Efficiency of Packed Fractionating Columns - Effect of Vacuum Operation

Excess Molar Volume along with Viscosity, Flash Point, and Refractive Index for Binary Mixtures of cis-Decalin or trans-Decalin with C9 to C11 n-Alkan...
2 downloads 0 Views 910KB Size
Efficiency of Packed

Fractionating Columns EFFECT OF VACUUM OPERATION ROBERT T. STRUCIC' AND CORLISS R. ICINNEY The Pennsylvania State College, State College, Pa.

%

McMahon saddles, and Cannon pertruded packing with Fenske helices. The most suitable system for an investigation a t low pressures, based on the available vapor-liquid equilibrium data ( 1 , 3-5, 10, I d , 18, 23, 96-38, 31, 8%), appears t o be the ndecane-trans-Decalin (decahydmnaphthalene) system developed by Fenske, Myers, and Quiggle (16). This system has a relative volatility which is constant with composition, but which decreases with decreasing pressure. Also, these hydrocarbons can be obtained readily and in a highly pure state by fractionation of the source materials. Since Myers (83) did not report results a t pressures below 50 mm. of mercury, the vapor-liquid equilibria for this system were determined a t an absolute pressure of 20 and 10 mm. of mercury as part of this work. This system was used for efficiency tests a t reduced pressures and the system n-heptane-methylcyclohexane at atmospheric pressure.

T h e effect of diminished pressure on the efficiency of packed fractionating columns has been investigated in a 0.75-inch glass column packed with Raschig rings, Fenslce stainless steel helices, Cannon pertruded packing, and McMahon wire gauze saddles. Column efficiency was found to vary little with pressure, being slightly higher at 50 and 100 mm. of mercury absolute pressure than at 10, 20, or 730 mm. of mercury. In addition to efficiency tests on the four packings, vapor-liquid equilibrium data and correlations of holdup and pressure drop are reported.

D

ISTILLATION under reduced pressureshas found increasing use both commercially and as an analytical tool. Aside from the inducement offered by possible increased relative Volatilities under vacuum operation, the necessity of avoiding thermal effects has given the growth of vacuum rectification great impetus. For example, coal tars and particularly water-gas tars are known to contain valuable heat-sensitive compounds which will polymerize to resins if raised to their boiling points a t atmospheric pressure. The degree of thermal decomposition Or Polymerization produced depends upon the. temperatures involved and the time of exposure. During fractionation the temperature level is controlled by the boiling points of the compounds a t the pressure used. The time required for a given separation depends upon the relative volatilities, the reflux ratio, the Separating ability (efficiency) of the column employed, and the rate of distillation. Very little knowledge is available on the effect Of reduced pressures on several of these factors. At ]OW Pressures the temperatures are lower, but rates of distillation are also much k!SS; this requires a longer distillation time for any given reflux ratio. The effect Of reduced pressure on relative Volatility may be either favorable or unfavorable and must be determined for each case. The other two factors are interrelated, the reflux ratio required being a function of the efficiency of the column. It is this last question, that of the effect Of VaCUUm Operation On the efficiency Of packed fractionating Columns, which has been studied and reported here. It has been generally considered that the efficiency of a frattionating column would be lower a t lower pressures due t o the lower ratio of liquid to Vapor v 0 h - m flowing, because this would tend t o Permit the rising vapor less contact with descending liquid. The validity of this assumption has not been checked until recently owing to the lack of adequate vrtpor-liquid equilibrium data a t reduced pressures. Recently, Fenske, Myers, and Quiggle (16)and Berg and PoPovac (3)have studied this problem and have reported the general conclusion that the effect of pressure on column efficiency is slight. These workers used a Podbielniak column and columns packed with Eknske helices. The work reported here was undertaken before the above results (3, 15) were published and also includes data comparing the operation of columns packed with Raschig rings, 1

