33
Ind. Eng. Chem. Process Des. Dev. 1980, 19, 33-40
output. This filter helped to sharpen the response as seen in Figure 6. The standard error of estimate for this system was decreased to 0.012 with an index of determination of 0.90. The average variability in the loading was reduced to 10% by employing this device. A question that is common to such photodetector devices is the amount of solid material that is deposited on the walls of the glass monitoring section. Throughout the runs conducted, the background absorbance of the empty glass tube was continually checked. This variation in absorbance due to solid deposition reduced the background signal on the average by 5%. In general, the deposition and retention of solid particles on solid surfaces depends on many parameters: the chemistry of the solid and surface, humidity of the carrier gas, electrostatics, size, and moisture of the solid particles. For use of the proposed light/sensor arrangement, a dynamic equilibrium is assumed to be established after the solid properties remain constant during a run; this equilibrium should not be upset and the output readings of the sensor should be reliable.
Conclusions The light/sensor device is an inexpensive arrangement for monitoring flow steadiness and measuring solid/gas loading. In addition, with knowledge of the solid to gas loadings and the flow rates of transport gas, the solid flow rate can be easily determined. For fine coal the light/ sensor system gives a larger response ratio with increased loadings than coarse coal.
Acknowledgment The support of Oak Ridge Associated Universities, under whose sponsorship this study was undertaken, is greatly appreciated. Special thanks to Mr. Daniel Bienstock of Pittsburgh Energy Technology Center (DOE) are noted. His laboratory provided the facility to test the meter discussed in this paper. Nomenclature R = voltage output reading from photodetector R, = voltage output reading, base W = air flow, kg/h b$ = coal flow, kg/h Literature Cited Babh, V. I., Etkin, V. B., Teploenergetika, 21(2), 62 (1974). Epstein, P. S., Garhart, R. R., JASA, 25, 553 (1968). King, P. W., "Second InternationalConference on Pneumatic Transport of Solids in Pipes", BHRA, p D2-9, Sept 1973. LeSage, L. G., O'Fallon, N. M.. ANL-FE-49622-6 (Sept 1977). Litchfield, E., USBMIDOE, Bruceton, Pa., personal communications, 1977. Merilo, M., Dechene, R. L., Cichowias, W. M., J . Heat Transfer, 99, No. 2 (1977). Micro Motion, Inc., Coriolis Force Meter, Boulder, Colo., 1978. O'Fallon, N. M., Duffey, D., Wiggins, P. F., Bull. Am. Phys. SOC.,21, 109 (1976). Perzewlocki, K., Nizegarodcew, P., LaHouille Blanche, 28, 59 (1973). Price, C. C., Kehler, P., Sackett, J. I., Trans. Am. Nucl. Soc., 23, 117 (1976). Raptis, A. C., Doolittle, R., Papper, G. F., Fitzgerald, J. W., Carey, W. M., ANL-CT-77-1 (OCt 1976). Soo, S. L., Trezek, G. J., Dimick, R. C.. Hohnstreiter. G. F., Ind. Eng. Chem. Fundam., 3, 98 (1964). Soo, S. L., Cheng, L., J . Appl. Phys., 41, 585 (1970).
Received f o r review August 11, 1978 Accepted July 9, 1979
Fluid Dynamic Observations on a Packed, Cross-Flow Cascade at High Loadings Louis J. Thibodeaux Department of Chemical Engineering, University of Arkansas, Fayetfeville, Arkansas 7270 1
The cross-sectional areas for flow of gas and liquid in conventional countercurrent packed columns are identical. A high turndown ratio is impossible and increased loadings results in high pressure drop followed by flooding. A device of simple design consisting of a cascade of cross-flow stages was developed with flow areas decoupled. Phase flow within individual stages was opposed at 90'; however, an overall countercurrent operation is maintained. This study reports on the visual observations of the completeness of interphase contact and quantitative measurements of flow rates, pressure drop, and holdup obtained from experiments on the device. Apparently operating characteristics, which include a broad loading range and low pressure drop, will possibly make this device an attractive alternative gas-liquid contactor. Other developments have shown that liquid phase mass transfer in cross-flow packed stages is equivalent to countercurrent and that theoretical stage efficiency is between countercurrent and cocurrent flow.
