Beruflich Dokumente
Kultur Dokumente
IN PACKED TOWERS
J. C. ELGIN AND F. B. WEISS’
Princeton University, Princeton, N. J.
The effect of a wide range of air and water rates, With or without gas flow, it varies linearly with
including the flooding zone, on the operation of a liquid rate except a t low flows of the latter. Meas-
3-inch-diameter glass tower successively filled urements of the effect of gas and liquid rate on the
with four types of packing has been studied pressure drop and liquid holdup a t the flooding
visually, and pressure drop and liquid holdup were point are presented.
determined simultaneously. The performance of PIotting flooding data as the square root of gas
a packed tower in the flooding zone is described velocity against that of the liquid on rectangular
more definitely than hitherto, and a visually de- coordinates gives a straight line for any one packing
termined flooding point characterizing a definite with all available data and offers a useful method
state of operation is defined. This visual flooding of comparison and interpolation. Reasonable
point is shown to correspond closely to character- agreement with the correlation method of Sher-
istic behavior of pressure drop and liquid holdup. wood, Shipley, and Holloway is found. A differ-
Data for flooding velocities of the four packings, ent empirical method of plotting gives just as good
both a t high and low rates and confirmed by the results with available data on similar size packings,
three methods of determination, are reported. and substantiation of a correlation method de-
With no gas flow and sufficiently high liquid rates, mands further systematic data determined by a
packings flood in essentially the same manner as reliable method on packings whose properties as
if gas is flowing. used have been accurately measured. The gener-
Except for combinations of low packing free ality and fundamental similarity of the flooding
space and high liquid rates, holdup is not appre- phenomenon in apparatus involving counter-
ciably increased by gas velocity until just before current flow of two fluids are emphasized, and a
the critical point is reached, when it rises sharply. mechanism to explain its basic cause is suggested.
Apparatus
Apparatus and flow lines are shown in Figure 1:
The tower was constructed of 3-inch-diameter (2.89-inch
i. d.) Pyrex glass tubing, 5 feet in over-all height and flanged at
the ends. To the lower end was attached a conical holdup
receiver constructed of sheet metal and designed to permit the
satisfactory discharge of water from the bottom of the column
in normal operation, the collection of holdup for measurement,
and a t the same time suitable distribution of the gas entering
the tower so as to avoid interference with the discharge of liquid
as far as possible. The gas was conducted from the top of the
tower through a tapered sheet metal pipe. Both exit devices
were connected to the ends of the tower by flanged joints in
which 0.25-inch brass plates and proper gaskets to ensure a
tight joint were inserted. Copper tubing connected to 0.19-
inch holes drilled in these brass plates led to a manometer on
which the pressure drop across the tower was measured.
Air was supplied to the bottom of the tower through a filter
and control valve provided with a needle valve by-pass for close
regulation, and from the top it passed through a trap to remove
FIGURE
1. DIAGRAM
OF APPARATUS
entrainment and a calibrated orifice for metering. Water was
supplied from storage to the top of the tower by a pump, first
passing through a calibrated orifice for metering. It entered
found that all experimental points approximated a single the tower through a shower head with enlarged openings placed
curve. Sherwood and co-workers obtained a satisfactory approximately 3 to 4 inches above the packing level and adjusted
in operation so as to distribute the water evenly over the packing
correlation of their data on air-water, hydrogen-water, carbon without touching the tower walls. The water issued a t the
dioxide-water, and air with various glycerol and butyric bottom through the holdup receiver as fast as it was received,
acid solutions and methanol in a 2-inch glass column by a and passed through a U-seal and valve to an auxiliary tank from
log-log plot of which it was returned by pump to storage. Inlet and exit water
lines were provided with quick-opening valves to permit the
simultaneous stoppage of water to and from the tower and the
collection of the quantity draining from the packing for weighing.
The water inlet was arranged to discharge only a constant pre-
determined quantity of water after stoppage. A small quantity
retained in the exit pipe between quick-opening valve and offtake
They also found a satisfactory general correlation of the data to weigh tank was determined in advance, and the holdup was
for dumped rings and miscellaneous packings in large columns corrected accordingly.
