Professional Documents
Culture Documents
OF PACKED
COLUMNS
Total and static holdups have been measured for %-, and 1-in. ceramic Berl
saddles, %-, I-, and I%-in. ceramic Raschig rings, and I-in. carbon Raschig rings
with air rates from 100 to 1,000 lbJ(hr.1 (sq. ft.) and water rates from 1,ooO
to 10,000 lb./(hr.) (sq. ft.).
The holdup measurements and motion picture observations of the flow of dye
solutions through packings provide an explanation for the great differences observed
when gas-phase mass transfer rates are measured by absorption and vaporization
methods. If the effective interfacial area for vaporization is assumed to be pro-
portional to total holdup and the area for absorption is assumed proportional to
operating holdup, the ratio of the two mass transfer rates should be equal to the
ratio of the two holdups.
The departure from equality of the two ratios can be explained by the observation
that the static holdup is displaced slowly, resulting in additional effective area for
absorption over that expected from the operating holdup alone.
In recent years investigations of literature(8,ll). The total holdup, 19,ZZ) have measured or estimated
mass, heat, and momentum trans- h,, defined as the total liquid in the liquid holdups, and two extensive
fer in fluid-soild systems have in- packing under operating conditions, studies have been reported by
dicated that the free space, or void is expressed as cubic feet of liquid .
Elgin and coworkers ( 3 , l l ) Un-
fraction, E , is an important variable per cubic foot of packing. The fortunately the previous investi-
in the correlation of data(l,.4,5, static holdup, h,, as used in this gations have dealt with operating
6,10,12). To apply these findings work is defined as the liquid in the holdups almost exclusively and no
t o an analysis of more complex packing which does not drain from investigation has been reported of
systems such as those occurring in the packing when the liquid supply the t h r e e types of liquid holdup in
packed absorption and distillation to the column is discontinued and packings for which extensive mass
columns, one must know the total is also expressed as cubic feet of transfer data are available, namely,
liquid holdup to determine the void liquid per cubic foot of packing. Raschig rings and Berl saddles.
fraction available for gas flow un- The operating holdup, h,, defined The object of the present work
der operating conditions. A knowl- as the difference between the total is the determination of the three
edge of liquid holdups is also of and static holdups, represents the types of liquid holdup f o r 0.5-, LO-,
interest in the study of batch dis- liquid which will drain from the and 1.5-in. unglazed white porcelain
tillation and effective interfacial packing and is also a measure of Raschig rings, 0.5-, and 1.0-in. un-
areas for diffusional operations em- the liquid flowing through the pack- glazed white porcelain Berl sad-
ploying packed columns. ing when the column is in opera- dles, and 1.0-in. carbon Raschig
Three different types of liquid tion. The relation between the three rings. These data will be used to
holdup have been discussed in the holdups is given by explain the differences observed in
gas-phase mass transfer rates when
Additional tabular material may be obtained
from the Photoduplication Service, American
ht = h, + h,. vaporization and absorption tech-
Documentation Institute, Library of Congress, niques are used. In P a r t I1 of this
Washington 25 D.C., as document 4565 for
$1.25 for micrdfilm or photoprints. Several investigators (2,9,13,15, paper the data will be applied to
the problem of separating t h e volu- and DeGouff(l8) except for minor packing, consisted of eight arms ex-
metric mass t r a n s f e r coefficients modifications, which will be discussed. tending radially from a special fitting
(k,a and k,a) into their com- A water tank and pump were in- and was 9 in. in diameter. I t had a
stalled to permit adjustmefit of water total of 160 0.1-in. holes evenly spaced
ponents, t h e effective interfacial temperature and recycling of water. in three rows on four alternate arms
area, a, and t h e mass t r a n s f e r Two calibrated rotameters were in- and in a single row on the remaining
coefficients, kG and kL. stalled to replace the orifice meters arms. Several holes were plugged with
previously used to meter the air. The solder as the result of tests made to
EQUIPMENT 10-in.-diam. by 36-in. packed glass col- ensure uniform distribution of liquid
umn was arranged so that it might be for the liquid rates employed.
