INDUSTRIAL A N D ENGINEERING C H E M I S T R Y
December, 1944
6. The tubing characteristics of a compound may be predicted with a reasonable degree of accuracy from knowledge of the viscosity or the plasticity of the finished compound in conjunction with an estimate of ita relative surface smoothness. 6. Surface smoothness may be achieved b hot breakdown, ncreasing softener ratios, or increasing ratios o f carbon black. 7. High-temperature plastication tends to exert a deleterious effect on the physical properties of the subsequent yulcanizate, 8. Further investigation is re uired to estabhsh a relation between the relative benzene sdubility of the unplasticated polymer and ita processing characteristias. 9. Plastication at low temperatures tends to increase benzene solubility of the olymer, whereas plastication at hlgh temperatures tenas to re&ce it.
1119
LITERATURECITED (1) Baker, W. O., and Mullen, J. W., 11, unpub. rept. on "Solubility Relationships in G R S Polymer". (2) Flory, P. J., J. Am. Chem. Soc., 63, 3083 (1941). (a) Naugatuck Chemical Div., U. 8. Rubber Co., Synthetic Rubber Compounding Bull. 2, 4 (Oct., 1942). (4) Nellen, A. H., India Rubber World, 96, No. 0, 4 3 6 , 62 (Sept., 1937). (5) Tingey, E[. C., unpub. rept. (0) Vila, 0. R.,IND. ENG.CBEM.,34, 1209 (1940).
(7) Vile,0;. R., unpub. data.
Passsmsr, before the fall meeting of the Diviaion of Rubber Chemiatry. AMlaarcAN CEmmic& SOCIETY. in New York, N.Y.. 1943.
PURE HYDROCARBONS from
PETROLEUM Development of Laboratory Pilot-Plant Screen-Plate Fractionating Columns JOHN GRISWOLD, J. W. MORRIS, AND C. F. VAN BERG* University of Texas, Austin, Texas BEVELLED
WIRE
O V E R FLOW 8/10"
IN
SCREEN
SYALL LOWER
END
ASSEMBLY
" Y
LL 0
HIS article describes the development and performance of screen-plate columns that have been used for generalpurpose hydrocarbon fractionations and in pilot-plant processes for separating pure hydrocarbons from petroleum* i n the University of Texas laboratories during the past several years. Laboratory fractionating columns may be classified as packed, h - t y p e , and plate-type. With the exceptions of Stedman embossed packing (I) and single-turn helices (4), high-efficiency film-type packings have been successful only in smaller sizes of laboratory columns. The efficiency of such columns is often sensitive t o throughput and even to the operator's technique. These considerations led to the development of screen-tray columns having the desirable characteristic of relatively constant plate efficiency over a wide range of liquid and vapor velocities. Bruun developed a 1-inch (25-mm.) diameter, all-glass bubbletray column @) that has found extensive use. Oldershaw reported comparable performance data on all-glaw perforatedplate columns (6). Because of the complexities of constructional
T
PIPE
TUBINB.
NOTCHES
FOR
Development and performance of three designs of allmetal screen-plate fractionating columns for general laboratory and pilot-plant use are reported. The characteristics of six different screens were determined, and performance of 1.5-inch and 2-inch diameter columns are given with plate efficiency-rate data on n-heptanemethylcyclohexanea t total reflux. The features of these columns are ruggedness, ease of construction, and relatively high capacities, with maintenance of efficiency at high rates characteristic of plate-type columns. Maximum boil-up rates are 73 ml. per minute (1.2 gallons per hour) for a 1.5inch column and 250 ml. (4 gallons) for a 2-inch column. Maximum HETP occurs a t maximum boil-up, and the corresponding values are 1.8 and 3.2 inches, respectively.
I
I STEEL
PIP6
address, Grasselli Chemicals Department, Oak Ridge, Tenn. Present address,+Iumble Oil and Refining Company, Baytown, Teur. Previous articles of this series appeared in Volume 35, pages 117-19,
1 Present
Figure 1. Details of Construction of Experimental Column for Wire-Screen Plates (No.1)
247-61,8b4-7
(1948).