Previous Work Single-stage cross-flow, fixed bed, packed scrubbers operate with the gas stream moving horizontally through the packing while the irrigating liquid flows by gravity vertically through the packing. This type of scrubber operates with a very low pressure drop and water requirements, both of which are about 40% of that required for counterflow operation. The leading face of the packed bed is usually slanted from 7 to 10" (depending upon the gas velocity) in the direction of the oncoming gas stream to ensure complete wetting and washing of the face of the 0019-7882/80/1119-0033$01 .OO/O
bed by the falling irrigated liquid. Liquid requirements range from 1 to 4 gal/ 1000 ft3 of gas, with pressure drops of from 0.2 to 0.5 in. of water per foot of bed (NAPCA, 1969). Certain of the "open" plastic shapes, such as Tellerettes and Pall rings are more adaptable to cross flow than to countercurrent contacting. The first cross-flow application was in a water cooling tower (Fordyce, 1957). Flow parallel to splash bars and shortened flow path allow a lower pressure drop than the comparable countercurrent operation. Sparse packings typify the tower fill in cooling towers. The slopes used for 1979 American Chemical Society
34
Ind. Eng. Chem. Process Des. Dev., Vol. 19, No. 1, 1980
this fill in cross-flow water cooling towers have evolved from actual test observations and not from analysis (Holmberg, 1973). The slope required is dependent upon the type of fill and on the air velocity. An angle of 10' from vertical has been satisfactory for film type fills for air velocities in the range of 500 to 700 ft/min. Dense splash (or drip) type fills need an angle of about 18' for velocities between 400 and 600 fpm. Sparse splash type fills require less slope, operating satisfactorily at 12.5O with air velocities from 400 to 550 fpm. Higher velocities require a greater angle. Oza (1974) attempted to predict pressure drop and liquid drift angle for a densely packed cross-flow cascade. Pressure drop predictions were made by incorporating countercurrent liquid holdup data into the Ergun equation for pressure drop in dry packings. Liquid drift angle was obtained from a force balance on the liquid held up in the packing. For lI2-in.Raschig rings the predicted pressure drop was roughly half of that observed in a comparable countercurrent operation. Predicted liquid deflection angles were 5' a t G = 200 lb/h ft2 to 3 5 O at G = 600. Pressure drop predictions were made for s/8-in. Pall rings which were significantly less than those predicted for 'I2-in. Raschig rings. Drift angles for Pall rings were 15' a t G = 400 and 55O a t G = 1200. Liquid flow rate increase results in slight angle increase. Pittaway (1976) obtained pressure drop and holdup data during oxygen desorption experiments (air-water) with a single stage cube packed with ll2-in. Raschig rings. A Burke-Plummer type of correlation satisfactorily reproduced the experimental data Ap = 1.76 X
G2 (1 - t ) / p C c 4
log 1000H, = -0.2767
Gas Outlet
4'
desntrarner Space (void)
-
Packed Section
Gas Turn- A .%ace
r
o
w
Front View
L
e
Gas Oefiscticn Bafriss
Side View
1 -
(1)
Liquid holdup was found to be a function of irrigation rate. A least-squares fit of the holdup data resulted in log 1000h, = 0.2128 + 0.4745 log L; G = 30 (2) log 1000h, = 0.2228
"p
+ 0.477 log L ; G = 100
(3)
+ 0.6048 log L ; G = 200
(4)
Operating liquid holdup was essentially identical with reported countercurrent data, while pressure drop was comparable. At L of approximately 9000, Ap was similar for cross flow and countercurrent flow, with G < 100, and for G > 100, Ap for cross flow was approximately twice that of countercurrent flow. At L of approximately 5000, A p for cross flow was the same or smaller than countercurrent for all G. Oza's equation for drift angle of the leading face was modified by Pittaway for the gas-phase static pressure drop with no liquid present and a term added for liquid dispersion in horizontal direction to yield
Values o f f = 20.57 and C2 = 4 x resulted in a reasonable representation of the observed drift angles for 1/2-in.Raschig rings. Measured liquid drift angles for 'I2-in. Raschig rings averaged 20" with maximum of 48'. See the Nomenclature section for dimensions of terms in eq 1-5. Experimental Methods The cross-flow cascade consisted of a narrow rectangular box constructed of Plexiglas and metal. The internals of the device were arranged to encourage gas-liquid contact in a general countercurrent fashion, with a liquid phase descending more or less in a vertical path over a packing
Scale:1 - 1
-
2 5 . 4 cm.