APPARATUS

A vapor-liquid equilibrium still of the Othmer type ( 2 4 ) was constructed using enlarged vapor tubes to ensure negligible pressure drop at' low pressures. Bumping was prevented by the use of a U-tube boiling leg filled with stainless steel helices. A more complete description of the apparatus used has been given elsewhere (300). The selection of the size of the column to be used was governed by the desire to use column which would be typical of those available in most laboratories, have low holdup, ]ow pressure drop, and fairly high efficiency. In order to have all these p r o p erties simultaneously it was necessary to use packing materials which would normally be considered too large compared to the diameter of the column for best efficiency. The resulting efficiencies, however, were very good, and show what can be expected of & typical 0.75-inch diameter column if packed with large (up to 0.25-inch diameter), low pressure-&op packings. The column was of the usual glass laboratory type having 0.75-inch diameter, 32-inch high packed section surrounded by air jackets formed by a concentric tube wound with resistance wire to provide adiabatic operation, and another outside glass tube to decrease heat loss and the effect of air currents. Thermocouples were fastened to the outside surfaces of the inner tube and the first jacketing tube. After calibration, these allowed determination of the proper heat input to ensure adiabaticitye The condensing head was of the usual type with a cold finger condenser displaced from the axis of the column and was fitted to the column with a standard taper 24/40 joint. Connection to the vacyum system was through a 20-mm. tube and a 35/20 ball and socket joint. The distillate receiver was connected to the take-off head by ground joints and was of the usual vacuum type permitting removal of a sample during fractionation by evacuation of a receiving bottle. The rate of distillation was measured a t the base of the column by means of a special rate-measuring device which fitted between the still and the column. This unit consisted

Present address, Pittsburgh Consolidation Coal Company, Library, Pa.

77

INDUSTRIAL AND ENGINEERING CHEMISTRY

78

TABLE I. PROPERTIES OF PACKING MATERIALS USED Packing type Soniinal size, inch Dimensions, inch Length (or 0.d.) Width (or i.d.) Height General form Material

Raschig rings

Mchiahon saddles

'/a

1/4

Single-turn helices 5/82 i.d. Made from No. 26 Brown and Sharpe gage wire

Cannon 0.18 X 0.16

Made from a 0.254 0.24 piece 0.25 X 0.116 0.15 0.16 X 0.003 0.232 0.09 Rascliig ring Saddle Helix Cupped strip Carbon 100 X 100 Stainless steel Pertruded mesh niokel stainless screen 609 4840 No. of pieces usedn 24,540 9447 0.22318 0.023218 0.017435 0.013934 Wt. of 1 piece, g. 0,000107 Volume of 1 piece, cu, inch 0.00R28 0.000179 0,000120 0.3097 0.1615 0,0270 Area of 1 piece, sq. inch 0.0825 188.5 781.8 662.5 Total packing area, sq. inch 779.4 263,O 856,3 737.0 Area of packing 853.9 column wall- sq. inch b 56.8 48.1 56.6 PaokiLiarea, sq. inch/cu. inche 13.7 19.1 62.2 53.5 62.0 Total area. sa. inch/cu. inclic 59.0 93.7 91.8 80.9 % free space5190 808 5320 3890 Packing area X % free space 1127 5830 5690 4330 Total area X % free space 0.736 0.811 0.403 0.213 Static holdup, c c . / o c . d a The column was continuously pounded during packing. b Area of column walls, 74.5 square inches. 0 Square inches of surface per cubic inch of packed volume. d Cubic centimeters of liquid per cubic centimeter of packed volume, after 15 minutes draining (n-heptane a t 90' '2.).

+

of three separate vertical tubes nhich came to common openings a t the top and bottom (having the appearance of 4 ) . The top and bottom were fitted with standard taper joints. The central tube contained a stopcock surmounted by a vacuum-jacketed bulb calibrated to measure the liquid return from the drip tip of the column. The two outside tubes provided passages for the vapor from the still to reach the column. These arms were lagged with a 0.75-inch thickness of asbestos to prevent excessive heat loss. To take a rate measurement, the normally open stopcock was closed and the time required to fill the calibrated bulb determined. The still was a standard 1-liter flask fitted with a thermometer well, a sampling device, and a connection to the pressure drop manometer. The flask was heated by a Glas-Col mantle connected through an ammeter to a variable transformer. The sampling receiver for the still was similar to that for the distillate except that the sample was removed through a dipping tube followed by a cooler. A lower vacuum was maintained in the receiver by an auxiliary vacuum source in order to raise the liquid into the receiver. The pressure in the still was transmitted to the pressuredrop manometer through a 2-mm. capillary tube followed by a condenser. Purge nitrogen was introduced into this pressure line through a sight-feed bubbler in order to prevent diffusion of vapors into the pressure connection. The pressure drop could be read on either a U-tube manometer or an inclined draft gage, both filled with n-dibutyl phthalate. The head pressure on the column was maintained by a ballasted vacuum system operated by an inclined-tube mercury regulator. The regulator actuated a normally closed solenoid valve in the line to a continuously operating pump. The solenoid had variable maximum and minimum valves which permitted control of the pressure to within 1 0 . 5 mm. The pressure was read on a U-tube absolute mercury manometer or a McLeod gage. Four different packings were tested: 0.25-inch carbon Raschig rings; 0.25-inch stainless steel McMahon gauze saddles; 0.16inch Cannon pertruded packing; and 6/32-inch inside diameter stainless steel, single-turn helices (Fenske packing). The properties of these packings are given in Table I. Boiling points of the pure materials and known mixtures were determined in apparatus of the modified Cottrell type similar to that described by Fenske ( I s ) ,except that there was no takeoff line and they were built to operate on a much smaller charge (8 ml.). A detailed description of these units have been given by Hinkel ( 2 0 ) . Boiling points were determined a t 760 mm. by ballasting with nitrogen and a t reduced pressures by using the balIa+d vacuum system. Precision instruments allowed determina-