The column was packed successively with nominal 0.5-inch
reported by Uchida and Fujita (IO), White ( I 2 ) , and Baker, clay Berl saddles, 0.625-inch (5/*-inch) clay Raschig rings (0.63-
Chilton, and Vernon ( 2 ) by this method. This general cor- inch 0. d., 0.31-inch i. d., and 0.63 inch long), 0.5-inch clay balls,
relation curve in large columns varied from that given by and 0.25-inch porcelain saddles to a height of 4.67 feet. The
their own data by a factor of approximately 2, which was at- packing support was a wire screen of mesh just sufficiently
smaller than the packing to retain it. The characteristic proper-
tributed t o wall effect: “The flooding condition was ascer- ties of the packings as actually employed in the tower are given
tained by visual observation of the liquid flowing over the in Table I, in which the method of determining each property
packing and down the walls of the column.” is also indicated. The drained free void in each case was de-
The correlation employed by these latter investigators termined by filling the tower to the top level of the packing and
measuring the water draining from the packing with proper
is based on the original suggestion of Sherwood (7) and of correction for that retained in the holdup receiver below the
Walker, Lewis, McAdams, and Gilliland ( I I ) , who obtained bottom level of the packing and the offtake. The saddle pack-
a reasonable correlation of a few of the available data by a ing was obtained from the Maurice A. Knight Company, the
g r n U O / U L . The latter authors
plot of p ~ U ~ ~ / p ~ 2against 0.5-inch size from a supply received in 1935, the 0.25-inch
porcelain, in 1937. The manufacturers recently stated that up
suggested that flooding in packed towers is caused by the to the present time there has been considerable variation in
friction of the rising gas holding u p a certain amount of the uniformity between different batches of saddle packing of the
liquid in the tower, and regarded the flooding velocity as the same nominal size. Since the rin packing was from a supply
point where the friction per unit length equals the hydro- obtained from the United States honeware Company in 1933,
it appears likely that this also may not be representative of
static liquid head in the tower. Mach ( 5 ) found, as a general more modern material. The tower was filled in each case by
rule, that the flooding velocity corresponds approximately t o dumping. Unfortunately it was learned subsequent to the
a pressure drop between 2 and 3 inches of water. Badger completion of the ex eriments that in the case of the 0.5-inch
and McCabe ( I ) concluded that the lower break in the saddles and 0.625-inci rings the packing had been leveled off and
otherwise adjusted to fill in large free voids appearing adjacent
AP-U, curve (loading point of White) represents ‘(that to the walls a t intervals during the process of packing. Because
velocity a t which drops of liquid are carried u p out of the of the relatively small tower diameter in relation to the three
tower by the gas” whereas the upper break or flooding point larger packing sizes the existence of “wall effect” to some degree
represents (‘the gas velocity a t which the liquid cannot pene- is also probable. These factors introduce uncertainty as t o the
extent to which the present pressure drop, holdup, and flooding
trate the tower but is held as a layer on top of the packing.” data would be reproduced in absolute magnitude by the same
They also concluded that these two points are independent nominal size and types of packing of recent manufacture in
APRIL, 1939 INDUSTRIAL AND ENGINEERING CHEMISTRY 437
regular commercial use in large-scale towers. They do not, and within these limits for the other packings. All experiments
however, affect the characteristic behavior of the present pack- were made a t room temperature, which resulted in air and water
ings, the relation of the present data among themselves, or the temperatures varying between 19" and 23" C . over the course
conclusions to be drawn. of the experiments.
Visual Observations of
Column Operation
I. DATAON PACKING
TABLE AS USEDIN THE TOWER ,#, I n normal operation well below flood-
0.5-111, 0.25-11. 0.625-In. ing, water flowed quietly down over the
0.5-In.
Clay Berl Porcelain Clay Raschig Clay
Type of Packing Saddles Berl Saddles Rings packing in a continuous film, gas being
Balls
No. packing units/cu. f t . of towera 11,650 118,700 4750 the continuous phase. When the gas
13,200
Packing surface 8,sq. ft./cu. f t . of packed volume
Fractional drained void, cu. ft./cu. f t . of packingb
llOd
0 543
285d
0.500 0.442
66s o,3ii8 rate was gradually raised with the water
Fractional dry void, cu. ft./cu. f t . of packingo 0.75 0.513 0.624 0.500 rate constant, no change was a t first
a Calculated from weight of packing in tower and weight of 100 packing units. visible but a point was eventually
b Calculated from the weight of water draining from t h e packed section after filling it.