I N D U S T R I A L A N D E N G I N E E R I N G CHEMXSTRY
1120
Vol. 36, No. 12
before the t b t set of ~smpleswas taken. Succeeding -plea (st different vapor velooities) were withdrawn one hour after attainment of steady operating conditions for eaoh new inpur setting. This procedure was found to give reproducible values with no change in sample compositionsresulting from longer NW. Refractive indices of top and hottom samples were taken at 20" C. with B Bausch & Lomb dipping refractometer. Mole percentages of n-heptane were read from a plot of the dats given by Ward (a), and the number of theomtical plates "88 dculated by his prooedure, except that the more accurate value of 1.083 was used &s the relative volatility of n-hepte,n*methylcyel* hexane (6). The number of theoretical plates was usually reproducible to one plate for B given soreen st a fixed net heat input. The pressure drop was mmured by the mnnometer shown,
Figure 2.
Sereen Plate and its I'aarts (No. 1)
detuilx, sniall-scale all-nietal buhble-tray columna have not found gerieral usage. Palkin reported B column consisting of 40-mesh wire gauze cups mounted in 8 g1m shell (7). Buhhlecap plates, perforated plat=, and Screen pletes &refundamentally aimilsr sa t o mcelianism of vapor-liquid contacting. Screen plates are the easiest to fahricate in amaU siaos and should be most readily adaptahle to all-metal bench-scale column construction. DEVEWPMENT OF SCREEN-TRAY COLUMNS
Preliminary observations, using a wide variety of soreens mounted in a glass tube with sir-kcroscnc as vepor and liquid, showed that the screen should he rsther eohut should have a relatively small free area (or he claaely woven). Several designs of liquid overftow arrangements indicated that the mmt a a t b factory was &o the simplest-a plain tube, notched at the bottom and resting on the screen below. No difficulty with liquid running t h u g h the screen at the hottom of the overflow pipe w a encountered. The construction of an &metal test column is illustrated by Figures 1 and 2. The overflow pipes were merely p r e f i t t e d into the screens, projecting '/,e inch shove the upper screen surfaoes in order to msintain B shallow but definite liquid seal. The plates are punched from the screen, using plunger-end-ring dies made for the purpose. The thirty-six-plate test seotion wae amembled from the eingle-plate elements cut from l'jrinch standard pipe and machined as shown in Figure 1. These provide B plate spacing of 1 inch. Esoh section constitutes both oolumn wall and plate spacer, and leakage to the outside was prevented by sealing the joints with a molasses-graphite mixture. T h e test &up is shown in Figure 3. Tho still was eleotridly heated, with other circuits for separate compenaating hestera placed in the insulation of still and column. Heat loss determinations (ss wnttape) wore made sa blank runs over the temperature range existing in the tests. The losses were suhtraeted from the tot4 heat inputs ior the rate caleulationa. At any given pot temperature, heat lorn deviation from run to run averaged 0.35 wstt. The power going to veporimtion ranged from 24 to 250 watts.
The still charge was p m n-heptane and TESTPROCEDURE. purified methylcyclohexane. The odumn was flooded, then operated under total reflux at a steady boiling rate for 4 hours
INCHES
/I I
KI
I COLUMN
STlLL POT HEaT
CONTROL
Figure 3. Experimental Fraotionating Column for Wire-Soreen Plates
*
I
December, 1944
INDUSTRIAL AND ENGINEERING
CHEMISTRY
1121
TABLE I. WIREISCREENSPECIFICATIONS No. of
Screen Mesh No. per In. 1 30 2 22 3 14 6 6
l8 14 14
Wire Diam., In. 0.014 0.023 X 0.026 0.041 X 0.036 0.028 0.036 0.041
Opening, In. 0.0193 X 0.0193 0.0226 X 0.0206 0.0304 X 0.0364 0.0276 X 0.0276 0.0364 X 0.0364 0.0304 X 0.0304
Individual Opening Area, Sq. In. X 106 3.73 4.61 11.09 7.67 13.3 9.24
Free
Area, % 38.6 22.4 21.7 24.6 26.0 18.1
TABLE11. SUMMARY OF TESTDATAON WHEPTANE-METHYLCYCLOHEXANE AT TOTAL REFLUX(a 1.083) L..C1. LIQUID
Figure 4.
