely lost. Table I
indicates that most of the usual variables which are
thought to affect mam transfer, under more or less
stagnant f ilm conditions, have no noticeable effect at a
pressure drop approximating 0.6 inch/foot packed depth.
The properties of latent heat of vaporization, heat
capacity, relative volatility, average change of concen-
tration per plate of separation, and the diffusion co-
efficient have been subjected to wide variations by a
selection of six different binary mixtures, only to find
that the HETP is really constant at this pressure drop
with the exception of the methyl chloroform-toluene
system. [See (70) on method of correcting for end
effects between 5 and 10 feet of packed depth.]
Complex variables are needed in an equation expnse
ing the mass transfer characteristics between a liquid and
a gas in a packed bed where the liquid is filmed over a
solid surface and agitated by the gas. The mass transfer
coefficientat low pressure drop Zone A (Figure 2) is under
the major influence of the Schmidt No. Ns,and is
basically a diffusional operation. In Zone B the mass
transfer is controlled by both the Schmidt No. and
the Reynolds No., N,; and as Zone C is ap-
proached, it is controlled almost entirely by the Reynolds
No. The shape of the HETP curve indicates that
the relationship of these two groups of variables on the
rn transfer coefficient is complex and will require
complex coefficients related to the pressure drop across
the packed bed.
Tepe and Dodge (77) determined that for the NaOH-
C O r H a system, the rate of mass transfer would in-
1.5
1.4
1.3
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
V O L 5 9 NO. 2 F E B R U A R Y 1 9 6 7 43
 crease by the sixth power of the absolute temperature.
In general, however, it is not presently known if this
effect will be the same with salts other than sodium
carbonate. Much research is needed to determine the
temperature effect on rate of mass transfer (on systems
other than NaOH-COz-HzO). The hot carbonate
system achieves a much higher mass transfer because of
the higher temperature of absorption ( 3 ) . Up to the
present time, generalized equations and mathematical
models have had little value in actual design work
because they contain relative system properties to which
the designer is unable to assign sufficiently accurate
values to yield worthwhile answers.
Absorption systems are best designed from either pilot
or existing plant data with modifications from this base.
For example, the Koa values are known for most of the
systems used in industry today a t specific operating
conditions (Table 111). The effect of various operating
conditions on this basic mass transfer coefficient is be-
coming more commonly known (Figures 3 and 6).
The systems which Lewis preferred to call “liquid filni
controlled” were gases of low solubility such as 0 and

TABLE I
~

I I I
Methyl
Chloroform Benzene i-Octane Methanol i-Propanol Acetone
Toluene Toluene Toluene Water Water Water
-

Latent heat of 156 169 117 474 296 220

vaporization, AH 156 156 1000 1000 1000

Heat capacity, cp. 0.26 0.24 0.38 0.599 0.596 0.51 1

0.26 0.28 1 .o 1 .ooo 1 .o
Relative volatility 1 3.4-4.6 25 1 .4-1 . 2 (740 mm.)
1.78-1.2 (100 mm.)
I
7.0-2.45
I
1 ,46-l .03
1
20.0-1.25
I
I I
Av., Aconcn./plate 21.1 11.82 4.3
3.82
10.5 1.55 j 5.35

Diffusion coefficient 6.58 X lo-’

HTU
H ET Pa 1
1.23
1.36b
6.66 X l o w 5
0.90
1.10
4.63 x
0.92/0.80

0.96/0.96
10-5

1
5.96
0.35

0.70*
x 10-5

1
3.87
0.60
1.00
x 10-5
1
1
3.73
0.68
0.98
x 10-5

a H E P T and HTU taken at 0.6 inch HzO/ft. Afi.

Fine-foot packed height. All other tests were run at IO-ft. bed height.

