INDUSTRIAL A N D ENGINEERING CHEMISTRY NOMENCLATURE
total surface of granular particles per unit volume, square feet per cubic foot total surface area of one packing unit, square feet particle diameter, feet arithmetic average diameter, feet diffusivity coefficient of transferable component in gaseow state, s uare feet/hour effective arealactor superficial maw velocity of flowing gas, pounds/(hour) (square foot) m w transfer factor heat transfer factor mass transfer coefficient for gas film, pound-mole/ (hour) square foot) (atmosphere) mean mo ecular weight of flowing gas, pounds/poundmole mean deviation number of experimental points partial pressure of transferable component in gas film, atmospheres mean artial pressure of nontransferable component in gas &m, atmospheres rate of ma= transfer, poundymoles/hour volume of granular bed, cubw feet
\
symbol for difference
c fi p
Vol. 43, No. 11
= packing porosity, void volume/total volume
+
= = =
absolute viscosit ounds/(foot) (hour) density, pounds$uEic foot shape factor
subscripts = film conditions m = mean
f
LITERATURE CITED
(1) Chilton, T. H., and Colburn, A. P., IND.ENG.CHEM.,26, 1183 (1934). (2) Gamson, B. W.,Chem.Eng. Progress, 47, 19 (1951). (8) Gamson, B. W., Thodos, G . , and Hougen, 0. A , , Trans. Am. Inst. Chem.Engrs.. 39,1 (1943). (4) Hobson, M., and Thodos, G., Chem. Eng. Progrees, 45, 517
.
119491. - .- - ,. (5) I b X , 47, 370 (1951).
(6)Hurt, D.M., IND. ENG.CHEM.,35,522 (1943). (7) Maisel, D. S.,and Sherwood, T. K., C h m . Eng. Pmgresa, 46, 1729 (1Q.m). ,-- (8) O'Brien, L. J., and Stutsman, L. F., IND. ENQ.CHEM., 42, 1181 (1950). (9) Powell, R.W.,Trans. Inst. Chent. Engra. (London), 18,36 (1940). (10) Resnick, W.,and White, R. R., C h m . Eng. Progress, 45, 377 (1949). (11) Taecker, R.G.,and Hougen, 0. A., Ibid., p. 188 (1949). RECEIVED December 18, 1950 - 8 .
Coking of Heavy Residual Oils cess
I
A LABORATORY-PILOT PLANT CORRELATIVE STUDY
I
LEO GARWIN' AND B. E. STEINKUHLER2 OKLAHOMA A. AND M. COLLEGE, STILLWATER, OKLA This study was undertaken to determine if the broad oven coking process could produce economically from residual oils, on a pilot plant scale, high grade coke and, as by-products, gas, gasoline, and a stock suitable for catalytic cracking, and whether a small bench test coking unit could be designed and operated to yield results correlatable with the pilot plant data. The 17 square foot Knowles oven used in the pilot plant studies produced high density coke consistently. The laboratory unit developed operated successfully, with very iow losses in coke, liquid, and gas products, at temperatures up to 1550' F. The data it yielded correlated well with the pilot plant results. I t should be possible to predict the coking characteristics of a charge stock in large-scale equipment from an experimental determination of its behavior in the laboratory coking unit.
I
NTEREST in the production of high grade electrode coke from heavy residual oils has grown recently because of present and anticipated market conditions. The method most widely used at present for the production of electrode coke is a two-step process. It consists of the production of raw petroleum coke by common refining coking methods, following by the calcination of the raw coke in rotating kilns a t approximately 2700' F. (8). 1 Prerert f
address, Kerr-McGee Oil Industries, Ino.,Oklahoma City, Okla. Present address, Monsanto Chemioal Co., Springfield, Mass.