Vol. 42, No. 1

tion of the temperature to *0.05" C. Refractive indexes were determined on a 4-place Abbe-type refractometer maintained a t 20.0" 0.1" C. Densities were determined with pycnometers of the modified Sprengel type (20). MATERIALS

The purification and properties of the materials used were as follows. The absolute ethyl alcohol gave an index of refraction, np, of 1.3615 and a density, die, of 0.7899. The n-heptane was taken from heart cuts of a fractionation of Phillips pure grade n-heptane through a 100-plate column at a reflux ratio of 100 to 1. The index of refraction was nSo, 1.3877; the density, die, 0.6839; and the boiling point, 98.43' C. Methylcyclohexane, a special product of the Dow Chemical Company, was percolated through silica gel to remove aromatics and unsaturates and was fractionated in the same way as the n-heptane. Its properties were index of refraction, n p , of 1.4231; density die, of 0.7692; and boiling point of 100.94' C. The n-decane was percolated through silica gel and fractionated through the same column at 20-mm. The index of refraction, n y , was 1.4119 and its boiling point was 174.10' C. Pure trans-Decalin was obtained by fractionation a t 20-mm. of a sample of Decalin. The fraction used had an index of refraction of 1.4692 and a boiling point of 187.30' C. Polyamylnaphthalene, sold under the trade name of Pentalene 92 by Sharples Chemicals, Inc., had an initial boiling point of 353" C. The refractive index v a s 1.5407. PROCEDURE

The vapoi-liquid equilibrium data were determined in the usual manner, allowing the still to run at steady conditions until the vapor sample had been replaced 30 times. A winding on the side walls was used to prevent condensation in the still chamber. The amount of heat put in through the side winding was sufficient to produce a superheat of 1 " to 2" C. in the vapors. Care mas taken to boil the liquids at such a rate as to avoid splashing on the heated walls. The accuracy of the still and procedure were checked by determination of the vapor-liquid equilibria for ethyl alcohol-water at 760 mm. The results checked the curves of Carey and Lewis ( 7 ) and Langdon and Keyes ( 9 1 ) with an average deviation of approximately 0.2 mole %. Tests were also made on ethyl alcohol-water at 95 mm. and on n-d~cane-trans-Decalina t 50, 20, and 10 mni. In order to ensure even, dense packing, the column was continually pounded with a piece of rubber tubing while the packing materials were slowly dropped into it. This resulted in slightly denser packings than have been reported in the literature for these materials. The rate of distillation in the rectifying column was measured a t the bottom of the column by the rate-measuring device inserted between the still and the column. The procedure was simply t o close the stopcock and measure the time required t o fill the calibrated bulb with liquid. The pressure drop (or back pressure) was determined as a function of throughput at each pressure for each of the mixtures used to determine efficiency. The rate-measuring device was never present during efficiency tests, the rate being found from previously determined curves of throughput versus pressure drop. The total amount of liquid present in the column during operation is of great importance in limiting the amount of intermediate fractions produced in separating two components. The total holdup was measured at atmospheric pressure by a mixture of polyamylnaphthalene in n-heptane. At reduced pressures, a mixture of polyamylnaphthalene in n-decane was used. Poly-

INDUSTRIAL AND ENGINEERING CHEMISTRY

January 1950

TABLE 11. RESULTSOF TESTSON EQUILIBRIUM STILL WITH ETHYLALCOHOI~WATER AT 95 MM. OF MERCURY PRESSURE Liquid Composition,

a

Vapor Composition, va XQ Experimental Literature (I) 14.5 49.8 49.5 15.3 50.8 50.4 21.0 55.4 54.9 29.6 59.4 58.8 36.1 62.3 60.8 45.2 65.3 63.9 52.3 68.1 66.9 55.1 69.3 68.2 55.7 69 4 68.4 69.3 75.9 75.8 Mole per cent alcohol.

