Moving Coke-Bed Gas Filter for Dust Removal G. C. EGLESON', H. P. SIMONSZ, L. J. KANE, AND A. E. SANDS3 Synfhesis Gas Branch,
U. S. Bureau of Mines, Morganfown, W. Va.
R
ESEARCH and development on processes for the production of synthesis gas from pulverized coal are a part of the synthetic fuels program of the United States Bureau of Mines. The Synthesis Gas Branch has been operating gasification pilot plants under a cooperative agreement with West Virginia University. At an early stage of this development work, it became apparent that removal of dust kntrained in the gas stream would be a difficult problem. The method of filtration of dust by a coke-filled shaft filter, developed by Sachsse of Oppau, Germany ( I ) , for the recovery of carbon black obtained in the partial combustion of acetylene with oxygen, appeared t o be applicable to the problem. Experimental work was thus begun on a gianular coke-bed filter, in order t o develop an efficient continuous process for the removal of dust entrained in synthesis gas. Specifications ( 2 ) for dust content have been set a t 0.05 grain per 100 cubic feet of gas before turbocompressors or synthesis catalpt and 0.25 grain per 100 cubic feet before dry box purifiers. The initial dust content of the gas will, of course, vary, depending on the specific method of gasification chosen. Gasification operations a t the Bureau of Mines pilot plant in Morgantown have shown dust contents varying from 900 grains per 100 cubic feet of gas a t the outlet of the generator to 40 grains per 100 cubic fec,t following a conventional dust removal train consisting of a baffle chamber and a water scrubbing system. Experimental Apparatus Provides Intermittent
Flow of Coke through Filter Column Preliminary experimental work was carried out with dust-laden air passed through a small, fixed-bed filter 3 inches in diameter and 4 feet long, filled to a depth of 3 feet with 0.1- to 0.3-inch coke. With this filter, dust removal efficiencies of 99.7% were obtained for short periods. At an air rate of 4000 cubic feet per hour per square foot of filtering area, the pressure loss across the clean bed was found to be less than 2.5 inches of water. Since these preliminary results were encouraging, a small pilot plant scale filter of larger capacity and greater depth, with a moving bed 'capable of continuous filtration, was designed and constructed. In such a large capacity filter the edge effect was considerably reduced. Figures 1 and 2 show the design of the experimental movingbed filter. The shaft and hoist bin were constructed of 10-gage steel sheet. The two grate plates were made from 7-gage high carbon steel. Standard 2-inch iron pipe was used for piping and 3 and 4 are photographs of the comDleted water seals. Figures - unit, including the filter and auxiliary equipment. Incoming dust-laden air was passed upward through the filterinn- medium of granular coke. When the Dressure drou across the
-
* Present
address. Dow Chemical Go., Freeport, Tex. 2 Present address, Department of Chemical Engineering, West Virginia University, Morgantown, W. Va. a Present address, Union Carbide and Carbon Corp., Oak Ridge, Tenn.
June 1954
WATER
Figure 1. Front View of Moving-Bed Filter
filter increased above an established value because of the clogging of the lower section of the bed, a portion of the coke was dropped through the grate, by reciprocating m n ~ n e n t 5imparted by hand to the movable plate, into the wash hopper below the flange. Here the dust was washed from the coke with water in swirling motion provided by a tangential water inlet line. The wash water was drained Offthrough the lower water seal and the
INDUSTRIAL AND ENGINEERING CHEMISTRY
1157
ENGINEERING, DESIGN, AND PROCESS DEVELOPMENT 2"
TOP I2 " D I A . A
SIDE
FRONT
LJGRATE
GRATE SECTION
2" WATER OVERFLOW d ( T H R 0 U G H 48" SEAL) A---
Figure 2.
Details of Shaft of Moving-Bed Filter
cleaned coke was dropped into the water hoist bin. From the bin the coke was elevated to the top of the filter shaft with water. The hoisting water passed through the separating screcn and was drained off through the upper water seal. The coke slid down the screen and dropped on the top of the filter bed. Preliminary Experimental Runs. The results of 12 experimental dust removal runs are given in Table I and shown graphically in Figure 5 . Air was used as the dust entraining and carrying medium in all cases. Dust was added to the air by means of a vibrating hopper feeder. A 15-inch-diameter cylindrical chamber was placed ahead of the filter, so that large particles capable of being removed by elutriation would drop out. The dust used in all tests, except run 5 , was a residue obtained from one of the pilot plant gasification runs. Since the residue showed a definite tendency t o agglomerate, the dust particles were undoubtedly considerably larger than those originally produced by the gasifier. Tl'hile the ultimate particle size of the dust, as usually determined, would lead to erroneous conclusions regarding the effect of particle size on the operation of the filter, knowledge of the size of the agglomerates as they actually exist in the gas stieam ~ o u l d be useful. However, no satisfactory method for making this determination could be devised. The apparent density of the residue was extremely low and varied from 0.035 to 0.050 gram per cc. The coke was 0.125 to 0.40 inch in diameter; about one third was in the 0.375- to 0.40-inch range. The particles were regular in shape and all diameters of each particle mere approximately equal. The average pressure drop during the experimental runs represented in Figure 5 was 4 inches of water. Run 12 was made at a constant pressure loss of 5 inches of
Figure 3.