TABLE II. KGUWITH ADDITIVES IN NAOH-CO, SYSTEM

I I
~~

Li uid Rate, Gas Rate, Pressure Drop, Amount

Lb.fSq. Ft. Hr. Lb./Sq. Ft. Hr. In. /Ft. Additive Additive, P.P.M.
-~
5000

5000

5000
760
520
1000
0.15
0.15
0.30
None

Detergent

None
1i :: 11 -

..
2.4
2.0

2.4
5000 1 900 1 0.30 Detergent 1 .. 1 2.3
1 I
~

5000 1025 0.30 Detergent and antifoam A 15 2.3

~ ~ ~~ ~

5000 1050 0.30 Antifoam B 2 2.3

5000 1050 0.30 Antifoam B 16 2.2

5000 1350 0.6 None .. 2.7

5000 I 1030 1 0.6 Detergent I * * 2.7

5000 1 1290 I 0.6 Detergent and antifoam A ~ 15 1 2.6

Additives:
Detergent. Alkyl benzoyl sulfonate.
Antifoam A. Unsaturated monohydric alcohol.
Antifoam 3. Silicone.

44 INDUSTRIAL A N D ENGINEERING CHEMISTRY

 TABLE 1 1 1

Gas Reagent Hydrate Ion Ionization “K”

NaOCl Na+OCI- 20.0

HCI H+CI- 16.0 7 x 10-4
NHOH “+OH- 13.0 1.8 X
MEA. SO3 MEA+2S03-2 8.0 ...
NazS03 Na +so,-’ 7.0 ...
DEA*S DEA +2S-2 5.0
KzC03 K+C03-’ 3.10 ...
MEA * C 0 3 MEA+’COg-’ 2.50 ...
Na2COa N~+COS-~ 2.25 ...
HzS H+S-2 0.400 1.1 x 10-7
HzS03 Hf S 0 3 - 2 0.317 5 . 6 X lo-*
HOC1 H+OCI- 0.138 3.2 X lo-*
HzC03 H +C03-’ 0.072 5.0 X
- H+OH- 0.0072 1 . 0 x 10-14

HzO * KOH KOH K+OH- Kaa above a t 25% <10

HzO. NaOH NaOH NafOH- completion 10
HzO MEA MEA*OH MEA-OH- 6 . 0 X 10-1
H z 0 -DEA DEA*OH DEA-OH- 1 . 6 X lod2

a 11/2-Inch Intalox saddles.

CO and those termed “gas film controlled” were of high
solubility, such as hydrogen chloride and ammonia.
I t may be concluded that all distillations would be
largely “gas film controlled,” while absorption processes
could be either.
With highly soluble gases, increasing the gas rate
through the tower increases the rate of absorption very
greatly. For gases of low solubility, it has little or no
effect in the pressure drop range where absorbers are
usually designed-i.e., less than 0.5 inch/foot of packed
depth. However, at higher pressure drop, the gas induces
sufficient agitation in the liquid that an increase in gas
rate will cause an increase in the absorption coefficient
even with gases of low solubility (Figure 4). The mass
transfer coefficient of highly soluble gases increases much
faster at this higher pressure drop than it does for the
gases of low solubility.
While the effect of the gas rate on the mass transfer
coeficient is rather selective, especially at low pressure
drop, an increase of liquid rate will always increase the
mass transfer coefficient (Figure 5) whether the gas is of
either low or high solubility.
There are certain additives which will influence the
rate of absorption at the interface, some of which have a
very great effect. The presence of a surfactant (Table
11) will slow down the rate of transfer very greatly a t
low pressure drop but will have little effect a t higher
pressure drop. If the surfactant happens to produce a
foam, then both capacity and efficiency may be reduced.
A technician skilled in the operation of a Tensiometer
can observe that the surface tension of a liquid contain-
ing a surfactant takes time to stabilize after the surface
has been disturbed.
This time factor, amounting to several seconds, tends
to give an unpredictable behavior to a packed bed at low
pressure drop, particularly with packings where widely
different friction factors exist on the surface, such as
Raschig rings. The interior surface of a Raschig ring,
on the average, is subject to much lower gas and liquid
flow than the outside. However, at higher pressure
drop, the interior of the Raschig ring tends to become
more uniformly efficient.
In most industrial absorption processes, the gas is
reacted with some substance to form a semistable com-
pound in the liquid phase. This technique permits a
great deal more gas to be absorbed per gallon of liquid
circulated, and, in most instances, will increase the mass
transfer coefficient. I t is possible to have the chemical
reaction in the liquid phase so slow that it will control
the mass transfer across the interface; a typical example
of this is the reaction of C 0 2 to form carbonic acid and
then sodium carbonate and sodium bicarbonate (Figure
3) as the 100% carbonation point is approached.
As carbonation proceeds, it has a profound effect on
the mass transfer coefficient, undoubtedly due to the
presence of HCO3 in repressing ionization of the carbonic
acid to Hf and -HCO3. This absorption is said to be
liquid film controlled ; however, when there is little
carbonate present, it can operate in a gas film controlled
manner (Figure 4). Here, at 0.5 inch/foot pressure drop