C Dry void = 1 - [(no. of packing units per cu. ft.) (volume of one packing unit)].
reached at which the tower slowly filled
d Calculated from manufacturer's figure. with water. Filling appeared to com-
6 Calculated from the average dimensions of a packing unit.
mence in the lower section of the tower
in the majority of runs, and with gas
flow maintained at the same constant
Procedure value, the water rose to a stationary level somewhere above the
Prior to quantitative measurements of pressure drop, holdup, packing. The filling or flooding process could be followed by
and flooding, the operation of the tower with each packing was an accompanying gradual rise of pressure drop (although gas
carefully studied visually. The water rate was set a t a pre- velocity was constant) until the water level stabilized a t a
determined constant value, the gas rate was adjusted to the constant point above the packing. The height of this head
desired value, and the behavior of the tower was then observed. of water above the packing could then be raised by further
The gas rate was increased in small increments until the tower
flooded, and the observations were repeated a t each value and hcrease in the gas flow. With extreme regulation on the gas
a t a series of water rates over the permissible range of operation. line and patience, the gas flow could be set so as to stabilize
Preliminary observations showed that the tower did not imme- the water level just above the top of the packing. With no
diately stabilize after the flow rates were set, but that pressure amountof effortwas it possible to adjust the gas flow so that
drop and holdup tended to vary to a constant value. I n all
subsequent experiments the tower was allowed to run con- a water level stabilized at intermediate positions in the pack-
tinuously for 30 to 60 minutes after flow rates were set, before ing. Either the water rose t o the packed level or the excess
pressure drop and holdup measurements
were recorded. This was particularly evi-
dent under certain conditions in the flood-
~~
ing zone where flooding did not appear TABLE 11. FLOODING P O I N T DATAFOR AIR-WATER SYSTEM AT 21" * 2" c.
until the elapse of an appreciable period -UQ a t Flooding- 7 AP/N--- Increase in H
subsequent to flow adjustment; in some From From H H at Fc/Fw at Upper Com- a t Flood over
cases times more than 30 minutes were L
AP/Nus. us. UO Complete Complete break plete H a t Zero
required for completion. Visual Uo plot plot Flooding Flooding point flood Gas Flow
Procedure in making simultaneous meas- 0.5-Inch Clay Saddles
urements of pressure drop and holdup was 733 2.91 .. ... ... ... .. 30.4 ..
similar except that the data a t each flow 1,100 2.93 ... 30.2
condition were the result of an independ- 1,160 2.64 2:55 2.55 1o:io 0:697 9: 5 28.6 7: 37
ent run. After the lapse of 30 to 60 min- 2,020 2.32 .. .. 28.3 ..
2,080 2.31 .. .. 28.8 ..
utes following the adjustment of flow rates, 2,380 2.03 .. ... ... . . . .. 28.1 ..
during which gas and liquid flow were 3,230
3,340
1.76
1.75
.. ...
...
... . . . .. 27.8
27.7
..
maintained constant, pressure drop, water, 4,210 1.42 1:38 1.42 13:25 0:6o8 6:3 26.4 6:63
and air flow meter readings were recorded, 5,140 1.48 .. .. 26.7 ..
the inlet and exit water lines were closed, 5,570 1.31 .. .. 26.2 ..
and the water held in the packed section 6,380
7,350
1.37
1.20
..
.. ... ...
..
....
25.9
26.3
..
..
was allowed to drain into the holdup re-
ceiver for 15 minutes, drawn off, and
7,860
8,140
1.14
1.13
..
..
...
...
...
...
I
...
.
...
.
. .
24.5
23.6
..