Lower Section of Moving-Bed Filter
Blower and dust feeder in foreground
1158
INDUSTRIAL AND ENGINEERING CHEMISTRY
Vol. 46, No. 6
PILOT PLANTS water; a small amount of coke was fed continuously. A comparieon of this run with runs made under the same experimental conditions in other respects shows increased efficiency. The capacity of a filter of this type may be limited by several factors in any particular application; but, in any case, the capacity is limited to the maximum gas flow that will permit downward flow of coke through the grate. Experimental work with a similar filter in Germany showed this flow rate to be about 3500 cubic feet of gas per (hour)(square foot of filtering area). At pressures considerably above atmospheric, the gas flow capacity would, of course, increase appreciably.
Curve A of Figure 8 shows the result of applying these formulas to the filter used, while curve B shows the result calculated as above, but modified by use of the particle diameter correction factor-Le., average particle diameter equals 0.82 D,. The experimental points are also shown on the diagram. A sample calculation is given below, using as an example the lowest point employed in graphing curve A-637 cubic feet per (hour)(square foot of cross section area of filter) on the abscissa versus 0.0394 inch of water pressure drop on the ordinate. Gas rate = 500 cu. ft./hr. Gas rate/sa. ft. cross-section area of emotv filter = _ 500_ - 637 cu. ft./(hr.)(sq. ft.) 0.785 I
Calculations of Pressure Drop through Bed Are Based on Operating Data
This value is used on the abscissa. To calculate the ordinate, it is first necessary to calculate the value of MRe.
From the operating results of runs 8, 9, 10, and 11, a r e l a t i o n s h i p between the amount of dust in the filter bed and the pressure drop was obtained. Figure 6 shows t h i s relationship-Le., increase in pressure drop above the resistance of clean filter bed as a function of the dust content of the bed. The equation for the curve is
(G
+ 0.591)2 = 0.517(P + 0.675)
.-
.
D,
=
0.30 inch = 0.025 ft.
'
=
500 cu. ft./hr. 0.785 sq. ft. X 3600 sec./hr. = 0'177 ft'/sec'
p
= 0.0808 lb./cu. ft. X ___
.
29.50 inches Hg 29.92 inches Hg 0.0748 lb./cu. ft.
492" R. 524" R.
(1)
The equation in this form is valid only for the experimental filter with a flow of 1275 cubic feet per hour per square foot. On the basis of 1 square foot of filter crosssection area, the equation becomes
(G
+ 0.752)2 = 0.838(P + 0.675)
(2)
Figure 4. Upper Section of Moving-Bed Filter
Sampling device in foreground The curves of Figure 7 show p r e s s u r e d r o p d a t a recorded a t various time intervals for runs 8 and 9. The dips of the curves represent the removal of fouled coke from the filter bed. With filter beds of fresh clean coke used for the first time, considerably more dust was removed per unit volume of coke. This was due to the fact tha: the recirculated coke was not completely free from residue after washing. Therefore, only the results of runs made with coke re-used several times, washed, and recirculated were employed for calculating the experimental points shown in Figure 6. The equation of the curve is empirical and should only be used for pressure drops ranging from 0.5 to 6 inches of water, The pressure drop through the clean bed may be calculated by means of the modified Fanning equation with a particle diameter correction factor of 0.82. The equation used was
dF
4f'f"V,z
m=2gD,
0.0110
'
400 1
I
'
6W
1
'
BOO
1
1000 1
1
DUST IN INLET AIR GRAINS/IOOSTD. cu. 'FT.
Figure 5. Variation of Dust Concentration in Filtered Air with Original Dust load Air Rate, Std. Cu. Ft./(Hr.) (Sq. Ft.)