VOL. 5 9 NO. 2 FEBRUARY 1 9 6 7 45


KING HtlCHI 10 h.
I. STEEL PALL kl'",6So 8o
00
=10.000 Ibr./ll.? hr.
COUC. 0.85 % LOG M o
IH CONC. 4% W' 0
CARBOULTtD
Figure 4. Kon us. air rote
Ewe 5. Koa V.I lipid rate. Gas and liquidflm control
0 1 2 3 4. 4.5
MOLES CO,/MW MEA
Figure 6. KGa US. cononrion
and beyond, the iilm resistance has been reduced to prac-
tically zero by internal mixing action resulting from high
pressure drop, and the Koa is regulated almost e n t i d y
by the rate of reaction of the carbonic acid in the liquid
phase. This is indicated by the Koa increasing rapidly
with increase in liquid rate.
Though generally the rate of absorption is controlled
by conditions in the liquid phase, there is one very
notable example of an absorption being controlled by
the rate of a reaction in the gas phase. In the absorption
of nitric acid to form nitric acid, the reduced oxide is
released to the gas phase to be reoxidzed:
3NOz+HzO*2HNOs+NO
2 NO + Oz+ 2 NO2
46 INDUSTRIAL A N D ENGINEERING CHEMISTRY
This gas phase oxidation is one of second order and can
best be speeded up by increasing the partial pressure of
the reactants. Nitric acid absorbers are characterized
by very large volume to give time for the gas phase
oxidation to take place. The sue of these units can very
materially be reduced by operation under pressure.
Because of the limitation of absorption rate by the gas
phase plus reaction, absorbers of this type seldom are
limited by interfacial area. Often they can be packed
with low performance packings, such as Raschig rings,
or can be built with very primitive-type trays. Heat of
reaction must be removed from these absorbers either
by high circulation rates with external heat exchangers
(packings) or internal heat exchangers (plates).
Liquid phase reaction after absorption is by far the
more common type encountered in industry. Since
most industrial gases are acid in nature, the scrubbing
liquors are characterized by being very strong bases.
A strong base tends to take up the acid ions formed from
the acids of the absorbed gases as rapidly as they are
formed. Of gases which are commonly encountered in
absorption processes, carbonic acid seems to be least and
hydrogen chloride the most reactive.
Gases which form acids or bases, which become highly
dissociated in solution, have higher mass transfer co-
efficients than those exhibiting low dissolution or none
at all (Table 111). K N values for some of these typical
gases using l'/*inch Intalox saddles are given in Table
111. I t is further evident that the salt reaction in the liq-
uid phase has a profound effecton the rate of absorption.
Figures 3 and 6 show how this chemical reaction affects
the rate of absorption for COz in relation to degree of
carbonation and concentration of reactants. The small
table on Figure 3 gives the Koa for NaOH-COp-HpO
system under the operating conditions at the reference
point for all of the commonly used packings as obtained
from published charts (78).
Figure 3 should be helpful to the designer in that it
would be possible to predict performance from one size
to another. When it is used in conjunction with Table
I11 and published charts (78), the engineer has available
the necessary information to design absorption towers
for any of the commonly encountered industrial gases.
Also, he can predict with reasonable accuracy the design
constants for absorption of gases which are not com-