Liquid Outlet
Figure 1. Packed cross-flow cascade.
material and a gas phase moving upward in a path perpendicular to the liquid flow direction. Deflection baffles positioned a t regular intervals on opposite sides of the packed section divert the gas phase into the packing affecting a crisscross gas-liquid flow pattern. The box was constructed of '/*-in. Plexiglas and had internal dimensions of 244 cm (96 in.) height, 25.4 cm (10 in.) width and 12.7 cm (5 in.) depth. The working internals consisted of a packed section 12.7 X 12.7 cm (5 X 5 in.) cross section and 152 cm (60 in.) in height. The packed section was bounded by Plexiglas on two sides, screened on the other two sides and top and bottom. Figure 1 is a detailed drawing of the cross-flow cascade column. The screen which formed two sides of the packed section was made of 0.088-cm stainless steel wire woven into a mesh of 0.55 X 0.55 cm squares. The projected open area of the screen was 74%. Half-inch ceramic Raschig rings and polyethylene Pall rings of porosity 63% and 87'70, respectively, were used as packing material. Two sides of the box were removable to facilitate packing material and baffle spacing changes. A (20 in.) high section a t the top of the column remained void and served as a water droplet de-entrainer. A (15 in.) section at the bottom of the column remained void and served as a liquid accumulator to measure liquid holdup. Metal baffles 12.7 X 6.35 X 0.32 cm (5 X 21/2 X '/g in.) of carbon
Ind. Eng. Chem. Process Des. Dev., Vol. 19, No, 1, 1980 35 Armngenent No
scale:
w 12
6
4
4
Total No. of Batflea
12
6
4
3
NO.
of X-flow
:i2
u,,
Figure 3. Cross-flow contactor-equipmentarrangement schematic.
S-8
Figure 2. Baffle arrangements. Table I. Tower Physical Dimensions overall height: 2.44 m packed height: 1.52 m working volume (packing plus baffled space): 4 9 160 em' packed volume: 24 580 cm' cross-sectional area, horizontal: total = 323 cm' packed = 161 cm' cross-sectional area per stage, vertical: baffle arrangement 1= 161 em' baffle arrangement 2 = 323 cm2 baffle arrangement 3 = 484 cmr baffle arrangement 4 = 645 cm' column pressure taps 127 cm apart
steel were positioned in the open plenum on either side of the packed section. Baffles were gasketed with soft ruhher to prevent air and water bypass. Four baffle arrangement patterns were possihle as shown in Figure 2. Tahle I contains the physical dimensions of the cross-flow cascade. Fluid dynamic observations were made with air and water. Figure 3 is a schematic of the equipment arrangement. The air was supplied by a 51/2X 6 in. Roots-type blower, operating at 1750 rpm, rated a t 280 ft3/min at 3.5 psig. Air flow was measured with sharp edge orifices sized for the appropriate flow range and installed according to ASME power test code, part 5. Orifice pressure and pressure drop was measured with water and mercury manometers. Water flow was measured with a rotameter capable of 264 gpm full-scale reading. A centrifugal pump irrigated the packing with water recycled from the liquid accumulator. Liquid holdup was obtained from water level on two selected bafflers located in middle and top of tower. Column pressure drop was obtained with a Pace transducer indicator model CD25 equipped with a Model KP15 cell (diaphram) of range 0 f 2.5 PSID. Transducer was calibrated daily with 2.95 sp gr fluid and confirmed upon completion of daily test. Column outlet pressure was atmospheric and temperature was 20 to 30 "C. The purposes of the experiment were (a) to conduct a visual study of the phase (gas-liquid) contacting phenomena, (h) to determine the flooding (i.e., phase inversion)
Liquid Oul
Figure 4. Five, cross-flow staged packed tower point and/or other limiting characteristics (Le., drift angle), and (c) to measure gas phase pressure drop and liquid holdup as a function of flow, packing type, and baffle spacing.
Theory The permeable cross-sectional area available for the movement of the gas and liquid phases in conventionally constructed countercurrent packed towers is identical and inseparable. An alternative countercurrent, packed mass-transfer device can be created by forcing the gas to flow in a direction perpendicular to the liquid flow direction. Conceptually and mechanically a series of cross-flow units can he interconnected to create a cylindrical packed cascade tower as shown in Figure 4. Once the flow areas are uncoupled a packed tower with definite advantages as a gas-liquid contactor can result. Flooding and pressure drop conceivably should he affected. Flooding, caused in the conventional operation when the collinear forces of gravity and momentum of the
36 Ind. Eng. Chem. Process Des. Dev., Vol. 19, No. 1, 1980 100
,
1
1
8
,
DRY PACKINQ
0)
bl G -
G-0
Short OIIDYI lor l i q u i d A s :A, i 161m 2
sod
IO
Long O r r o w Z t o r
- I30
OBI
c l G - 100-1000
1
qome
10
lhrec
coles rnd
Figure 5. Gas-liquid contact patterns.