RATE AT TOTAL REFLUX, IN 1.61" I.D. COLUMN
ML./ MIN.
Over-all Plate Efficiencies for Usable Range of Rates
?$Jr
~~~ufnBv
Tz;-
Refraotive Index, n y Velocity, Hd'per DistilStill retical hciency, Cm./Sec. Plate late pot Plates % Column 1. 36-Plate Seotion, 1.61 In. I.D. (Screen No. 3) 60.0 6.9 1.73 0.16-0.16 1.39622 1.40966 21.6 11.0 2.74 0.16-0.18 1.39446 1.40928 22.6 62.8 17.6 4.39 0.18 1.39262 1.40838 26.6 70.9 0.21 1.39198 1.40877 6.19 24.8 28.6 79.2 0.21 1.39367 1.41148 8.41 27.8 26.7 77.0 1.39270 1.40840 0.22 33.8 8.42 26.8 71.6 42.4 0.24 1.39646 1.41093 10.67 23.6 66.6 1.59437 1.40899 0.26 61.6 12.81 22.2 61.6 62.8 16.40 0.26 1.39508 1.40917 21.0 68.4 0.27 1.39662 1.40932 16.34 86.6 20.4 66.6 1.39682 1.40961 0.28 17.77 20.1 66.9 71.3 Column 2,100-Plate Section, 2.07 In. I.D. (26 Plates Screen No. 3 , 7 6 Plates Screen No. 6) 1.64 0.23 1.39681 1.42207 66.6 56.6 10.1 0.28 1.39362 1.42172 2.42 68.6 16.9 68.6 0.28 1.39906 1.42263 3.30 67.2 67.2 21.7 66.6 55.5 4.40 0.29 1.39328 1.42132 28.9 36.1 6.49 1.39661 1.42184 64.2 64.2 66.0 9.89 0:31 1.39977 1.42172 46.4 46.4 108.3 16.60 0.32 1.40061 1.42149 42.8 42.8 0.46 1.40091 1.42120 40.1 40.1 137.1 20.83 Column 3, 16O-Plate8, 2.07 In. I.D. (Screen No. 6) 77.2 28 4.1 0.20 1.38869 1.41929 61.6 1.38822 1.41917 4.4 0.21 63.6 42.3 30 4.4 0.20 1.39026 1.42086 71.3 30 47.6 67.9 10.4 0.25 1.39020 1.42123 46.2 71 16.7 0.28 1.39102 1.42134 64.6 43.1 114 27.0 0.30 1.39172 1.42178 66.0 44.0 0.31 1.39413 1.42196 62.7 41.8 221 lS4 32.4 0.38 1.39099 1.42068 60.3 40.2 232 34.0 268 39.4 0.32 1.39274 1.42178 62.6 41.7 ~i~~~ Rate' Ml./Min.