+0 0
1275 1000 650
(3) p
The values of f' and f" were obtained from curves given by Walker, Lewis, McAdams, and Gilliland ( 4 ) by the use of the modified Reynolds number, hlRe, which is calculated from the formula (4)
June 1954
" 200
=
0.018 centipoises X 2.42 lb./(hr.)(cu. ft.) centipoise
=
1.21 X 10-6 lb./(sec.)(ft.)
1 3600 sec./hr.
Therefore, MRe =
0.025 ft. X 0.177 ft./sec. X 0.0748 lb./cu. ft. = 27,3 1.21 x 10-6 lb./(sec.)(ft.)
INDUSTRIAL AND ENGINEERING CHEMISTRY
1159
ENGINEERING, DESIGN, AND PROCESS DEVELOPMENT
Table
1.
Experiments with Dust-Laden Air
(Depth of filter bed 8.5 it.; temp. approx. 75' F.)
Air Flom R a t e Cu. ft./ hr. 530 1060 ion0 ion0 1000
Run NO.
g.j!:i. ft.)
,
Dust Concn. a t Inlet Conditions, Grains/100 Ft. Inlet Outlet
Dust ~~~~~~l Efficiency,
ing separation to remove the deposited dust, probably would facilitate hot dedusting and recirculation. Further, dry hoisting would simplify residue recovery should the dust thus removed be of value.
%
0.088 0,176 0,042 0.112
0,101 0,027 0.125 0.001 0.043 0.110
1200
500 750 1000 750 1000 ion0
0.109
0.120
E 5
E
TIME, MINUTES
8
Figure
m 5
7.
0 I
-
Time Variation of Pressure Drop across Filter
g4 z
2
z
;3 0 m Y
2 2 w Y 4
-E
l
$00
Figure
400
600 800 1000 DUST IN BED. QRAINS
12.00
6. Effect of Dust in Filter Bed on Pressure Drop
Air rate 1 2 7 5 std. cu. ft./(hr.)(sq. ft.) Filter diam. 1 2 inches
B using the above value in the curves in Walker, Lewis,',bIcAdams, and Gilliland ( 4 )
I"
= 37 f " = 0.95
Therefore,
The desirability of heat recovery or the simultaneous use of catalytic reaction has also been mentioned. With this in mind, several other materials have been investigated to ascertain their efficiency as filtering media. I t v a s found, in general, that rough, porous materials were much more efficient, even in larger size ranges, than smooth, regular-shaped filter media. For instance, a 4- to 8-mesh broken silicon carbide was found t o be as efficient ay coke, and a 4- to 8-mesh bauxite showed some promise as a filtering medium. On the other hand, 0.125-inch-diameter mullite spheres readily allowed the escape of dust, as did larger alumina balls. X slightly modified design proposed for use in further study of the filter under actual gasification conditions is discussed in a previous Bureau of Mines publication ( 3 ) . Coke Hoisting. A detailed study of the required coke circulation rate has not yet been made. During run 12, made a t a constant pressure loss of 5 inches of water, coke was recirculated a t the rate of 0.5 cubic foot per hour. On the basis of this coke rate and other data given in Table I, about 1.0 pound of dust was removed per cubic foot of coke and 2000 cubic feet of gas was treated per cubic foot of coke, under the conditions of this particular test.
-
4 X 37 ft.-lb./lb. X 0.95 X (0.177 ft./sec.)* dF 2 X 32.17 ft./sec.l X 0.025 ft. dT = 2.74 ft.-lb./lbo =
2.74 ft.-lb./lb. X 0.0748 lb./cu. ft. X
0.192 inches HzO lb,,sq. ft.
= 0.0394 inches HzO
Estimates of pressure drop through fouled beds a t flow rates other than 1275 cubic feet per (hour)(square foot) can be made by using Equation 1 and multiplying the result by the ratio of the pressure drop a t the given gas flow rate to that a t 1275 cubic feet per (hour)(square foot) for a clean bed. This ratio can also be obtained graphically from Figure 8. !
400
Investigation of Operating Variables Results in Design Modifications
Several modifications in filter design and operation have been suggested t o make the equipment usable under various conditions. One possibility is removal of dust from hot gas by a dry filter bed. The use of a conventional mechanical hoisting apparatus to return the hot coke to the top of the bed, after a screen1160
800
I
1200
GAS RATE,STD. CU. FT./(HR.)(SQ.
Figure 8.
I6m
2mc
FT.)