Deviation from Literature Value +0.3 +0.4 +0.5 +0.6 +1.5 4-1.4 f1.2 +1.1 f1.0 +o. 1

amylnaphthalene has a refractive index high enough (1.5407) to permit analysis of mixtures with paraffins by refractive index. To determine the holdup, a mixture of approximately 40% polyamylnaphthalene in n-heptane or n-decane was placed in the still, the column flooded, and allowed to come to steady conditions. A sample was then taken, weighed, and its analysis determined by refractive index. Assuming no polyamylnaphthalene was distilled into the column and, knowing the amount of polyamylnaphthalene present originally, the total amount of liquid in the still during operation was calculated and subtracted from the original total weight to give the amount of holdup in the column. All determinations of efficiency were made a t total reflux. Between 400 and 500 ml. of the test mixture were placed in the flask, the column was flooded several times, and allowed to reflux a t steady conditions until equilibrium was assured. After purging the lines, reflux, and still, samples were taken. Analysis of these by refractive index permitted determination of the separation obtained as measured by the number of theoretical plates or transfer units required. Since the relative volatility for the system used here is constant with composition, the number of theoretical plates could be calculated by Fenske's equation (16) and the number of transfer units by the equation of Chilton and Colburn (8). The time allowed before taking samples a t atmospheric pressure was from 2 to 5 times that required for equilibrium as pre-

79

dieted by the equation of Berg and James (2). Equilibrium was reached much sooner a t lower pressures than was calculated by the Berg and James equation. This seems entirely reasonable due t o the much greater vapor velocity at reduced pressures for a given weight throughput. A value of 1.074 (83)was used throughout for the relative volatility of the n-heptane-methylcyclohexane system. The relative volatilities for the n-decane-tmm-Decalin system a t reduced pressures were taken from a curve of the relative volatility versus pressure plotted from the data obtained here and those of Fenske, Myers, and Quiggle (16). The values used to plot this curve were as follows: Pressure, Mm. Hg 100 50 20 10

Relative Volatility 1.22 1.19 1.145 1.11

The relative volatility corresponding to the arithmetical average of the head and still pressures was used throughout, although this was practically the same as that corresponding to the head pressure except a t 10 and 20 mm. DISCUSSION OF RESULTS

As a check on the operation of the vapor-liquid equilibrium still under vacuum conditions, the vapor-liquid equilibria on ethyl alcohol-water a t 95 mm. were determined. The results, Table 11, agree with those of Beebe et al. (1) a t both ends of the composition range, but show a systematically greater separation in the central portion. In an attempt to check the results of Beebe et al., a number of runs were made with a high current to the side winding and a t a high rate of boiling, both of which tend to give less than one theoretical plate separation. As the results did not change from the previous values, it seems possible that the data of Beebe et al. are somewhat low in this region. After completing the preliminary work, vapor-liquid equilibrium tests were run on n-decane-trans-Decalin a t 50-, 20-, and 10-mm. pressure. The results of these tests are shown in Table 111. The tests a t 50 mm. check the results of Fenske, Myers, and Quiggle (16)fairly well, the present data showing an average relative volatility of 1.20 as compared with their value of 1.185. An average value of 1.19 was used for the efficiency tests. The tests a t 20 and 10 mm. gave average values of 1.145 and 1.11, respectively. At all pressures, the TABLE 111. VAPOR-LIQUID EQUILIBR~UM DATAFOR ~-DECANE-~~UTL~-DECALIN relative volatilities were independent of MIXTURESAT REDUCED PRESSURES composition. Mole Fraction Vapor Pressure Activity The boiling points, corresponding t o n-Decane in Relative ~ ~ i lat iB.P., ~ Mm. ~ Hb Coefficientd Liquid, Vapor, Volatilitya, Point, Decane, Decalin. Decane, Decalin, compositions listed in Table 111and nec2 II a 'C. PAb Psa YA YB essary for the calculation of activity coAt 50-Mm. Absolute Pressure efficients, were taken from large scale 0.256 0.295 1.215 95.35 59.6 47.4 0.967 0.999 -0.0146 0.0000 curves drawn from the data of Table IV. 0.257 0,295 1.21 95.35 59.6 47.4 0.963 1.000 -0.0164 0.0000 0.469 0.516 1.205 94.0 1.011 56.4 45.1 0.974 -0.0114 0.0048 The values for pure n-decane agree with 0.637 0.676 1.19 93.0 54.3 43.2 0.978 1.032 -0.0097 0.0137 results given by Stull (SI),but those for 0.637 0.678 93.0 1.20 54.3 1.026 43.2 0.980 -0.0088 0.0112 0.844 0.866 1.195 91.95 51.9 41.6 ' 0.988 1.033 -0.0052 0.0141 trans-Decalin were considerably different At 20-Mm. Absolute Pressure from the generally accepted values of 0 . 2 5 8 0.285 1.145 72.9 22.96 18.92 0.963 1.019 - 0 , 0 1 6 4 0.0082 Seyer [see reference @ I ) ]which appear to 0.288 0.262 1.14 72.9 22.96 18.92 0.958 1.019 -0.0186 0.0082 0.460 0.494 1.145 72.03 22.07 18.24 0.973 1.026 -0.0119 0.0111 have been determined on impure mate0.466 0.502 1.155 72.0 21.98 18.20 0.980 1.025 -0.0088 0.0107 rials. The values of Fenske, Myers, and 0 . 7 3 3 0.759 1.145 70.9 20.97 17.34 0.988 1.041 - 0 , 0 0 5 2 0.0175 0.842 0.858 1.135 70.45 20.51 16.98 0.995 - 0 , 0 0 2 2 0.0249 1.059 Quiggle (16)for trans-Decalin seem to be At 10-Mm. Absolute Pressure more reliable because of greater purity 0.262 0.282 1.105 58.15 11.40 9.66 0.945 1.006 0.0246 0.0026 and are in agreement with those found 57.3 0.497 0.523 1.11 9.29 0.960 10.96 1.021 -0.0177 0 . 0 0 9 0 0.500 0 . 5 2 8 1.12 57.3 10.96 9.29 0.964 1.017 -0.0159 0.0073 here. The vapor pressure of trans0.684 1.105 56.65 0.705 10.63 9.02 0.969 1.036 -0,0139 0.0154 Decalin from 10 t o 200 mm. can be 0.732 1.115 0,753 10.53 56.5 8.94 0.977 1.031 -0.0101 0.0133 0 . 8 7 4 0.884 1.10 66.1 10.30 8.76 0.982 1,052 -0.0079 0.0220 represented by the equation I