A batch process for the production of high grade coke in one operation makes use of the Knowles coking oven. The operation of several such commercial units has been described (6-7, 9 ) . In one of these (ff), a correlation was noted between the carbon residue values of the charging stocks and the coke yields. It is the purpose of this paper to report comparative results obtained in a pilot plant-size Knowles oven that has 17 square feet of floor area and in a simple laboratory counterpart. The pilot plant study was conducted jointly by the Deep Rock Oil Corp. and Godfrey L. Cabot, Inc. The plant was located at the refinery of the Deep Rock Oil Corp. in Cushing, Okla.; it was built to ascertain if the broad oven coking process could produce economically from residual oils a maximum amount of high grade coke and, as by-products, gas, gasoline, and a stock suitable for catalytic cracking. The laboratory work was undertaken to determine whether a small bench unit, properly designed and operated, could yield results correlatable with the pilot plant results, and thua serve as a means for the rapid evaluation of the coking characteristics of any new charge stock. FEED STOCKS
PILOT PLANT.The properties of the stocks used for the pilot plant studies are given in Table I. Vis-breaker tar was the bottoms stream of a viscosity-breaking unit. Only one batch of this tar was used. Thermal tar waa the bottoms stream of a Dubbs thermal cracking unit. Different batches of this tar were used, hence the variation in analyses.
November 1951
INDUSTRIAL A N D ENGINEERING CHEMISTRY
LIOUID 8 L Y I l . t
Figure 1. Flow Diagram of Pilot Plant IPG.
Indieating remure .age
LLC. Liquid leve?oontrol MCV. Motor aontrol valve PC. Syitem premure aontrol valve
Virgin asphalt waa the bottoms stream of a vacuum fractionator. Several batchea of this tar were used, again accounting for the variation in analyses. Vi-breaker tar used here waa identical with LABORATORY. the pilot plant stock. It waa secured directly from the pilot plant storage tank. In thie o w , the thermal tar waa not identical with the ilot plant stock. It was obtained from the Dubbs thermal c r a c k g unit several week after owation of pilot plant operation. The Conradson carbon value waa 16.4%. Virgin asphalt waa also not identical with the pilot plant stock, being obtained after an interval of several weeks. The Conradson carbon value waa 19.8%.
condensable gases. The pressure on the system waa manually controlled by a valve in the aa line leading to the ejector. For the removal of coke %om the oven a hand coke pusher was used. The coke, aa it fell, waa caught in a coke cart, EO constructed as to fit under the lip of the oven. Chromel-alumel thermocouples were used for most of the temperature measurements. Higher temperatures were read with an optical pyrometer. OPERATION. A run consisted of a charging period, a degassing period, and a calcinin period. The oven was first Erought to a floor temperature of 1500' F. The oil charge was reheated to 550' F. by circulation throu h the preheater, the ctarging loop, and back to the char tan% The two oven doors were lowered into place and sea& with
PILOT PLANT
A flow diagram of the pilot plant is given in Figure 1. EQUIPMENT. *The oil charge waa stored in three 40-barrel tanks each eqwpped with a bayonet-type s t e a m heater. Movable feed linea o o n n e c e the storage tanks to the charge tank, which ww rovided m t h a steam coil. All tran&r lines were s t e a m traced and insulated. Steam a t 125 ounds pressure waa wed throughout the plant. 08 entered the oven from the oil loop through two 0.26-iich Xauck linear discharge valves. The pressure in the oil line was manually controlled b a valve in the recycle portion of the line. The o v a consistedrof four gaa-fired flu- conne0ted.t~an exhaust stack and a coking chamber, with a 17 8 uare foot floor area, directly above the flues leadin to the con8eming system. Each of the flues waa provided wit% a 530000-B.t.u. per hour natural gaa burner. Both the coking chamber and combustion flues were l i e d with silicon carbide brick, The two oil inlet porta were located diagonally opposite each other in the arch of the coking chamber. The chamber waa provided m t h two mortar-filled steel doors with cable winchea for lifting. The top of the coking chamber stack was rovided with a heavy, maohinefitted, alloy cap; this was openetand closed by a lever and chain system. The vapor main leaving the coking chamber waa construoted of heavy %inch steel ipe. It was lined with 1 inch of hi h temperature mortar whici was held in place by a &inch steefliner. The main condenser contained 16 rforated baffle platas spaced a t ap roxbnately 1 foot. A a m a l ~ a t e rcoil waa installed in the top opthia condenser. The after-condenser consisted of a short section of &inch pipe containin a 0.375-inch pipe water coil and a wider bottom portion, w h i 8 provided approximately 10 gallons of starage capacity. An entrainment separator consisting of a &foot length of 6-inch pipe containing small pebbles followed the after-condenser. Raflux li uid for the system was cooled in a finned tube exchanger. '&e oil passed throu h the Fnular section containing the fine and water was used as &e cooling me&um. Most of the reflux liquid waa pum ed to the top of the main condenser; a small amount was use$as quench in the va or msin. A %inch steam ejector waa used for t i e disposal of non-
f--
-VAPOR OUTLCT
c--.---*Figure 2. Details of Coking Unit
2588
INDUSTRIAL A N D ENGINEERING CHEMISTRY
Vol. 43, No. 11
input by control of the gas burners had little effect because of the high heat content of the oven refractory. LABORATORY COKING INSTALLATION EQUIPMENT. The laboratory counterpart of the pilot plant consisted of a coking chamber, a condenser, a liquid product receiver, a gas meter, a gas receiver, and a gas sampling system. Details of the coking unit are shown in Figure 2. Figure 3 is a flow diagram of the laboratory coking process.