..
weighed. Small predetermined correc- 8,410 1.09 ... 28.3
tions were applied to this quantity of 8,800 0.89 0:88 0.83 15:so 0.'533 4 :2 21.9 6 .'69
13,400 0.50 0.50 0.46 18.02 0.466 5.7 23.4 6.57
water to correct for that draining from 18,100 0.39 ... 19.10 0.436 24.0
the inlet line and retained in the receiver 16,610 0.38 0:33 0.37 19.65 0.420 419 24.6 6145
19,580 0.25 0.21 0.23 21,30 0.370 5.5 25.0 6.57
between quick-throw valve and drawoff. 22,700 0.09 .. ... 22.00 0.350 .. 26.4 ..
Water level in the discharge line below 28,000 0.034 .. ... 25.10 0.225 .. 25.5
.. ..
the column was maintained a t a constant 33,2005 0 .. ... 21.60 0,363 .. ..
predetermined level on a gage glass during 40,7006 0 .. ... 23.20 0.327 .. .. ..
the runs. For the next measurement a t a 0.625-Inch Clay Raschig Rings
different flow condition, the water flow was 4,070 1.66 1.59 1.64 6.56 0.762 7.4 13.4 2.37
reset a t the same value and the gas rate 4,070 1.68 . . I ... 13.5
at a higher value, and the above procedure 6,230 1.35
1.13
1:33
1.08
1.32
1.09
7.64
8.46
0:722
0.692
7:l
7.0
13.2
12.5
2:45
2.32
was repeated for a series of gas rates a t 8,160
11,460 0.75 0.72 0.64 10.10 0.633 7.6 12.3 2.72
each of a series of constant water rates 15,780 0.50 0.49 0.46 11.74 0.573 9.3 12.4 2.46
extending through the flooding zone of the 20,600 0.25 0.24 0.24 14.61 0.468 13.2 16.1 3.55
packing. O.5O-Inch Clay Balls
Holdup determinations a t each water 20.4 2.86
rate with no air flow were usually made 3,170 1.29 1.27 1.26 5.32 0.740 9.0
5,030 0.84 0.79 0.81 7.10 0.654 5.5 19.7 3.55
prior to those with air circulating. Flood- 7,880 0.56 0.55 0.55 8.06 0.607 6.8 19.8 3.52
ing observations were made with no air 11,480 0.33 0.31 0.31 9.97 0.514 10.1 22.4 3.55
flow, both by starting with an initially dry 0.25-Inch Porcelain Saddles
packing and by first flooding the column
a t a high water and low air rate, shutting 1,928 0.50 0.50 0.495 19.05 0.593 5.5 40.5 2.96
off the air, and then increasing the water 3,480 0.41 0.34 ... 21,55 0.513 3.5 39.2 2.48
6,500 0.24 0.24 0.24 24.65 0.215 7.5 47.2 2.06
rate to flooding. Determined with packing initially dry.
0
Water rates varied between approxi- b Determined by flooding the packing a t a very low gas flow, discontinuing the gas, and increasing
mately 730 and 41,000 pounds per hour per t h e liquor flow t o flood.
square foot with the 0.5-inch clay saddles
438 INDUSTRIAL AND ENGINEERING CHEMISTRY VOL. 31, NO. 4
Pressure Drop
Pressure drop data on the four
packings are presented graphically
in Figure 2 as the customary log-log
plot of pressure drop per foot of
packing vs. apparent gas velocity
a t constant liquid rates. The ma-
jority of the data were taken in the
vicinity of flooding since this was
the point of present interest. I n a
few cases measurements were ex-
tended well below this. Following Dotted ourves show holdup at the visually defined flooding point.
440 INDUSTRIAL AND ENGINEERING CHEMISTRY VOL. 31, NO. 4
drop bears a different relation to gas velocity a t high liquor with the critical flooding point of the tower previously
rates under conditions of large holdup and probably low described, and confirms the view that this is the true flooded
packing free space. point of a packed tower.
The pressure drop curves of Figure 2 show the same be- I n general, depending upon the location of the experimental
havior for all four packings and agree in type with similar points, there is obviously some latitude in locating a break
measurements of other investigators on various packings. point. The latter and its proximity to the true flood point
The data for the 0.5-inch saddle and the ring packing appear, are determined primarily by the size of the increments by
however, to be appreciably higher than those available for which the gas velocity has been varied for the pressure drop
similar packings of the same nominal size. Similar dis- determinations in the flooding Bone. The break points, as
crepancies appear between data of other investigators. I n shown in Figure 2 or Table 11, always fall slightly below the
the present case this might be explained by “wall effect” or true flooding point of the tower. They may be regarded as
probably by differences in the packing. It does not seem setting a lower limit to the flooding velocity which does not
impossible that differences in the method of measuring the vary appreciably from it if pressure drop measurements have
flow of gas to the tower may account for some divergence. been properly located.