~

~

-

a

Relative volatility =

-.

the data of Boord in reference (SI). Taken from a large scale plot of present data. d Activity coefficient = 3. b Taken from a large so&-plot"Af C

Px

where P

= absolute pressure, mm.

T = absolute temperature, 'K.

I

INDUSTRIAL A N D ENGINEERING CHEMISTRY

80

The complete holdup data for all of the packings a t various pressures and throughput rates (50) have not been included, but typical curves are shown in Figures 1 and 2. Total holdup was found to be fairly linear with distillation rate for all packings except at low rates and a t atmospheric pressure. This is in agreement with the results reported by Collins and Lantz ( 9 ) and Elgin and Weiss (11). The total holdup of the 32-inch packed section at atmospheric pressure is shown as a function of distillation rate in Figure 1 for n-heptane. Except a t very low rates, Cannon packing with hole size A has the highest holdup, followed in order by McMahon packing, Raschig rings, and helices. [Cannon has informed the authors ( 6 ) that a new packing with Iarger perforations, which he has designated as hole size B, gives the lowest pressure drop per plate of any packing reported t o date.] It should be pointed out that the effect of packing size would be considerable. Thus, if 3/32-in~h helices were used, the holdup would be higher than for the 6/sn-inch size used here. Likewise, if the 0.25-inch Cannon packing were used, it could be expected to have a lower holdup than the values found for the 0.16 X 0.16 inch size. 60

b'

Vol. 42, No. 1

packing to be used for vacuum frnctionation are very important. While a lower pressure drop for given distillation rate may be obtained by the use of larger diameter columns, this cannot be done without lowered efficiency (I?') and greater holdup, both particularly detrimental in small scale analytical columns. For all packings, pressure drop data versus distillation rate a t constant head pressure were correlated well by log-log plots. Straight lines resulted in all cases except a t atmospheric pressure and a t very low rates. Of the packings tested, Cannon packing showed the greatest pressure drop a t a given rate, followed by helices, MclIahon packing, and Raschig rings. The greatest pressure drop for a given throughput rate was at the lowest pressure. The maximum pressure drop before flooding was also found to increase with decreasing pressure. I n common with Hagy ( l Y ) , a log-log plot of distillation rate versus head pressure at constant pressure drop was found to give a straight line. This is shown in Figure 3. The efficiencies of the four packings a t various pressures are given iri Tables V to T'III. For all tests the number of theoretical plates was very close to the number of transfer units found. No subscripts have been used on the height of transfer unit (H.T.U.) since the only values used are the over-all results based on the gas film. The results for the various packings at atmospheric pressure are shown in Figure 4 where height of transfer

B

$ 50

y&

w

at 7

I

55

13O

b-

40

LL

2 J

a3 30

4 r

"0

402

800

0 CANNON 0

MCMAHON

1200 1600 2000 LIOUIO RATE, ML/HR. a RASCHIG RINGS 0 HELICES

2400

Figure 1. Holdup us. Throughput for'several Paclcings at 730 M m . of Mercury Head Pressure