The equipment used for these weighed with its contents while in
studies, shown schematically in Fig- operation. This was done by sup- PROCEDURE
ure 1, is the same as that described in porting the column by means of four Preliminary tests made possible
detail in a recent paper by Shulman 0.25-in. steel rods connected at the the establishment of a procedure
top by a yoke and at the bottom by which gave reproducible data even
a steel ring on which the glass col- when the column was emptied and
umn and the four legs of the packing repacked. When the column was
support plate rested. In operation the packed with dry packing and the
DIAGRAM
FIG.1. SCHEMATIC OF
1 SCALE
glass column was suspended in air
except for a few inches a t the bot-
tom which were immersed in a water
seal. The column and its contents
were counterbalanced by weights on
a pan connected t o the yoke a t the
top of the column by the lever arm
and knife edges obtained from a
platform scale. The pan of weights
rested in turn on the pan of a 20-kg.-
capacity platform scale which could
water rate was set at a fixed value,
the total holdup was found to in-
crease with time until a constaxt
equilibrium value was reached. The
same equilibrium value could he
reached in a much shorter time by
supplying water a t a very high rate
for a short time, to wet the packing
thoroughly, and then running a t the
desired rate. Figure 2 shows the total
holdup for 1.0-in. Berl saddles as a
APPARATUS. be read to within 1 g. The weighing function of time when the two meth-
system was capable of detecting a ods are employed. The effect of time
change in weight of the column of is even more noticeable with the
0.10, , , I , , , , , , , , I , ] , a s little a s 3 g., and it was therefore smaller packings, which have greater
possible to determine holdups to holdups. The buildup with time is
within 1% under the poorest con- believed to be due t o the gradual
ditions, Le., static-holdup determina- accumulation of water in pockets at
tions of the larger packings. The points of contact, inside rings, or on
column was open at me top to the portions of the surface which are not
atmosphere, and so a draft gauge continually covered by flowing liquid.
employed for reading the pressure at The splashing of liquid as i t falls
the base of the column measured the from one piece of packing t o another
pressure drop across the packing. and the random motion of the mov-
- 0 D
. (PO 110 1.0 300
The weighings were corrected for ing liquid over the surface of the
m.
TWE.
the buoyancy effect of the portion of packing not only causes this gradual
FIG. 2. TOTALHOLDUPVS. TIME AT the column in the water seal and for accumulation, until an equilibrium
L = 3 , 5 0 0 FOR 1.O-IN. BERL SADDLES. the upward thrust due to the pres- value is reached, but also results In
sure drop across the column, which a slow turnover of the liquid in what
was operated at all times with 28 might otherwise be stagnant pockets.
in. of packing using water a t 1 9 O to Observations based upon motion pic-
25°C. and saturated air of approxi- ture studies of the flow of dye solu-
mately the same temperature to pre- tions in packings will be discussed
vent vaporization of water. later in connection with the interpre-
The water distributor, which was tation of mass transfer data.
supported independently of the col- The total holdup was found to vary
umn approximately 1 / 2 in. above the with the density of the packed bed
FIG.4. TOTALHOLDUP
I N 0.5- FIG.5. TOTALHOLDUP
I N 1.0- FIG.6. TOTALHOLDUP I N 1.5- FIG.7. TOTALHOLDUPI N 0.5-
IN. RASCHIGRINGS. IN. RASCHIGRINGS. I N . RASCHIGRINGS. IN. BERLSADDLES.
noticed that different packings re- FIG.8. TOTALHOLDUPI N 1.0- FIG.9. TOTALHOLDUPI N 1.0-
quire different lengths of time to IN. BERL SADDLES. IN. CARBON RINGS.
drain t o fairly constant weight.