For example, with the orifice before the tower, the conditions The authors conclude from a careful analysis of available
under which the gas is measured obviously vary with pressure data that the log-log plot of pressure drop vs. gas velocity is
drop in the tower and depth of packing. Other experiment- in reality a continuous curve such as is represented schemati-
ers, as for the present data, have measured the gas volume cally by the solid curve ABHC,Figure 3. Flooding is visual-
after leaving the tower. I n either case the linear velocity ized as a continuous process rather than an abrupt transition,
of the gas varies in the packing (although its weight rate of commencing gradually somewhere near B and continuing a t
flow is constant) and bears a complicated relation to the an accelerating pace until completed a t C where liquid fills the
volume measured a t the orifice and the calibration of the tower to the level of the packing. According to this view, a
latter. Whether the linear gas velocities calculated by dif- zone of flooding exists rather than a distinct flooding point.
ferent investigators are on a comparable basis is unknown. The velocity a t the upper limit marking the completion of
the process and characterizable as described above, may,
Flooding Velocity from Pressure Drop however, be so regarded. Curve AKHC, composed of
straight lines, is drawn to show two break points, K and H ,
The highest point on each curve of Figure 2 is the pressure and indicates the common method of representing such data.
drop obtained a t the visually observed flooding point de- The narrow range of the flooding zone under commonly
fined above. The upper break point on the pressure drop studied conditions, together with the usual method of plotting
curves a t each water rate has been located as accurately as on logarithmic coordinates, makes this possible. Points on
possible in accordance with the procedure of Mach (6) and the sharply rising portion of the pressure drop curves above
White (22). Gas velocities from the curves of Figure 2 the break point, Figure 2 (portion HC of Figure 3), appear a t
corresponding to this point, defined by White as the flooding random. This is also the case in the data of other investiga-
velocity of the tower, are recorded in column 3 of Table 11. tors and is to be ascribed to the sensitivity of pressure drop
The agreement between the values thus obtained and those to gas velocity in the flooding zone, combined with the
measured visually demonstrates clearly that the break point spacing of pressure drop measurements a t widely separated
of the pressure drop curve properly located coincides closely gas velocities in this region.
SHOPASSEMBLY VIEW
OF THE WORLD’S FIRST
F U L L YA U T O M A T I -
CALLY CONTROLLED
SOLVENT RECOVERY
ADSORPTIONSYSTEM;
THE CONTROL MECHA-
N I S M SHOWN H A S
BEEN SIMPLIFIED AND
REDUCEDIN SIZETO
FIT INTO A SMALL
CASE FOR PANEL
MOUNTINQ
Courtesv, E. L. Luaces
Associates
APRIL, 1939 INDUSTRIAL AND ENGINEERING CHEMISTRY 441
Pressure drop readings were unfortunately not continued mutually dependent. Some parallelism between the be-
after the visually defined flooding point was attained. Hypo- havior of the holdup and pressure drop curves can be dis-
thetical curve CR of Figure 3, representing increase of pres- cerned. Simmons and Osborn (9) and Furnas and Bellinger
sure drop with gas velocity as the liquid head builds up (4) reported that holdup in packings is unaffected by gas
above the packing, is intended t o indicate what may happen flow, but they gave no measurements and evidently em-
in locating erroneous break points ( P or 8) if measurements ployed low flow rates. With 0.5-inch paper rings, Mayo,
are made a t too wide an interval on either side of the true Hunter, and Nash (6) obtained a curve similar to those in
flood point. Figure 4 for one low water rate.