liquid (downward) are opposed by a force (upward) due to pressure gradient and form and frictional drag upon the liquid due to the moving gas phase (Hutton et al., 19741, may be forestalled because the forces upon the liquid in the cross-flow operation are not collinear and cannot have a zero resultant. Pressure drop which is increased by increased holdup at high gas rates, due in part to the same collinear forces, may be reduced. Further advantages may accrue once the gas cross-sectional area, A,, is independent of the liquid cross-sectional area, Ai. By choosing baffle spacings such that A, > Al pressure drop should be significantly reduced. Although open spaces (i.e., non-packed) exist in the baffled plenum, liquid bypassing and short circuiting should not occur; however, liquid drift within the packing should. Liquid drift is manifest by a horizontal liquid velocity component due to the moving gas phase. In general the crisscrossing gas phase tends to deflect the liquid toward the next lower baffle as it moves across a stage. However, due to the constant change in direction, the gas naturally redirects the liquid back into the packed space as it flows beneath each baffle and therefore prevents bypassing and short circuiting. Results Qualitative. The cross-flow cascade appears to be an effective gas-liquid contactor. No liquid bypassing or short circuiting occurs in the baffled plenum area. For the most part the liquid remains in the packed section. The fluids appear to flow perpendicular to one another while in the packing. The gas-liquid flow pattern is apparently crisscross, interweaving in nature. The liquid flow is apparent and the gas flow pattern can only be inferred from its effect upon the liquid. Detailed fluid behavior is shown by sketches in Figure 5. This sketch is typical of all four baffle spacings with either packing. At low gas rates (Figure 5a) the liquid is not deflected and moves straight down the tower, all of it remaining in the packing. At intermediate gas rates (Figure 5b) the liquid becomes definitely deflected into the gas flow direction. Standing liquid appears upon each baffle and a portion of the packing on the opposite side is unirrigated. Gas bubbles through the standing liquid from the screen side and this action appears not unlike that on a bubble tray. Gas liquid contact within the packing is vigorous and phase flows appear perpendicular. At still higher gas rates (Figure 5c) the liquid has a pronounced deflection. Horizontal liquid movement appears a t some locations. Standing liquid in the baffled plenum is for the most part absent; however, the entire plenum space is filled with a spray of entrained, circulating liquid droplets undergoing vigorous agitation. The packing appears irrigated throughout; however, entrainment recycling likely occurs opposite baffled points. Extremely vigorous gas-liquid contact is evident throughout the tower. Quantitative. Measurements of gas and liquid flow rates, pressure drop, and holdup were made in conjunction
500
10,cx
I,OOO
Gas R a t a , g / m Z s e c
Figure 6. One-half inch Raschig rings.
tor c o u n t e m r r m
300
1000
5
P
Goa R a t e , g/m2 roc
Figure 7. One-half inch Raschig rings.
with the visual observations. The effect of baffle spacing arrangement was the independent variable of primary interest. The baffle spacing variable is CY A,/A1 where A , is the packed cross-sectional area for gas flow created by baffle spacing and Ai is the packed cross-sectional area for liquid flow. This variable reflects the nominal area ratios available for gas and liquid flow in the packing. For the case of CY = 1,the areas are identical and the cross-flow tower is “equivalent” to countercurrent flow where CY = 1 is the only choice. Gas-phase pressure drop was measured between taps located in the packed section 127 cm apart. Gas and liquid loading rates upon the tower are based on the total (packing plus plenum) cross-sectional area of 323 cm2 without regard to baffle arrangement. Pressure drop data for selected experiments are indicated in Figures 6 to 9. The method of plotting is based upon the conventional log A p = f(1og G / $ , L ) relationship employed for packed beds.
Ind. Eng. Chem. Process Des. Dev., Vol. 19, No. 1, 1980
t 7 DRY
6
" a
37
L
PACKING
1
1 s
c
i
t 01
500
lop00
1,000
Go$ R a t e , p / &
sec
Figure 8. Five-eighth inch Pall rings.