A typical curve showing tho effect of head pressure on holdup at constant pressure drop is shown in Figure 2 for hfcMahon packing. A similar effect was found for all of the other packings tested. The lower the head pressure, the lower the holdup at any given pressure drop and the higher the pressure drop a t a given head pressure, the higher the holdup. A plot of the logarithm of the head pressure against holdup was found to be a straight line a t any constant pressure drop for McMahon saddles and helices. Tho values for Cannon packing showed some curvature. For Raschig rings tests were made a t only 10 and 100 mm. and, consequently, an insufficient number of points are available to determine whether curvature occurs. The pressure drop or back pressure obtained during rectification rapidly assumes increasing importance as the head pressure is lowered. Consequently, the pressure drop characteristics of a

TABLE IV. PROPERTIES OF ~-DECANE-~TCL~S-DECALIN MIXTURES Refractive Boiling Points, O C . , at the Pressure Given Index, (Mm. Hg, Absolute) n %o 10 20 50 100 200 1.4692 74.0 0.00 0.00 59.2 97.1 116.9 137.25 1.4612 12.34 12.03 55.75 7 3 . 5 96.3 114.9 136.0 1.4830 25.27 58.2 72.95 95.4 24.73 113.85 134.75 36.71 1.4458 57.7 94.65 112.8 72.4 37.38 133.6 1.4382 50.42 49.71 57.25 7 1 . 9 93.9 111.9 132.45 1.4313 71.35 93.1 62.62 61.95 56.9 110.95 131.4 1.4247 75.05 74.50 70.8 56.6 92.45 110.2 130.4 1.4181 87.23 87.54 7 0 . 3 ' 91.8 56.1 109.3 129.3 1.4119 LOO. 00 100.00 69.8 55.7 91.2 108.6 128.3 a Calculated from the weight per cent values using the following molecular weights: n-decane = 142.275, trans-Docalin = 135.244. n-Decane Present Weight Mole

%

%"

PRESSURE DROP, INCHES OF WATER

Figure 2.

Variation of Holdup with Pressure Drop for AIcMahon Packing

n-Heptane at 730 mm.; n-decane at 100, 5 0 , Z O . and 10 mm.

TABLE.'1

EFFICIENCY TESTSO S 0.25-IXCHRASCHlG RINGS -N,irnher .-.. - -.

Pressure of Number Liquid Drop, Theoof H.E.Rate, Inches of retical Transfer T.P., H.T.U MI./Hr. Water ma zob Plates Units InchesC I n c h e d Atmospheric Pressure, Tested with n-Heptane-Methyloyolohexane 596 0.21 0.762 0.861 9.2 8.2 3.9 3.5 595 9.1 0.21 0.758 0.857 8.1 4.0 3.5 940 7.4 0.39 0.762 0.854 8.4 4.3 3.8 940 0.39 0.845 7.4 0.762 8.4 4.3 3.8 1680 0.762 7.3 1.43 0.844 6.4 5.0 4.4 1800 1.76 0.844 0.762 7.3 6.4 5.0 4.4 100 Mm. Hg, Tested with n-Decane-trans-Decalin 460 0.129 0.12 0.607 10.8 10.6 3.0 3.0 500 0,137 0.659 11.6 0.16 11.4 2.8 2.8 530 0.21 0,137 0.665 11.7 11.5 2.7 2.8 650 0.49 0.137 0.603 10.4 10.1 3.1 3.2 680 0.57 0.137 0.602 10.4 10.1 3.1 3.2 930 2.26 0.139 0.606 10.4 10.1 3.1 3.2 960 0.139 0.603 10.3 2.62 10.0 3.1 3.2 10 Mm. Hg, Tested with n-Decane-trans-Decaiin 115 0.31 0,131 0.280 8.2 7.5 3.9 4.1 140 0.45 0.131 0.280 8.2 7.5 3.9 4.1 340 0.131 2.46 0.321 10.0 9.8 3.2 3.3 360 2.87 0.131 9.6 0.311 9.3 3.3 3.4

d

H.T.U.

=

height of a transfer unit.