Mach (6) concluded t h a t the flooding velocity (break HOLDUP AT FLOODING is represented for each packing by
point of the AP-U, plot) was attained a t a relatively con- the dotted curves of Figure 4. The tower operates less full
stant pressure drop between 2 and 3 inches of water (10 to a t the visually defined flooding point as the water rate to the
15 pounds per square foot) per foot of packing. The present packing is diminished. This behavior would be expected,
data do not substantiate this as a general rule for all flow since the packing must accommodate the passage of a cor-
conditions, although it may serve as a rough empirical guide. respondingly greater volume of gas. Calculated values of
the fractional drained void a t flooding unfilled by water are
recorded in Table 11.
When plotted against water rate, the holdup curves in each
case coincide closely with that a t zero gas flow until the up-
turn a t the beginning of the flooding zone is reached. Several
interesting facts are revealed by the holdup data, the physical
significance of which are not yet understood. With all
packings, if holdup a t the flooding point is plotted against
the water rate, the line obtained parallels almost exactly
that with no gas flow reported below; that is, water rate
,u FTJSEC.
Holdup
FIGURE
6, EFFECT
OF WATERRATEON HOLDUP IN VARIOUS
The water retention in each packing measured simul- WITH No GAS FLOWING
PACKINGS
taneously with pressure drop is shown in Figure 4 as a func-
tion of gas velocity a t different constant water rates. Except
a t high rates with the ball and ring packings which have varies the holdup a t the flood point to the same extent as it
relatively low free voids, gas velocity has very little effect on does with no gas flow. The amount by which the holdup a t
holdup until the flooding process begins. The upward turn flooding exceeds that without gas flow was calculated in the
of the holdup curves evidently marks its beginning. Once the last column of Table 11. The relative independence of water
gas flow begins to effect an increase in holdup, the process and gas flow exhibited by this quantity in the case of each
continues a t a rapidly accelerating pace since the two are packing indicates t h a t the column floods when the normal
442 INDUSTRIAL AND ENGINEERING CHEMISTRY VOL. 31, NO. 4
L I I I I I I I I I I I I I I I I I I I I 1 1 1 1 1 1 1 I I I I I I I I I I I pounds per square foot per hour their.data ap-
pear to be practically linear. Simmons and
Osborn (9) and Mayo, Hunter, and Nash (6)
also concluded t h a t holdup varied as the 1.0
power of water rate except in the very low flow
range.
Holdup in packed towers may be related to
liquor rate by
H = L/V (1)
where V defines the vertical rate of passage
through the tower; that is, 1/V is the time re-
quired to traverse one foot of packed height.
Since liquid is a discontinuous phase, V is less
than directly proportional to L and holdup must
increase with the latter. As pointed out by
Furnas and Bellinger, V can be related to VO,the
average velocity of the liquid layer over the pack-
00001 ing surface, by
0 00008
0.00006 V = Vaesina
0 00004
0.00 002 ! I Illllll I 1 IIIIIII ~ @ l l l l l l I I IIIYIIR I If the entire holdup measured by drainage is
assumed to represent a liquid layer on the pack-
0.0 000 I ing surface, the average layer thickness is given
0.01 .oz .o3.w.o6~soI 0.2 a40.6081
4 6110 20 4060~)100 zoo z l
.
--
UY
7. COMPARISON
FIGURE OF PRESENTFLOODING VELOCITYDATAWITH
CORRELATIONMETHODOF SHERWOOD,SHIPLEY,AND HOLLOWAY, SHOW- W P )
ING AGREEMENTWITH LINE FOR DUMPEDRINGS t = H/pS = (2)
S(Vosin a)
holdup of the packing a t a given water rate is exceeded by a The slope of the holdup curves of Figure 6 is 1/V. Hence,
fixed constant value. above rates of approximately 4000 pounds per hour per
FLOODING VELOCITY FROM HOLDUP.Comparison of the square foot, for a given packing the water layer must evi-
pressure drop curves of Figure 2 with
those for holdup in Figure 4 shows
that the abrupt rise of holdup coin- 2.6
cides closely with that of pressure drop.
It cannot be doubted t h a t these factors
are closely related to each other and
to the flooding process. The close
correspondence is still more evident
if the holdup data are plotted against
gas velocity on logarithmic coordinates.