monly encountered if something of their dissociation
constants when in solution and the rate of removal of
their ions by chemical reaction are known.
The rate of removal of the ions of the dissolved gas by
chemical reaction is illustrated by the mass transfer
coefficients realized with potassium hydroxide, sodium
hydroxide, and monoethanolamine. The higher the
ionization constant of the reagent, the greater will be
the rate of mass transfer.
Highly soluble gases give higher mass transfer co-
efficients than the less soluble ones-for example, SO,
is very soluble in HISO, and exhibits a very high K N ,
while 01is very sparingly soluble in HzO and exhibits
a very low Kon. Increasing the pressure on the system
' decreases the Koa by the 0.5 power of the pressure (76)
 but increases the mass transfer itself because of the extra
drive resulting from increased pressure of the soluble gas.
The hot carbonate process for C02 absorption derives
its efficiency from increased temperature (usually 300-
350' F.), high pressure (350-450 p.s.i.g.), and use of
potassium ions in the reagent. There are some proprie-
tary processes wherein a catalyst is added which in-
creases the mass transfer coefficient and reduces the
amount of steam required to effect regeneration of the
carbonate.
The Giamarco-Vetrocoke (7, 75), Benfield ( 7 4 , and
Catacarb ( 7 7) processes all use hot potassium carbonate
as the reagent salt. The principal difference in the
three processes is the use of various additives which
improve the mass transfer coefficient, reduce steam re-
quirements for regeneration, and reduce corrosion of the
system.
Diethanolamine (DEA) and monoethanolamine
c (MEA) are commonly used as water solution reagents to
react with COz and H2S in absorbers. I n recent years
DEA has tended to lose popularity except for gases con-
taining carbonyl sulfur resulting from cracking opera-
tions, such as those found in refineries. Since DEA does
not lend itself to easy reclaiming methods, it proves to be
more expensive than MEA. The mass transfer coeffi-
cient of MEA tends to be much higher (Figure 6) than
either the sodium or potassium carbonate systems. MEA
systems are generally not run as hot as potassium
carbonate because they tend to break down, with con-
sequent back pressure of COZ at a lower temperature than
potassium carbonate.
In recent years two new organic reagent-type processes
have appeared. The Fluor process (2, 73) uses mixtures
of solvents of the type consisting of propylene carbonate,
glycerol triacetate, butoxy diethylene glycolacetate, and
methoxy triethylerie glycol acetate. This process can
be used as a dehydrator as well as an absorber. Shell's
Sulfinol process (7) is similar to the Fluor process in
that it uses a mixture of organic reagents; however, it
operates as a water solution and will not dehydrate the
treated gas.
The performance of these latter two systems is some-
what obscure. The cited references give most of the
information which is available to the public.
Conclusions
It may be concluded that the absorption process is
, somewhat more complex than the diffusional resistance
encountered at the interface between the gas and the
liquid. Gas induced agitation of the interface will in-
crease the rate of mass transfer so much that the influence
of the factors affecting mass transfer based on film resist-
ances can no longer be detected. At lower operating rates