INDUSTRIAL AND ENGINEERING CHEMISTRY

January 1950

TESTSON 0.26-INCH MCMAFION TABLE VI. EFFICIENCY PACKINQ Pressure Liquid Drop Rate Inches ’of MI./&. Water a75 540 575 1580 1610 2210 2340 290 520 670 900 315 560 710 85 270 620 100 165 245 465

50

XU

Number of Number Theoof H.E.. retical Transfer T.P., H T.U Plates Units Inches Ihches’

n-Heptane-Methylcyclohexane a t 730 Mm. 0.17 0.372 0.917 40.0 40.6 0.80 0.97 33.0 33.4 0.880 0.392 0.25 31.9 32.2 1.00 0.372 0.861 0.27 0.774 22.4 22.6 1.43 0.392 1.72 1.46 0.754 21.9 22.0 0,372 1.76 0.722 19.5 19.6 1.64 3.28 0.375 20.6 1.56 0.395 0.752 20.5 3.68

0.79 0.98 0.99 1.42 1.45 1.63 1.55

n-Decane-trans-Decalin a t 100 Mm. 0.998 48.7 47.3 0.034 0.062 0.990 35.9 36.8 0.990 35.9 36.8 0.062 0.981 32.6 33.1 0.063

0.64 0.87 0.87 0.97

0.27 1.07 2.05 4.10 0.74 2.95 5.12

0.68 0.89 0.89 0.98

0.054 0.057 0.062

At 50 Mm. 0.996 47.2 0.977 36.5 0.962 32.9

48.6 37.0 33.3

0.68 0.88 0.97

0.66 0.86 0.96

0.049 0.054 0.059

At 20 Mm. 0.998 66.3 0.977 46.9 0.890 33.0

68.1 47.5 33.2

0.48 0.68 0.97

0.47 0.67 0.96

0.044 0.044 0.041 0.056

At 10 Mm. 0.988 68.5 0.964 57.0 0.915 47.8 0.813 34.6

69.0 57.0 47.5 34.1

0.47 0.56 0.67 0.93

0.46 0.56 0.67 0.94

0.25 1.76 6.03 0.74 1.52 2.74 7.13

81

unit is plotted as. a function of liquid rate on logarithmic scales. These produce straight lines for all packings as suggested by Molstad, McKinney, and Abbey ($9). Under the conditions used here, the McMahon packing is the most efficient, followed in order by Cannon packing, helices, and Raschig rings, although there has been some indication that in larger diameter columna the Cannon packing may be more efficient than the comparable McMahon packing (6). Another point shown by these curves is that the height of transfer unit decreases with decreasing rate of distillation. This same type of curve was found by Richards (99)and others using small columns. The same general relationship among the packings holds at reduced pressures also, aa can be seen from the data in the tables. The superiority of McMahon packing over helices and Cannon packing in these 100-

50

HTU DETERMINE0 ATA CONSTANT RATE OF 250 MLlHR

i

0 30 8 z eo

2 10

05

046

IO

20 30 50 100 200 300 AVERAGE COLUMN PRESSURE, MM HG

0 RASCHIG RINGS 0 HELICES

500 730

A CANNON PACKING 0 McMAHON PACKING

Figure 5. Variation of H.T.U. with Average Column Pressure at Constant Throughput

P x

tests increased slightly a t lower pressures. This can readily be seen in Figure 5 which shows the change in height of transfer unit with the average column pressure. An unusual feature of these curves is the fact that the lowest heights of transfer units were under vacuum conditions. As the tests at 50 t o 100 mm. showed the lowest height of transfer unit of any of the vacuum 6

10 20 50 50 IM) ZOO HEAD PRESSURE, MM.HG RASCHIG RINGS a HELICES 0 MoMAHON PACKING A CANNON PACKING

F i g u r e 3. V a r i a t i o n o f Distillation Rate with Head Pressure a t Constant Pressure Drop of 2 Inches of Water

1 - 1 0.5

0.4

100

200 300

500

1000

2000 3000

TABLE VII. EFFICIENCY TESTS ON ‘/=-INCH SINGLB-TURN HELICES



Pressure Liquid Drop Rate, Inches’of MI./Hr. Water

ZY

140 270 290 305 1030 1940 2000

n-Heptane-Methylcyclohexane at 730 Mm. 0.12 0.295 0.692 22.6 22.5 1.42 0.17 0.642 20.5 20.4 1.56 0.279 0.295 0.18 0.645 19.6 19.5 1.63 0.18 0.628 19.6 0.279 19.5 1.63 1.15 0.544 0.279 14.8 14.6 2.16 4.10 ’ 0.279 0.527 2.32 13.8 13.7 2.32 4.47 0.527 13.7 0.279 13.8

275 670 1140

0.23 2.05 7.38

160 385 750

0.25 1.52 5.91

0.050 0.049 0.046

At 50 Mm. 0.935 31.2 0.907 29.2 0.856 26.6

60 260 270 540

0.16 1.85 2.18 5.94

0.063 0.062 0.065 0.065

60 175 255 260 390

0.49 2.09 3.57 3.64 6.52

0.059 0.047 0.049 0.049 0.046

n-Decane-trans-Decalin a t 100 Mm. 0.064 0.954 29.3 28.0 0.064 0.937 26.2 26.8 0.064 0.917 24.7 24.6