LIQUID RATE
i
6110 p/m'
aec
IS"
c
~9~
1,dO'S Gas R a t e , q/m2sec
Figure 9. Five-eighth inch Pall rings,
Table I1 contains the essence of the experimental observations for both II2-in. ceramic Raschig rings and S/s-in. polyethylene Pall rings. This compact tabularized form was chosen to save space. Liquid holdup was obtained from inventory changes in the accumulator and checked periodically by direct weight. Total holdup is presented in Table 11. Liquid holdup a t standing water upon baffles was obtained from observations at two baffles, one located near the top and the other a t midcolumn. Baffle holdup is represented as a function of total holdup in Figure 10. Discussion Loading rates, both gas and liquid, measured in this study were higher than normally encountered in conventional packed tower operations. In the case of 1/2-in. Raschig rings, handbook data are available for L(max) = 11700 g/s m2 and G(max) = 810 g/s m2. In this study successful operations were attained for 6100 IL i 61 100 and 50 5 G 5 2700. It is tempting to compare the
cross-flow cascade with the conventional operation; however, this is impossible for the most part. The device employed in this study has twice the volume of a countercurrent device with identical packed volume. The tower volume that is void of packing undoubtedly contributes markedly to increased loading rates. No attempt was made to uncover an optimum packed volume to void volume ratio or to compare flooding limitations. It is possible to make a direct comparison on a single cross-flow stage. Pittaway (1976) reports that a single crossflow stage can be operated a t loading conditions which would cause flooding in the conventional operation. The pressure drop for the unirrigated operation shown in Figure 6 is proportional to G1.6-2,2which is typical of packed beds. The Pall ring unirrigated data show similar behavior and as expected, pressure drop is less than for the rings. Oza (1974) calculated the pressure drop in the plenum and found it to be insignificant when compared to that in the packing. The pressure drop for irrigated packing is reduced as a increases just as with the unirrigated case. At CY = 2 the pressure drop is significantly lower than CY = 1. The reduction between CY = 2 and 3 is less, and that between CY = 3 and 4 is less still. It appears that some limit is being approached whereby additional increase in CY does not result in a measurable I p reduction. This study suggests that an CY slightly greater than 4, possibly 5 or 6, may be the limiting value. This limiting CY appears the same for both packing types; however, the effect of the construction, particularly plenum volume, tower shape, screen open area, etc. on CY is unknown a t this point. Total liquid holdup, h,, accounts for the liquid holdup within the packing and liquid residing in the plenum as entrained droplets and standing pools upon the baffles. Baffle liquid holdup, h b , is a measure of the liquid holdup observed as distinguishable pools of liquid standing upon the baffles. It is a product of the average volume of liquid per baffle and the number of baffles. Baffle liquid holdup is reported as volume of water per volume of packing. Figure 10 shows the scatter diagram of h b vs. h,. In general h, is larger than for conventional countercurrent; however, approximately half of this is due to the baffles. The relatively high baffle holdup should provide fluid for side stream withdrawal. Pressure drop measurements appeared to be sensitive to h b and this resulted in dispersion of the data as typified by the a = 3 observations in Figure 7. For the case of column pressure drop a somewhat convincing argument can be constructed for comparing crossflow and countercurrent flow. For the arrangement CY = 1,the cross-flow environment is equivalent in the sense that the gas length-of-travel path in the packing is nominally identical with the countercurrent path. Ignoring for the moment the fact that loading rates are based on total cross-sectional area, the experimental data locus in Figure 7 suggests that conditions are equivalent. The cross-flow
38
Ind. Eng. Chem. Process Des. Dev., Vol. 19, No. 1, 1980
Table 11. Fluid Dvnamic Measurement@,b
G l a p l h , for a = 1
L I
0
7 330 1 2 200
18 3 0 0 3 0 500
01-111 7 010.52 I259012.3101-10.05 48014.41* 01-10.14 240/*10.15 280/4.8/* 120/2.5/* 190/5.6/0.28 90/8.8/*
54010.14 1111010.74/338013.81280/0.52/*
01-11380/0.23101-10.069 630/0.691* 93011.381* 138 012.6 5 10.24
860/0.058/1480/0.23/39010.3510.069 740/0.92/* 105011.67/* 1380/2.08/0.14 260/0.58/0.11 1 0 5 0 / 1 .S S / O . 21 160/0.63/0.25 63Ol2.76lO.34 80/0.69/0.21 4 5014 .08 10.35 16Oll.73 10.29 3 3 0 16.38 /=6 150 IO. 5 8 1 m o