1
COND. T O DRAIN
OXOAOTLIIR WI T O ORAIM
I
I VI*?
Holder
TO
''
ELIVATION
OVERFLOW SALT W A T L I TO R F C L I V E R
Figure 3. Flow Diagram of Laboratory Coking Installation
asbestos rope and fire clay. The charging period was started by opening the charging valves and regulating the feed line ressure. For about 5 to 10 minutes the coking vapors were al6wed to p""s out the open oven stack. When carbon black disappeared rom the smoke issuin from the stack, the stack damper was shut, and circulation ofreflux-Le., a portion of the liquid yield from the preceding run-was started. The pressure on the system was regulated manually t o get a small plume of smoke from a 0.25-inch hole in the cap of one of the oven peepsights. Oil was charged a t the desired rate until the charging was complete. At the close of the charge period, the oven inlet valves were closed and the preheater shut off. Degassing then took place for 0.5 t o 1.5 hours. When the gas production was ractically complete, the condensin system was shut down, a n l t h e oven stack damper opened. T%e run was terminated when the coke was calcined a t the desired temperature. The coke was pulled from the oven with a rake. It was waterquenched, dried, weiFhed, and analyzed. The collected liquid was gaged in the receiver. Spot and composite liquid samples were taken for analysis. ASTM distillation data for these samples are reported as: initial boilin point (I.B.P.) 400" F. (gasoline), 400" F. t o cracking (gas oil) a n i residue (recycle stock). Hempel distillations were also made on several selected samples. Spot gas samples were taken in drums by water dispIacement. The temperature of the exhaust gas was about 90' to 100' F. The gas was noticeably wet. Carbon black formation in previous Knowles coking units had been a serious problem. This was eliminated here by operating the oven a t an initial floor temperature of 1500' F.and by maintaining a slight positive pressure on the oven to prevent air Ieaka e into the oven. Carbon black formation was serious for initiakfloor temperatures above 1500' F.; a t 1800' F. the system was rendered inoperable. The most serious operational difficulty encountered in the pilot plant was that of oven door leaks. These accounted for some of the relatively high losses experienced. Other factors affecting the losses were floor leaks, vapor purging a t the start of a run, and vapor losses during calcination. Some trouble was initially experienced with tar frothing into the vapor line but proper charging rate remedied this. Essentially, the only effective controls on the coking process were the initial floor temperature and the charge rate. Variation in heat
The coking unit proper consisted of a covered iron Skidmore crucible of 180-ml. capacity set in a mild steel, open top cylindrical heating chamber. A No. 3 wide-form Coors porcelain crucible containin the charge was set in the Skidmore crucible. A yoke, fitted wit% a thumbscrew, was su plied with the Skidmore crucible for keeping the cover of tgis crucible in place. This was not satisfactory, however, in the prevention of leakage a t high temperatures. The use of an asbestos gasket either plain or im regnated with water glass did not remedy the situation. A tigft metal-to-metal seal, obtlined by grinding the cover and crucible together with a mixture of glycerol and emery powder, sto ped leaks for short times but warping of the crucible cawed leafage t o recur. It was found that warping and leakage could be reduced to a minimum with a tight metal-to-metal seal, if maintained by bpring loading of the crucible cover from the yoke, no pressure being applied t o the thumbscrew. The screw wa8 turned in until it barely made contact with the crucible cover; it was used primarily to keep the s ring in place. The spring was 0.5 inch in diameter and 1 inci long. A new spring was used for each run. An additional advantage of a spring loaded crucible cover over a rigidly clamped one is the safety factor. Accidental plugging
t 6 1
8
+\+
j'
VlROlN ASPHALT
e.4
TNCRMAL TAR
r t c o v c n la41
14
I
I
1000
I
I
I
I
1
1400
1300
I
IS00
Maximum Temperature,
I 1600
F.