On the basis of these preliminary
tests the following procedure was
adopted to ensure obtaining repro- brium was reached and data for a the static holdup, as will be shown
ducible data. The air was saturated drainage curve were recorded. It was later. At low liquid rates the total
and cooled to the temperature of the soon found that static holdup was
water in the tank. The weight of the holdup is almost independent of
independent of liquid and gas rates gas rate up to the loading point,
column and the dry packing was ob- when this procedure was employed,
tained and the packing was thorough- and so i t was not necessary to in-
but for higher liquid rates, espe-
ly wetted by setting the water rate vestigate the two ranges so thmough- cially for the smaller packings, and
up t o over 10,000 lb./ (hr.) (sq.ft.) for ly as for total holdup. for gas rates a t approximately the
a minimum of 1 hr. The water rate loading point the effect of gas rate
was then decreased to the rate to be in- cannot be ignored. This sudden in-
vestigated and the air rate was set at EXPERIMENTAL RESULTS crease in holdup as gas rate in-
100 1b.i (hr.) (sq.ft.) When the plat- Total Holdup. The data obtained creases causes the break in the
form scale indicated constant weight from the total‘ holdup measure- pressure-drop curve which has been
f o r more than 3 min. the readings of ments are plotted in Figures 4 t o taken as the definition of the load-
weight, pressure drop, and tempera- 9 f o r the six packings studied. The
ture were taken and the air rate was ing point. It is interesting t o note
increased to the next higher value. experimental points are shown, and t h a t the total holdups for 1.0-in
For each packing a series of runs was additional lines, obtained by in- carbon rings are greater than those
made in this fashion a t constant liquid terpolation, extrapolation, and cross f o r porcelain rings of the same
rates with increasing gas rates. The plotting, have been added t o make size. This difference is accounted
highest liquid rate, 10,000 lb./ (hr.) the plots more useful f o r estimat- for by the differences in static
(sq.ft.), was used first, and the gas ing total holdups in t h e ranges of holdup for the two packings; thus
rate was increased up to 1,000 lb./ gas and liquid rates covered. The the operating holdups are of the
(hr.) (sq.ft.) or the flooding point, loading-point line on each plot was same magnitude. Apparently the
whichever was reached first. The obtained from the data of Tillson
lowest liquid rate employed was 1,000 nature of the surfaces causes a
Ib./ (hr.) (sq.ft.) because i t was felt (21) f o r the same packing. The greater accumulation of semistag-
that good liquid distribution could shapes of the curves are quite simi- nant water in carbon than in por-
not be ensured a t lower rates. lar to those obtained by Elgin and celain; therefore the total and
F o r the static holdup tests a simi- Weiss ( 3 ) for operating holdups. static holdups reported here prob-
lar procedure was followed, but the This is t o be expected because the ably do not apply t o surfaces other
water supply was cut off when equili- two holdups differ by a constant, than ceramics similar to unglazed
4.F '
10
W h o
'
PO
I
3.0
I value for ammonia absorption in
water a t the same gas and liquid
rate, as obtained by Fellinger (7).
FIG. 12. AMMONIA ABSORPTION
SULFURIC
ACID.
IN
FIG.11. (kGa),,aD/
(kGu)absVS. k,lh,. The ammonia and water data were
corrected to a common basis by Another packed-column mass
correcting f o r gas diffusivity t o transfer anomaly which may be
limited amount of data suitable for the 213 power, employing the values
this purpose. An excellent set of explained eventually with holdup
of 0.236 (cm.sq.) / (see.) for am- data is the reason for the different
data given by Fellinger(7') for monia and 0.256 (cmsq.) I (see.)
ammonia absorption in water can effects of gas diffusivity obtained
for water a t 25°C. and 1 atm. It when vaporization and absorption
be corrected f o r liquid-phase re- can be seen t h a t the points fad1 in
sistance by means of the Sherwood techniques are employed. A large
a band which is relatively narrow amount of mass transfer data ob-
and Holloway (16) correlation and when one considers that the magni-
for gas diffusivity to the 213 power tained with aQ sorts of materials
tude of some of the ratios repre- and techniques indicates that the
to provide k,a for absorption for sents otherwise unexplainable dif-
a11 the comparisons desired. Re- gas-phase mass transfer rates
ferences of well over 200%. The should be proportional to gas dif-
liable vaporization data are diffi- best line through the points can be
cult to obtain because end effects fusivity raised to the 213 power.
represented by However, two seemingly reliable
are extremely important when the
short packing heights required are sets of data obtained by vaporiza-
employed. Surosky and Dodge(Z0) tion techniques by Surosky and
give water-vaporization data f o r Dodge(20) and Mehta and Parekh
1.0-in. carbon rings which have The broken lines on the diagram as reported by Sherwood and Hollo-
been corrected for end effects. Sher- 15% above and below the best line way(l6) indicate the power to be
indicate the magnitude of the 0.15 to 0.17, The vaporization tech-
wood and Holloway (16 ) report
agreement of most of the points. nique assumes that the effective
water vaporization data for 1.5-in.