Effect of Gas Flow Rates on Pressure Drop across Filter Bed
A. 8.
Calcd. by Fanning equation Calcd. by Fonning equation with porticle diameter correction factor 0 . Experimental
INDUSTRIAL AND ENGINEERING CHEMISTRY
Vol. 46, No. 6
PILOT PLANTS , , , ,SEPARATING
Figures 9 and 10 show that, with the equipment used, the optimum coke to hoisting water ratio was about 0.20 pound of dry coke per pound of water. The curve of Figure 9 reveals that, as the fraction of undershot water used (water passing a t a right angle to the outlet pipe from the coke bin) was reduced, the hoisting efficiency increased and tended toward a maximum a t a relatively low ratio of undershot water to total water used. However, when less than about one third of the total water used was undershot, jamming of coke a t the base of the hoist bin and in the tee below it caused difficulties. The curve of Figure 10 shows the effect of using different velocities in the hoisting pipe while holding the undershot water to one third of the total. The optimum velocity occurred a t approximately 30 inches of water flow per second. It appears that, in a larger pipeline, a smaller fraction of undershot water may be used successfully.
FUNNEL
(J/DEFLECTING
T
Synthesis Gas Is Cleaned by Moving-Bed Filter in Coal Gasification Pilot Plant
SCREEN
GAS I N
Modification of the filter apparatus was based on experience gained from the experimental work. h motor-driven rotary coke extractor mas substituted for the grate plates, a larger hoist bin was installed, and the inlet point for the gas stream was placed above the feeder. The changes are shown in Figures 11 and 12. The filter unit was installed as part of the dust removal train in a small scale pilot plant gasifying powdered coal at a rate of about 50 pounds per hour and producing approximately 1300 to 1400 cubic feet of synthesis gas per hour. The gas, after passing through the filter, was stored in a single-lift holder and subsequently compressed to 300 pounds per square inch for sulfur removal studies in the purification pilot plant of the Bureau of Mines.
11 ,-ROTARY
EXTRACTOR
WASHING CHAMBER WASH WATER OUT
WASH WATER IN PLUG V A L V E WATER DRAIN TANGENTIAL L I F T WATER I N L E T HOIST B I N S E A L POT PLUG VALVE
L I F T WATER I N L E T
Figure WATER, FRACTION UNDERSHOT
Figure
9.
WATER VELOCITY IN HOISTING PIPE, INCHES/SEC.
Coke Hoisting Efficiency
One third of water admitted to tee a t base of bin Two thirds of water admitted to bin
The coke filter apparatus, thus modified for the removal of dust from synthesis gas, remained a semicontinuously operated unit. The length of the operating cycle (continuous use of a given coke bed) ranged from 15 minutes to 6 hours, depending on the dust content of the gas and the maximum allowable pressure drop. The magnitude of the latter, in turn, is determined by economics of the operation, such as power requirement for gas blower, etc. In any case, the coke extraction, washing, and recycling must last long enough to decrease the pressure drop to the desired level. During this period, usually a few minutes, the gas flow through the coke bed is uninterrupted, which makes the filtration, in the sense of the gas flow, continuous. June 1954
for Pilot Plant Operation
Coke Hoisting Efficiency
W a t e r velocity 28.5 inches/sec.
Figure 10.
1 1. Moving Coke-Bed Filter Modified
Substitution of a rotary coke extractor for the grate plate may not be practical, or even feasible, for large commercial scale operabe more advantageous and preferable for operation on any scale, but, unfortunately, the relatively small size grate with mechanical reciprocating device needed for the work described could not be obtained. The use of a rotary coke extractor appeared to be the only way to substitute a mechanical extractor for the manually shaken grate. Table I1 summarizes the performance of the modified filter unit in several gas making runs. In the first three runs, several difficulties were encountered, such as bridges formed across the shaft, which prevented the free movement of coke, Also, the gas distribution was poor in spite of the deflecting funnel above the gas inlet point. As a result, an excessive amount of coke had t o be circulated. Most of these difficulties have, however, been eliminated by perforation of the vanes of the rotary extractor and placement of the gas inlet below the extmctor. In run 82, a high pressure drop vms allowed to build up across the bed by permitting the dust to accumulate in the lower portion of the bed. This improved the dust removal efficiency but made it impossible to recirculate coke during operation. However, when the gas flow was diverted from the filter, the coke moved and discharged without difficulty. Following run 82, a water spray was installed above the bed, just below the gas outlet point, and during subsequent runs 4 gallons per minute of water was continuously sprayed on the top
INDUSTRIAL AND ENGINEERING CHEMISTRY
1161
ENGINEERING, DESIGN, AND PROCESS DEVELOPMENT The efficiency of the filter is improved by using a water spray at the top of the bed. Use of a mechanically operated reciprocating grate appear8 to be more advantageous and feasible than a rotary extractor for removing coke from the filter bed for reactivation. Operation of the coke-bed filter is continuous in the sense that the gas flow through the filter bed need not be interrupted for the periodic extraction, washing, and recirculation of the filter medium. The necessity for cyclic reactivation of the coke makes the process of filtration semicontinuous.