Figure 4. Laboratory Coke Yield us. Maximum Coking Temperature I8
I
I
I
I
I
I
VIS-BREAKER TAR
TWERYAL 7AR
4
IC00
1100
1400
IS00
4 1600
Maximum Temperature, ' F.
Figure 5. Laboratory Gas Yield vs. Maximum Coking Tehperature
INDUSTRIAL A N D ENGINEERING CHEMISTRY
November 1951
2589
of the vapor line due to frothing over of the char e can be dangerous, since the heat content of the chamber wib cause coking to continue for a few minutes after the source of heat is removed. A fire hazard would be created by the sudden release of large quantities of vapors from the crucible, but this could be handled easily with the proper equipment. The vapor line consisted of a %inch length of 0.25-inch 22-gage brass tubing brazed t o a 2.5-inch length of copper tubing of the mme dimensions. This line was connected to the condenser tube bv a flare fitting. The heating chamber consisted of a short section of 3-inch standard pipe coupling with a plate welded to the bottom t o form a cylindrical crucible. The bottom of the chamber was covered with about a 0.375-inch bed of fine sand. The chamber was insulated with a 0.5-inch layer of asbestos cement. A 20-gage chromel-alumel thermocouple was inserted so that ita bead w-as just above the sand layer and about 0.125inch from the coking crucible. The thermocouple wan connected to a millivoltmeter and the pyrometer calibrated as a unit. The entire coking unit was surrounded by a thin ga e, sheet steel cylindrical section 6 5 inches in diameter and 9 inties hie;h. The bottom end was open while the top end was provided m t h a 3.5-inch hole for the ventin of burner gases. Appropriate slots were cut in the side of t f e cylinder to accommodate the coking unit. With this ohield there was a material reduction in heat losses. The heatin? chamber was supported by two ring stands and a refractory rmg of the type specified in the Conradson carbon residue test (2). Heat was uupplied by three natural gas burners, each having a heat output of approximately 7500 B.t.u. per hour.
04.
I
I
Figure 8. Typical Laboratory and Pilot Plant ASTM Distillation Curves of Liquid Produced f r o m Vis-breaker Tar
1 SO
I
I
I6
IC00
I
IS00
I
I
,.do .
I
I
moo
I
I
no0
Maximum Temperature, F. Figure 6. Laboratory Liquid Yield we. Maximum Coking Temperature
1
s
// /
I . S S C
1.40
1.40
1.3s
IC00
moo
Figure 7.
1000
1400
Msdmum Temperature,
le06
F.
Laboratory Coke Density us. M a x i m u m Coking Temperature '
The condenser was of the conventional ASTM distillation ty e ( 1 ) . It was equipped with a steam heating coil of copper tu&, as well as with circulating cooling water. The liquid product receiver was a Berzelius beaker with a 100-ml. capacit , This was followed by a wet-test meter, which contained distiged water saturated with gas from several preliminary runs. There were two Florence flask gas holders, each of 0-liter capacity. The liquid displaced by the gas was a saturated a ueous solution of salt. Reguyation of pressure on the system waa accomplished manually by the vertical positioning of a siphon-overflow flask. Flexibility of the siphon was provided by a short section of rubber tubing. The pressure a t the meter was maintained in this way between atmospheric and 0.2 inch of water vacuum. The gas was sampled for specific gravity determination by passing through a silica gel drying tube into a previously evacuated gas balloon of borosilicate glass. OPERATION.A standard type of run waa developed from the results of some preliminary runs. It consisted of a preheating period, a coking period, and a blasting period. A 50-gram sample of the oil stock was weighed into the porcelain crucible, the latter placed in the Skidmore crucible, the crucible cover fitted in place and spring loaded. The va or line was connected to the condenser tube, the burners were [ghted and the coking crucible was preheated. Water was circulated through the condenser box. The appearance of the first drop of li uid condensate in the receiver signified the beginning of the coping period. The gas burners were regulated to control the coking rate as indicated below. m e n the li uid make was ractically complete, the temperature was raise! to the desirea maximum value and maintained there until the gas make was complete. The burners were shut off the condenser water was shut off, and steam turned on into the heating coil. The water in the condenser box was allowed to boil for about 10 minutes to ensure the complete drainage of liquid from the condenser tube. The time-temperature sequence followed for the vis-breaker and thermal tar samples was as follows: The samples were preheated to lOOO0 F. for 0.5 hour, coked
000-
700
I
I
I
I
800
I
-
-
RUN 41
700
-
600
-
I
1
I
I
I
I
'
-
/ O ?