~
~
~
~
aud the liquid holdup for each set of screens was determined by adding a small amount of nonvolatile mineral oil to the still and measuring its concentration before, during, and after test runs. The procedure was to weigh a 10-cc. sample from the still, evaporate the volatile hydrocarbons on a steam plate, and weigh again. Determinations before and after the run gave weights of oil within 0.1% of the value calculated from the amount added in each case. *Theholdup of all screens ranged from 3 to 6 ml. per plate, increasing with boiling rate. This compares favorably with Oldershaw's values when the column diameters are considered. It has been shown that the effect of holdup on s h a r p ness of separation is not great unless the amount of a key component held up in the column is a large fraction of the amount of the same component in the still liquid (3). CHARA~ERISTICB OF SCREENS. Complete efficiency-pressure drop-vapor rate curves were run on the six screens described in Table I, which had been selected on the basis of observations in the glass column noted earlier. Fourteen-mesh screen was the coarsest obtainable having a low-percentage free area, Since these six include all available coarse screens having low percentthe level of the overflow pipe. The extent of this interaction ages of free area, the optimum screen must be among them. should depend upon both bubble size and foam height. Foam The over-all plate efficienciesover the entire usable range of height was limited to one inch per plate, which waa reached rates are shown in Figure 4. For five of the screens, a definite a t a relatively low rate. At higher rates the amount of liquid peak in efficiency occurred at some intermediate rate, Of the carried into plates above by the foam reduced the vapor-liquid six screens tested, the size of opening appears to control the vapor separation and the observed plate efficiency. Pressure drop capacity. At rates above the peaks, the efficienciesof screens would be expected to increase with foam height. Confirming 2 and 4 fell off mfich more rapidly than did the others. these conjectures, screens No. 2 and 4 showed highest pressure Pressure drops ranging from 0.15 to 0.4 inch of water per plate were recorded. Highest pressure drops occurred on screens No. 2 and 4 a t rates immediately below the flood points. No breaks were found in the pressure drop-rate curves to indicate load points such as exist in packed columns. Definite surging or intermittent vapor flow was observed for each screen a t low rates, but in no case did surging affect the efficiency appreciably. The surging mechanism is explainable on the basis of surface tension. Study of the foregoing observations leads to postulations concerning the mechanism of the bubbling action, which were in part confirmed by visual observation in the preliminary Figure 5. Comparison of Plate Efficiencies glass apparatus. Since the efficiencies No. of Holdup Range, Citawere high for such a small liquid s a l , Column Plates Test Mixture Ml./Plate tion 6 n-Heptantr-toluene 0.7-1 (9) +4 Glsss bubble oa ('3) much if not most of the vapor-liquid Glaae perforatefplate 37 Carbon tetrachloride-benzene 3.0-6.9 0.9-4.6 Authors interaction must occur in foam above 0 Wire soreen ( 001. 1) 36 n-Heptane-methylcyclohexane
g
INDUSTRIAL A N D ENGINEERING CHEMISTRY
1122
design is shown by Figure 6. These changes approximately doubled the efficiency of the 2-inch plates. A 150-plate column was then made up in six sections of twentyfive plates each, using the design of Figure 6. It W M planned to use S/a-inch steel tubing for the overflow pipes; but since tubing waa not obtainable, the pipes were made by boring out standard '/,-inch pipe with a 7/l&ch drill. The supply of No. 3 screen on hand was inadequate t o make 150 plates, and no more was obtainable. The final column consisted of two sections (fifty plates) of No. 3 screen and four sections (one hundred plates) of No. 6 screen. The sections made with the No. 3 screen were installed a t the top and bottom of the column. Preliminary tests on the 150-plate column usually gave a nearly pure product a t one end of the column, which is not satisfactory from the analytical standpoint for calculating plate efficiencies. The procedure finally adopted was to superheat the top two sections and.introduce the reflux at the one hundredth plate from the bottom, thus testing 100 plates. Plate efficiencies of the 2-inch column were lower than those obtained with the 1.5-inch column, but the trend of plate efficiency with rate observed on the 2-inch column (Figure 9) corresponds in shape to a composite of the curves for screens 3 and 6,as evident from a study of Figure 4. The plate spacing of 1.25 inches on the larger column permitted a higher vapor velocity than was obtainable with the 1-inch spacing in column 1. (This 2-inch column is hereafter designated "column 2".) After about six months of intermittent use in a pilot plant using solvents, the maximum capacity of the column decreased aa a result of accumulation of sediment at the bottom of the overflow pipes. An occasional stoppage necessitated reconstruction of an entire 50-plate section, since each individual tray was fuse-welded and an integral part of the section. To avoid extensive loss of time from this source, a third design was evolved, Further observations in a g l w section with glass downflow pipes
OVERFLOW PIPE