Table II.
Run
KO. 4d 52% 80 81 82 83 84 85
Filter Operation with Pilot Plant Gasifier
Dust Concn., Graind100 c u . Ft. Inlet OutleF 798 1.4 1.4 108
...
47,l 57.6 130 48.0 44.0
...
1.1 0.57 0.59 1.3 1.8
Coke Particle Diam., Inch 0 . 1 -0.4 0.1 -0.4 0.09-0.25 0.09-0.26 0.00-0.25 0,09-0.25 0.09-0.25 0.09-0.25
Pressure Drop across Bed, Inchee HzO 3.5 3.5 3.5 3.5 7.5 3.5 3.5 3.5
Duration Dust of Run, Removed, Hr. % 20.2 99.8 20.4 98.7 , . . 10.9 97.4 14.8 4.4 99.0 99.5 3.8 4.7 07.0 3.5 96.4
The results of the experiments described were promising enough t o indicate that further work should be done to determine the effects of all variables under an established and controlled set of conditions. This investigation should be carried out on a somewhat larger scale t o obtain the necessary design data for constructing commercial size coke-bed filter units for practical industrial applications. Acknowledgment
Figure 12. Lower Section of Moving Coke-Bed Filter Removing Entrained Dust from Synthesis Gas
of the bed. Comparison of run 83 with runs 81 and 82 sliows thc increased efficiency. After run 83 was completed, a new generator head was installed in the gasifier and the coal used for gasification was changed from 63 to 91% minus 200 mesh. Because of this, and since the percentage of carbon converted to gas during these runs was considerably higher than for earlier runs, the dust entering the filter was undoubtedly finer than in previous runs. The lower dust removal efficiency during runs 84 and 85 is attributed t o these factors.
Nomenclature
G
P
= =
F N V,
= = =
g
= = = = = =
D,
’, f” R4Re
Summary and Conclusions
It was not feasible to operate the gasifier a t a constant set of conditions because of the nature of the gasification and sulfur purification programs. Therefore, the type, quality, and amount of dust entering the coke filter were subject t o considerable change. Nevertheless, the following general conclusions can be drawn from the results of the tests discussed in this paper: Thc reduced efficiency of the filter, compared to earlier work with recovered residue entrained in air, is attributed to the smaller particle size of dust entrained in synthesis gas. The filter operates with negligible coke attrition losses. It is essential for proper distribution t o introduce the gas below the coke extracting mechanism unless a radical improvement is made in the design of the gas distributor. The efficiency of the filter is improved by operating a t higher pressure differentials. However, the maximum pressure drop is limited by the amount of dust that may be allomTed to accumulate in the lower portion of the bed without interfering with rocirculation of the coke. 1162
The authors are indebted t o John J. S. Sebastian, U. S. Bureau of Mines, Rlorgantonx, W. Va., for the thorough revision of the manuscript,, including several additions and alterations in the text, as d l as for valuable suggestions,
p
p
dust in the bed, 1000 grains increase in bed resistance above resistance of clean bed, inches of water friction loss, ft.-lb. forcelib. mass (feet of fluid flowing) depth of bed, ft. superficial velocity based on total cross-section area, ft. /see. acceleration due to gravity, 32.2 ft./sec.* diameter of packing, ft. factors in modified Fanning equation, dimensionlesmodified Reynolds number fluid density, lb../cu. ft. absolute viscosity of the fluid, lb./(sec.)(ft.)
literature Cited (1) Rachsse, E., Technical Oil Mission, Reel 132, Ammonia Lahoratory, Oppau, Germany, 1941 (available from Library of Congrem, Washington 25, D. C.) (2) Sands, 9. E,, Wainwright, H. W., and Schmidt, L,D., ZND. EKG. CHEM., 40, 603 (1048).
(3) Strimbeck, G. R , Holden, J. H., Rockenbach, L.P., Cordiner J. B.. J r , and Schmidt, L. D., U. 8. Bur. Mines, Rept. laeest. 4733 (1950). (4) Walker, W.H., Lewis, X’, K., McAdams, W. H., and Gilliland, E. R., “Principles of Chemical Engineering,” 3rd ed., pp, 95-6, ITew York, McGrsw-Hill Book Go., 1937. RECBIVED for reiiew November 9, 1953.
INDUSTRIAL AND ENGINEERING CHEMISTRY
ACCEPTED
February 8, 1952.
Yol. 46,No. 6