This has been done for all packing,
and the plot for the 0.5-inch saddles
is shown in Figure 5. Two break
points are readily distinguished; the
highest corresponds closely to the gas
velocity a t the upper break points of
F i g u r e 2. V a l u e s of t h e flooding
velocity determined from holdup in
this way are recorded in column 4,
Table 11. It is evident that holdup
s u p p l i e s a m e a s u r e of f l o o d i n g
velocity.
HOLDUP AT ZERO GAS FLOW.Data
for the present packings together with
those reported for water with various
packings by other investigators are
presented in Figure 6. I n all cases
the holdup varies with the 1.0 power
of the water rate when the latter ex-
c e e d s a p p r o x i m a t e l y 3000 to 4000
pounds per hour per square foot. At
water rates between 430 and 6800
Furnas and Bellinger (4) found holdup
with ring and saddle packings to de-
pend on the 0.54 to 0.74 power of jI V&
the water rate; but above 3000 to 4000 1 2 4 6 8 10 12 14 16 18 20 22 2 4 ;5
APRIL, 1939 INDUSTRIAL AND ENGINEERING CHEMISTRY 443
crease of holdup. Since the countercurrent velocities are gives a straight line for any packing with all available flooding
mutually interdependent, the effect is multiplied at a rapidly velocity data. An empirical plot is shown which differs from
accelerating pace with further increase in volume of either that of the above authors but which correlates well the present
data together with most of those available on similar size pack-
phase fed to the apparatus and must eventually culminate in ings, but this is suggested primarily as a basis for future study
flooding. When t h e flooding point is reached, either of two and because i t emphasizes the necessity for additional data
things may happen. Depending upon the rate of passage of on packings whose properties as used have been accurately
the continuous phase if made discontinuous in the other measured.
The generality and fundamental similarity of the flooding
and its corresponding holdup a t the relative volume flow phenomenon in any apparatus which brings two fluids together
rates at which flooding has occurred , either the discontinuous in countercurrent flow are pointed out, and a mechanism which
phase is reversed, and the only evident change is a transition, explains its basic cause is suggested.
say, from a spray to a bubble device; or if the flow conditions
are such that the corresponding holdup of each would be Acknowledgment
sufficient to fill the available volume, a competition results
and either is discharged more or less violently from one or The authors wish t o thank t h e Maurice A. Knight Com-
both ends of the apparatus. Both conditions occur in a pany, Akron, Ohio, who donated the supply of Berl saddle
packed tower or a spray type liquid-liquid tower. Only the packings utilized in the present experiments.
latter situation is observed i n a film type or a bubble type
gas-liquid tower. T h e first may be regarded as flooding with
respect to liquid, the second, to gas. Nomenclature
T h e experimental facts for packed towers as well as those Fc = wet void during operation with liquor circulating,
known to the authors for other apparatus are in accord with cu. ft./cu. f t . of packed tower volume
this hypothetical mechanism. F D = fractional dry void in the packing, cu. ft./cu. ft. of
packed tower volume
F , = fractional drained void in the packing, cu. ft./cu. ft.
Summary of packed tower volume
g = acceleration of gravity, ft./(sec.) (sec.)
That packed towers cease to operate satisfactorily when rate G = superficial mass velocity of gas = 3600 U Q ~ Q ,
of gas and liquid feeds exceed a certain limit, usually designated lb./(hr.) (sq. ft.)
as the flooding velocity, is well known. Published descriptions H = liquid holdup in packed section, lb./cu. ft. of packed
regarding flooding characterize it variously and indefinitely, and tower volume
the relatively few data for flooding velocities do not show good IC’, IC = proportionality coefficients
agreement. Different criteria have been used for its measure- L = superficial mass velocity of liquid = U L ~ Llb./(hr.)
,
ment and these have not been related to a definite state of opera- (sq. ft.)
tion or to one another. Data on the effect of gas and liquor m = mean hydraulic radius (free vol. divided by total con-
velocity on holdup in packings extending to the flooding zone tact area)
are completely lacking. N = height of packed tower, ft.
A 3-inch diameter glass tower filled successively with four A P / N = pressure drop, lb./(sq. ft.)(ft.)
kinds of packing has been employed to study visually the effect S = surface area of packing, _. sa.