/ : +
PILOT PLANT
-
-
'L
LABORATORY R U M 3 1 8 3l~ll8eOoF.)
-
LMORATORY RUN f 2 0 ?.l,(lll0%1
-
4
100
I00
-
-
I
I
I
I
I
I
I
I
IO
ro
ao
40
60
eo
70
80
,,loo*
eo
100
Figure 9. Typical Laboratory and Pilot Plant ASTM Distillation Curves of Liquid Produced from Thermal Tar
a t 1000" to 1200' F. for 0.5 hour, and then blasted to maximum tem erature of 1200', 1350', or 1520' to 1550' F. for 1 hour. T i e procedure was modified somewhat in the case of the virgin asphalt samples because of the tendency for the material to froth. The sample was coked directly in the Skidmore crucible under the followin conditions: The sampfe was preheated to almost 900' F. for 20 minutes, coked a t 880' to 900" F. for 0.5 hour, coked a t 900' to 1200' F. for 0.5 hour, and then blasted to maximum temperature of 1200",1350°, or 1520" to 1550' F. for 1 hour. One virgin asphalt run was made with a smaller sample of 20 grams coked in the porcelain crucible instead of the Skidmore crucible. No significant difference in product distribution was noted. Two runs were made for each set of conditions and the coke and liquid products pooled for analysis. The highest temperature attained in the laboratory coking unit was 1550' F. Higher temperatures would have been desirable during the blasting operation, as the production of high density coke requires temperatures above 1550" F. However, at 1500' F. the crucible scaled badly; there was also a tendency toward warping of the crucible rim. Although leakage was small, it probably could have been reduced still further by the use of a vapor ]in$ of greater crosseectional area. During the initial phase of the coking operation, the volume of vapor produced is large, and the back pressure developed was responsible for most of the leakage. There was no observable tendency for the coking vapors to polymerize or to form coke in the vapor line or condenser tube. The condenser-heating coil combination proved satisfactory. All liquid drained completely and the gas leaving the collector appeared to be very dry. The use of a larger coking crucible, capable of handling larger samples and yielding larger amounts of products for analysis, would have been desirable. The rest of the laborat,ory coking
I IO
I
eo
I
I
ao
40
60
I 60
1
ro
I
eo
I
eo
100
Figure 10. Typical L:horatory and Pilot Plant ASTM Distillation Curves of Liquid Produced from Virgin Asphalt
installation possessed suf5cient capacity for a t least a 100-gram charge. ANALYTICAL.The following tests were made on the laboratory samples: Feed stocks, Conradson carbon residue ( 2 ) . Coke, real density. The coke was crushed to pass a 100-mesh screen and the real density determined on a 2-gram sample by kerosene immersion under vacuum. Coke, volatile combustible matter (VCM) (8). Liquid, API gravity ( 4 ) . Liquid, ASTM distillation ( 1 ), A modification was introduced in that the sample was 50 nil. instead of 100 ml. The distillation data were divided into 3 cuts; I.B.P. to 400" F., 400' F. to cracking, and residue. Gas, specific gravity. This was determined by direct weighing in a gas sampling balloon of known capacity. RESULTS
PILOTPLANT.Forty-eight pilot plant runs yielding operutional data, product distribution, and quality information Kere made. Table I gives the best process data available from the pilot plant while it was operating near maximum capacity to produce maximum liquid recovery and high density coke from the three charge stocks. LABORATORY. Eighteen standard laboratory runp were made, six on each of the three feed stocks. Operational data, product distribution, and quality results are presented in Table 11. The product distribution results are also reported in this table on a loss-free basis. The loss was always less than 2%, and in the majority of cases less than 1%. The losses were considered to be two thirds gas and one third liquid, and apportioned accordingly. These figures were arrived a t by an inspection of the product distribution data for a particular set of run conditionn as affected by the magnitude of the total loss.