The average deviation from the interfaaial area is the same when
ceramic rings for which end effects
line given by Equation (7) is 2 different liquids wet the packing.
are estimated to be 35 to 40%. An 8%. The inequality of the two
equation was given for the uncor- In view of the fact that static hold-
rected data by Pigford and Colburn ratios may be caused by the failure ups are a function of surface
of the effective areas to be directly properties such as interfacial ten-
(14) which is equivalent to proportional to the holdups as as- sions, as illustrated by the dif-
sumed, a different proportionality ferences obtained with carbon and
for the two areas, surface tempera- porcelain, one would expect Iiquids
ture variations making one area with different physical and chemi-
Applying a correction of 37.5% less effective than the other, and cal properties to wet solid surfaces
based on the investigators' esti- the possibility that the slow, but differently. At the present time this
mate of end effects results in the finite, turnover of the semistag- explanation cannot be tested ; how-
corrected equation nant liquid makes some of its area ever, work underway to study the
effective for absorption with water effects of surface tension, viscosity,
work. The departure from equality and the nature of solid surface on
is not great, however, and i t seems holdups may provide the required
This equation as well as the cor- a s if Equation (7) can be used data.
rected experimental data can be empirically with Fellinger's data
used for the comparisons. Some and the holdups to predict mass SUMMARY
data for the absorption of ammonia transfer rates for vaporization and Total, static, and operating hold-
in sulfuric acid by use of 1.0-in. absorption with concentrated ab- ups, obtained for six ring and sad-
carbon rings were obtained by sorbents. dle packings, are represented by
Doherty and Johnson and reported A test of the ability of Equakion Figures 4 to 10 and Equation (1).
by Sherwood and Holloway ( 1 6 ) . (7) to predict the limiting values Equations (21, (31, and (4) may
The experimental points of the of kGa for a kGa vs. acid concen- be used to estimate holdups up to
f o u r sources mentioned were used tration plot is shown in Figure 12. loading conditions for rings and
if the combination of liquid and The acid-absorption KGa data are saddles larger than those investi-
gas rates fell in the regions in- plotted, and Fellinger's water-ab- gated. Total holdups are approxi-
vestigated in the holdup work re- sorption value of k,a as well as mately proportional to the number
ported in Figures 4 to 9. It was the k,a obtained by multiplying i t o f pieces of packing per cubic foot.
felt t h a t excessive extrapolation of by 0.85 (h,/h,) are shown as limit- Static holdups are independent of
the holdup curves was not justified ing values. It can be seen that all gas and liquid rates but are de-
because poor liquid distribution a t the experimental data fall between pendent upon the nature of the
low liquid rates and rapidly chang- the two as expected, and a t high surface of the paekings; operating
ing holdups a t gas and liquid rates concentrations there is close agree- holdups are independent of the
above loading would add to the ment with the limiting value. packing surface.
A study was made of separating the volumetric mass transfer coefficients, kau and kLa, into their components ko, h,and
u so that the effects of variables might be determined separately for each component. Mass transfer rates for four packings, %-
and 1%-in. Raschig rings and Yz-in. and 1-in. Berl saddles, made of naphthalene, were determined by vaporization into air
at gas rates from 100 to 1,OW lb./(hr.) (sq. ft.).
The correlation for ka was used to determine the wetted areas of those packings when irrigated with water and to calculate
the effective interfacial areas, u, from Eellinger’s data for ammonia absorption. These effective areas were then used to evahate
RL from previously published kLu data, and a correlation was obtained for all packings.
The correlations far ka and k~ and the effective-interfacial-area data make possible a more rigorous method for the design
of packed columns than was heretofore available.
To predict the performance of liable data such as those of Fellin- by the following relation t o give
columns employing ring and saddle ger(4) for volumetric gas-phase for a limited number of packings
packings for design purposes re- coefficients, kGa, and of Sherwood the over-all mass transfer coeffi-
~ and Holloway(8) f o r volumetric cients desired for design.
Additional tabular material may be obtained liquid-phase coefficients, &a, a r e
from the American Documentation Institute
Auxiliary Publications Project Photoduplication
used. Such coefficients can be esti- -
1 - 1 1 - 1
(0
Service Library of Congress’ Washington 25 mated f o r aqueous Systems from KGa kGa HkLa HKLa
D.C., a‘s document 4566 by r;mitting $2.50 fo; these sources of data and combined Although these sources of data are
photoprints or $1.75 for microfilm.