- ft./cu. ft. of Dacked
of air and water flow on the operation of such equipment up to tower volume -
the flooding point and simultaheously pressure drop and water t = average thickness of liquid layer on packing, f t .
holdup in the packing were measured. UG = superficial gas velocity based on entire tower cross
The true flooding point of a packed tower is characterized as section, f t./sec.
that at which the liquid fills the tower to the top level of the UL = superficial liquor velocity based on entire cross
packing. This is a definitely measurable and reproducible state section, f t./hr.
of operation. The behavior of the tower operated both below Y = vertical rate of liquid passage through tower, ft./hr.
and above this point has been described. Violent entrainment VO = average velocity of liquid layer over packing surface,
of liquid occurs at this point when the water rate is low, but the ft./hr.
only change at high rates is the transition from water to air as ff = average inclination of path of flow to the horizontal
the discontinuous phase. I n no case is water flow through the PG == density of gas, lb./cu. ft.
packing stopped. A packing floods in essentially the same PL = density of liquid, lb./cu. ft.
manner at sufficiently high liquor flow and no gas flow. Data P = viscosity of liquid, centipoises
for flooding velocities of the packings determined a t this visually
defined point are reported.
I n general, gas velocity has little effect on liquid holdup until Literature Cited
just before a critical gas velocity is reached, a t which it rapidly
rises. With and without gas flow, holdup varies linearly with Badger, W. L., and McCabe, W. L., “Elements of Chemical
water rate except at very low rates until the critical water rate Engineering,” 2nd ed., New York, McGraw-Hill Book Co.,
is reached, when i t rises rapidly. On a logarithmic plot of 1936.
holdup us. gas velocity, an upper break point occurs which Baker, T., Chilton, T. H., and Vernon, H. C., Trans. A m . Znst.
coincides closely with the visually defined flooding velocity. Chem. Engrs., 31, 296 (1935).
The observed effect of gas and liquor velocity on pressure Elgin, J. C., and Browning, F. M., Zbid., 31, 639 (1935).
drop in the flooding zone agrees with that of previous investi- Furnas, C. C., and Bellinger, F., Zbid., 34, 251 (1938).
gators. When represented by straight lines on logarithmic Mach, E., Forsch. Gebietezngenieurw., 6,375-9 (1935) ; Dechema,
coordinates, the data for pressure drop vs. gas velocity show an 6, 38 (1934).
upper break point a reeing closely with the visually determined Mayo, F., Hunter, T. G., and Nash, A. W., J . SOC.Chem. Znd.,
flooding velocity. I% is concluded that the break point always 54, 375T (1935).
falls slightly below the true flood point but supplies a reasonably Sherwood, T. K., “Absorption and Extraction,” 1st ed., New
accurate measure of it. York, McGraw-Hill Book Co., 1937.
Data for pressure drop and water holdu at the flood point Sherwood, T. K., Shipley, G. H., Jr., and Holloway, F. A. L.,
are given for various air and water flows. Sressure drops both IND. ENG.CHEM.,30, 765 (1938).
at this point and the break point on the logarithmic plot vary Simmons, C. W., and Osborn, H. B., Jr., Zbid., 26, 529
with gas and liquor flow and pass through a minimum. The (1934).
significance of the pressure drop at the break point and its Uchida, S.,and Fujita, S.,J. SOC.Chem. Ind. Japan, 39, 886
reliability as an estimate of flooding are questioned. (1936); 40, 238 (1937).
All packings show similar behavior. Walker, W. H., Lewis, W. K., McAdams, W. H., and Gilliland,
The data show reasonable agreement with the correlation E. R., “Principles of Chemical Engineering,” 3rd ed., New
method proposed by Sherwood, Shipley, and Holloway, but it York, McGraw-Hill Book Co., 1937.
is emphasized that insufficient unquestionable data are available White, A. M., Trans. A m . Inst. Chem. Engrs., 31, 390
to substantiate this method or to locate the theoretical line (1935).
definitely. Plotting the square root of gas velocity vs. the square White, A. M., Neel, R. M., and Welsh, A. J., chem. eng. thesis,
root of liquor velocity a t flooding on rectangular coordinates Univ. North Carolina, 1936.