INDUSTRIAL AND E N G I N E E R I N G CHEMISTRY
November 1951
TABLE I. SUMMARY OF AVERAGE PILOT
8.4 18.5-21.9 1.13 0 075-0.08 900
Operating conditione Temperatures F. Oven floor, initial Feed preheat Coke Vapor average Total c h k m pounds Total charge, gallons Material balance Gas weight % on ahawe Li Aid weight % on charge C a e , kei h t % on aharge LOSS,weig%t % on charge Product inspections Li uid bravity, 0 A.P.I. Volume at400° F Hem el Volpme at 400° F:, A.S.%.M. dlstillatfon Volume 4cLstillation 9 " F.,.to orsoking, AsT Volime 's, &due. A.S.T.M. distillation
"
,
Xi
GiM Average moleoular weight Coke VCM, weight %
RESULTS
Viabreaker Tar
Virgin Asphalt
Feed Stock Gravit A.P.I. Conragon carbon, weight % Sulfur weight % Ash ;eight o Visdosity 8.i7.8. at 2100 F. Visaosity: 8.F.S. at 250' F. Characterization factor Cycle Charge period, hour8 Degas period, hours Calcining period, houm
PLANT
l!:t
Thermal T ~ ,
13*i6_;t:g
0.96 15o!i 10.95
o .88
o.:!&tg
11.05
4.0 1.2 2.0
4.0 0.6
g.8
1500 560 875830 1910 227
1600 650 850870 1910 227
1500
19 63 21 7
8 72 14
Ti
6
4
26.4 I9 11 82
15 7 5 3 86
16,s
7
11
26.0
24.0
24 0
0.95
0.35
1.15
1.6
4.0
660
8o08t{ 18QO
224 g
3 87
259 1
t o be approximately equal t o the Conradson carbon residue value of the stock. The laboratory coking data show slight decreases in the coke and liquid yields with increasing temperature, a corresponding increase in the gas yield, and a marked increase in real density with increasing temperature. The increase in real density of the coke is accompanied, as expected, by a corresponding decrease in coke volatile combustible matter. The increase in gas yield takes place mostly a t the expense of coke, particularly in the case of virgin asphalt. This is because the liquid make is practically complete a t 1200' F., and temperatures higher than this result in gas production from the remaining coke. If this were completely true, however, there would be no effect of temperatures above 1200" F. on theliquid yield. Variations in the rate of coking between 900" to 1000" F. are believed responsible for the small changes in liquid yield observed. There appears to be no noticeable effect of maximum coking temperature on the quality of the liquid produced. There was also very little change in the gas density. The laboratory coke yields at 1520" to 1550' F.compared well with the Conradson carbon residue values except for the virgin asphalt. This may be explained by the fact that the Conradson carbon test was difficult to run on the virgin asphalt. This test specifies a standard rate a t which the temperature must be raised. The resultant frothing produced a low Conradson carbon value. LABORATORY-PILOT PLANT CORRELATION. The Comparative laboratory-pilot plant results in Table 111 show that the agreement is satisfactory, particularly when the following points are noted: A higher calcining temperature was employed in the
TABLE 111. COMPARISON OF LABORATORY AND PILOTPLANT DATA
Figures 4,5, 0, and 7 show, respectively, the coke yield, gas yield, liquid yield, and coke density as a function of the maximum coking temperature. A comparison of the quality PILOTPLANT AND LABORALTORY. of the liquid produced in the pilot plant and in the laboratory for each of the three charge stocks is given in Figures 8,9, and 10. A summary of comparative yield and coke property data is given in Table 111.
Coke vi~~~:~$rF Lab., 15450 F: Lab"1350OF p i 1 2 lant Thermaf tar Lab 1200" F Lab:: 1350'F: Lab 1535O F. Pilzplant Virgin asphalt Lab 12000 F
DISCUSSION AND CONCLUSIONS
High grade electrode coke was made consistently in the pilot plant oven. The coke yield for a given feed stock was found
Yields, Weight % Liquid Gas
,15.g 15, 15.4 14
78 78 .2 77 0 72
5,4 6 7 ., 6 8
17 0 16.8 16 4 14
77.6 76.4 76.2 73
5 4 6.8 7.4 9
26 6
61.8
11.6
Less ,,
.. .. 6
.. .. .. 4 ..
3211 ii$$ $ i::: 85t.g :$: Pilot plant
21
53
19
Coke Properties VCM Real we@' density, grams/ml
7.77 3.69 6.34 0.35
1.44 1.53 1. 72
7.53
1.41 1.61 1.71
5152 1.15
..