where this gas induced turbulence is not so great, there
are many other factors than relative volatility and
diffusion coefficient which become a factor. The
presence of a surfactant will have a definite retarding
effect unless there is a large amount of agitation from gas
friction.
Gases whose solutions exhibit a high degree of dis-
sociation have high mass transfer coefficients. Reagents
whose solutions have a high degree of dissociation cause
the gases to have higher mass transfer coefficients.
Gases which do not ionize in solution show mass transfer
coefficients proportional to their solubility. The very
soluble gases exhibit high mass transfer coefficients.
Increasing temperature increases the rate of mass trans-
fer; however, because it usually reduces the solubility
of the gas, there are limitations on the extent to which
the temperature can be profitably raised. Increasing
the temperature of absorption will increase the rate of
mass transfer only if it does not develop back pressure in
the liquid because of thermal degradation of the salt,
or reduce the solubility of a nonreactive gas. Several
proprietary processes are available for removal of COr
and HZS which employ catalysts or special mixtures of
organic reagents.
Nomenclature
A = area of packed bed, sq. ft.
A = atmospheres partial pressure soluble gas vapor pres-
sure in liquid phase
4 = atmospheres partial pressure in gas phase of soluble
gas
D = hydraulic radius, ft.
Dv = diffusion coefficient, sq. ft./hr.
F = driving force atmospheres, lb./sq. in.
G, = molar gas rate, lb. mol./sq. ft. hr.
H = height of packed bed, ft.
HTU = height of a transfer unit, ft.
&a = overall gas mass transfer coefficient N/cu. ft. AY
KLa = overall liquid mass transfer coefficient N/cu. ft. AX
N = lb. mol./hr.
NSO = Schmidt NO., I .1
pi DU
V = vol., cu. ft.
koa = gas film mass transfer coefficient, N/cu. ft. AI'
kLa = liquid film mass transfer coefficient, N/cu. ft. AX
-
(1 y), = log mean (Yz- Yl)/(ln YZ/Y1)
m = scope of operating line
U = velocity ft./sec.
AX = liquid phase driving force mol fraction
AY = gas phase driving force atmospheres
APm = log mean
P = viscosity, lb./ft. hr.
1 = liquid density, Ib./cu. ft.
LITERATURE CITED
(1) Anon., Chm. Week, 42-4 (Jan. 28,1961).
(2) Anon.,Ppt. Rcf. 40 (51,288-90 (May 1961).
(3) Benson H. E., Field, J. H., Jimeson, R. M., Chem. Eng. Progr. 50 (7), 356
(July 1994).
(4) Bienstock, D., Field, J. H., Corrorion (NACE) 17,87 (Dec. 1961).
(5) Colburn, A. P., IND.END.CHEM. 33 (4),459 (1935).
( 6 ) Danckwerts, P. V., Ibid., 43,1460 (June 1951).
Q7) Dum, C. L Freitas, E. R Goodenbour, J. W Henderson, H. F., Papa-
dopoulos, M. ,.; H P / P R 43 (3x 150-4 (March 1964):)
(8) Eckert, J. S., C h m . Eng. Progr. 57 (9), 54 (1961).
(9) Ibtd., 59 (5),76 (1963).
(10) Eckert, J. S. et al., Ibid., 62 (l),59 (Jan. 1966).
(11) Eickmeyer, A. G., Ibid.,58 (4), 89 (1962).
(12) Higbie, Ralph, Trans. A.I.Ch.E. 31,365 (1935).
(13) Kohl, A. L.,Buckingham, P. A., Pet. Ref. 39 (5), 193-6 (May 1960).
(14) Lewis, W. K.,Whitman, W. G., IND.ENC.CHEM. 16,1215 (1924).
(15) Riesenfeld, F. C.,Mullowney, J. F., Pet. Ref. 38 (5),161-8 (May 1959).
(16) Spector, N. A., Dodge, B. F., A.I.Ch.E. 42, Nos. 5-6, (0ct.-Dec. 1946).
(17) Tepe, J.B., Dodge, B. F., Trans. A.I.Ch.E. 39 (Z), (255).
(18) U. S. Stoneware Co. Data Sheets.
(19) Whitman, W. G., Chpm. &? Mrt. Engr. 29 (4),16C (1923).
VOL 5 9 NO. 2 FEBRUARY 1 9 6 7 47