::
5:t(l
7
0.95
..
1.40 1.50 1,64
TABLE 11. SUMMARY OF LABORATORY RUNDATA Run Ne.
12
17 18 14 16 Feed stock Vis-breaker Tar Temperature6 ' F. 990 1020 1040 1010 990 Inititrl cokidg 980 Maximum cokinp 1540 1660 1350 1350 1200 1200 76 83 79 82 84 79 Condensate reaeiver Product distribution 15.5 15.4 15.8 16.0 15.2 15.0 Coke, weight % of charge 76.6 77.2 77.6 77.5 78.3 78.4 Liquid,weight%ofchsrge 5.8 5.6 4.5 5.3 6 . 4 7.6 Gas weight $7 of charge 1.4 0.3 1 . 1 1.6 1.8 0 . 2 LOB;, weight of charge Product distribution on lom-free basis 15.2 15.0 15.5 15.4 15.8 16.0 Coke, weight % of charge Liquid, weight % of oharge 77.2 77.3 78.0 78.0 78.8 78.5 7.7 6.5 6.6 5.4 5.5 Gas, weight % of charge 7.6 Produot inspeotions Coke density, gyarns per mi. 1.72 1.63 1.44 Coke VCM, weight % 3.69 6.34 7.77 Li uid 9;'ravit 0 A.P.I. 16.5 16.2 16 1 A.8.T.k distillation Volume I B P tc 4ooo F. 7 6 6 Volume$:4bObF. toarack81 ing 81 82 Volume %, reaidue 12 12 13 Gas Specific gravity 0.72 0.82 0.82 0.80 0.74 0.85 Average moleaular weight 20.8 22.8 23.8 23.3 21.4 24.6
%
19
22
23
26 27 Thermal Tar
24
25
32
33
1010 1020 1020 1020 1010 1020 905 880 1535 1536 1350 1350 1200 1200 1525 1520 78 76 77 79 74 78 79 76 16.4 76.0 7.3 0.3
37 38 35 Virgin Asphalt 920 1350 81
920 1350 84
16.4 16.8 16.7 16.9 17.1 23.8 23.5 24.9 25.1 76.2 7 6 . 3 76.4 77.3 77.4 58.5 58.7 60.4 60.2 4.9 5.0 16.0 16.2 14.4 13.9 7.2 6 . 8 6.7 1.6 0.3 0.8 0.2 0 . 9 0.5 1.7 0 . 1 0.2
16.4 16.4 76.1 76.3 7.5 7.3
30
900 1200 78
920 1200 76
26.6 26.7 61.4 61.6 1 1 . 0 11.2 1.0 0 . 5
16.8 16.7 16.9 17.1 23.8 23.8 24.9 25.1 26.6 76.3 76.5 77.6 77.6 59.1 59.2 60.5 6 0 . 5 61.7 5.3 17.1 17.3 14.6 14.4 11.7 6.8 0.5 6.9
26.7 61.8 11.5
1.71 5.62
1.51
..
1 41 7.53
1.64 5.81
1.50
..
1 40
17.0
16.9
33 1
0.70 20.4
16.7
33.0
32.9
7
7
7
23
22
22
78 15
77
77 16
71
70 8
71 7
0.74 21.6
16
0.84 24.2
0.81 0.74 0 . 7 8 0.82 23.5 21.4 22.7 23.8
6
0.83 23.9
0.84 0.84 0.83 24.4 24.4 24.2
0.84 24.4
I N D U S T R I A L A N D ENGINEERING CHEMISTRY
2592
pilot plant than in the laboratory. This would tend to produce a higher gas yield and a lower coke yield. The laboratory data are presented on a loss-free basis; the pilot plant results involve losses of 4 to 7%. The laboratory gas was condenscd a t a lower temperature than that prevailing in the pilot plant after-condenBer. This is reflected in a slightly lower specific gravity for the laboratory gas as compared with the pilot plant gas. The results indicate, therefore, that the coking characteristics of a particular stock in the pilot plant can be reliably evaluated by its performance in the laboratory coking unit. Figures 8, 9, and 10 show that the quality of the liquid produced in the laboratory compares well except for the virgin asphalt with that of the comparable pilot plant stock. In the case of the virgin asphalt, thc laboratory unit produced a much lower boiling and lower specific gravity liquid. hlore efficient condensation of the liquid product in the laboratory may account for some of this, but it is possible that a good part of the discrepancy is due to differences in the charge stock.
Enginnyring
Vol. 43, No. 11
ACKNOWLEDGMENT
The authors are greatly indebted to N. K. Anderson of the Deep Rock Oil Corp., Cushing, Okla., for making possible participation in the pilot plant operation, and for making available the feed stock samples, the pilot plant data, and certain analytical information. LITERATURE CITED
(1) "A.S.T.M. Standards," Part 111-A,
Test 086-46,p. 155,Philadelphia, American Society for Testing Materials, 1946.
(2)Zbid., Test D189-46, p. 120. (3)Zbid., T e s t D271-46, p. 31. (4)Ibid., Test D287-39, p. 191. (5) Curran, M. D..Oil Gas J.,48,No. 15, 100 (1949). (6) Foster, A. L..Nutl. Petroleum News, 25,26 (1933). (7)Petroleum Refiner, 16, 63 (1937). (8) Watkins, J. C.,Chem. & Met. Eng., 44, 153 (1937). (9) Ziegenhain,W.T . , Oil Gas J.,30,16 (1931). RECEIVED December 12, 1950. Presented before the Sixth Southwest ReSOCIETY, San Antonio, Tex.. gional Meeting of the AMERICAN CHEMICAL December 1950.
Effect of Column Holdup in Batch Distillation
pocess development
I AND C. J. GARRAHAN* E. I. DU PONT DE NEMOURS & CO., INC., WILMINGTON, DEL.
R. L. PIGFORD', 1. B. TEPE,
T h e effect of column holdup on the sharpness of separation in the batch distillation of binary mixtures at constant reflux ratio was investigated through calculations involving a differential analyzer. Instantaneous values of distillate and residue compositions and of fraction of original kettle charge distilled over were calculated for values of column holdup from 14 to 56Yo of the initial charge, reflux ratios from 3.2 to 10 ( L I D ) , and relative volatilities from 1.15 to 3. A general tendency for the size of the intermediate frac-
tion to increase with increasing holdup was observed. Io the case of certain easy separations characterized by high relative volatility, however, the size of the intermediate cut decreased at first and then increased with increasing holdup. The optimum value of holdup was found to correspond roughly to values which are encountered in commercial batch-distillation operations. In the presence of appreciableholdup, the effect of reflux ratio on size of intermediate cut was found to be less pronounced than would be expected with negligible holdup.
M
In the design of a batch distillation unit the quantity of the intermediate fraction must be determined. This fraction must be collected and redistilled with the next charge in order to obtain semicontinuous production of two nearly pure product fractions. Important variables which determine the size of the intermediate cut include the relative volatility, the reflux ratio, the number of theoretical plates, and the liquid holdup of the column. I t has been customary to neglect the effect of column holdup because of the mathematical complexity which is introduced into the design calculations by taking this variable into consideration. There are a number of different ways of approaching the problem of determining quantitatively the effect of column holdup on the sharpness of separation in batch distillation. The most direct method is by experiment. Alternatively, the physical system can be analyzed mathematically. A system of differential equations which expresses the relationships between the important variables can be derived without difficulty. These equations, however, cannot be reduced to useful algebraic form by analytical methods without making certain undesirable assumptions. They can be solved for specific casea by tedioua numerical methods or by graphical methods, but these methods
ETHODS for the deeign of batch distillation equipment are less rigorous than those for the similar operation of continuous distillation, largely because variations of plate compositions with time lead to complex mathematical relationships which cannot be handled conveniently by the techniques of formal mathematics. The extensive development of large-scale computing machinery, which started in recent times with the so-called continuous integraph of Bush et al. (g) and which was accelerated during the war, has made available today a number of large scale computing machines capable of solving mathematical problems heretofore considered too difficult or too time-consuming for solution by conventional methods. In the present work, a typical large-scale computing machine, the differential analyzer located a t the University of Pennsylvania, was used to investigate the sharpness of separation in the batch distillation of binary mixtures. I t is hoped that this work will contribute to a better understanding of the batch distillation operation and to the refinement of batch distillation design methods. 1 Present address, Department of Chemical Engineering, University of Delaware, Newark, Del. * Preeent address, Department of Eleotrical Engineering, Swarthmore College, Swarthmore, Pa