DISTILLATION RATE, MWHR. 0 RASCHIG RINGS P HELICES

A CANNON PACKING 0 McMAHON PACKING

Figure 4. Results of Efficiency Tests at Atmospheric Pressure Using Log Scales

OI

Number of Number Theoof H.E.retical Transfer T.P., H.T.U PlatUnits Inches Inahes’ 1.42 1.57 1.64 1.64 2.19 2.34 2.34

1.09 1.22 1.30

1.14 1.19 1.30

31.3 29.1 26.1

1.03 1.10 1.20

1.02 1.10 1.23

At 20 Mm. 0.838 31.3 0.678 24.2 0.665 23.5 0,609 20.8

30.7 23.6 22.9 20.2

1.03 1.32 1.36 1.54

1.04 1.35 1.40 1.58

At 10 Mm. 0,591 28.9 0.417 23.0 0.353 19.4 0.361 19.7 0,295 16.4

28.2 22.4 18.7 19.0 15.7

1.11 1.39 1.65 1.63 1.95

1.13 1.43 1.71 1.69 2.04

INDUSTRIAL AND ENGINEERING CHEMISTRY

82

TABLEVIII. EFFICIENCYTESTSON CAKNONPERTRUDED PACKISG Pressure Liquid Drop, Rate, Inches of Ml./Hr. Water

Number

zto

20

of Number Thew of H.E.retical Transfer T.P., H.T.U., Plates Units Inches Inches

n-Heptane-Methylcyclohexane a t 730 M m. 215 440 730 1200

0.14 0.35 1.01 3.77

0.278 0.278 0.275 0.278

180 250 380 530

0.13 0.33 1.15 3.16

0.057 0.062 0.056 0.057

0.996 0.990 0.964 0.964

140 320 440

0.29 2.01 4.22

0.060 0.052 0.046

0,994 0.948 0,917

0.765 0.715 0.678 0.642

28.9 25.3 23.0 20.6

1.11 1.27 1.39 1.55

1.11 1.27 1.40 1,56

42.4 36.8 30.3 30.1

0.78 0.89 1.07 1.08

0.75 0.87 1.06 1.06

45.4 32.3 29.8

0.72 0.99 1.06

0.71 0.99 1.07

28.9 25.2 22.9 20.5

n-Decane-lrans-Decalin a t 100 M m . 40.8 35.8 29.8 29.6

Vol. 42, No. 1

velocities resulting in better film coefficimts. Under the conditions used here, McMahon packing was found to be the most efficient, closely followed by Cannon packing at all pressures. Helices were somewhat poorer, and Raschig rings much poorer. This same relationship held at each pressure, the effect of pressure on column efficiency apparentlv being independent of the packing material used. ACKNOWLEDGMEYT

The authors acknowledge the kindness of 11.R. Fenske and of hI. R. Cannon of The Pennsylvania State College for making available the h ~ l i c eand ~ the pertruded packings used in this study.

A t 50 Aim. 44.2 32.4 30.1

A t 20 N m . 60 110 160 350

0.20 0.74 1.44 6.27

0.057 0.050 0.050 0,047

0.971 0.913 0.876 0.716

35 90 170 260

0.41 1.48 2.30 5.82

0.070 0.066 0.062 0.060

0.678 0.603 0.585 0.580

45.7 37.8 34.8 26.6

46.2 37.5 34.4 25.0

0.70 0.85 0.92 1.20

0.69 0.85 0.93 1.28

30.1 27.0 26.0 23.5

1.05 1.16 1.20 1.32

1.06 1.18 1.23 1.36

A t 10 M m , 30.6 27.6 26.6 24.2

results, one might be inclined to blame the highei height of transfer units a t atmospheric pressure on different test mixtures used. However, Myers ($3)reported that the efficiency of a column tested with n-decane-trans-Decalin at atmospheric pressure was the same as when tested with n-heptane-methylcyclohexane. I n connection with the efficiency tests on helices, it was found that upon refluxing overnight at very low rates aftpr preflooding, the thoroughly flooded condition was lost and thp packing behaved as dry packing if the rate was subsequently increased. This is in line with previous reports of tests a t atmospheric pressure ( l 4 ) ,but is contrary to the findinga of Myers (23)who reported that preflooding made little or no difference for helices when operating a t reduced pressures. I n vien- of the much greater difficulty of flooding helices thoroughly at reduced pressures, it appears possible that none of Myers’ tests were made with a thoroughly wet packing. Although no specific tests were made, the other packings tested gave no indication of any difficulty at low rates. SUMMARY

Vapor-liquid equilibrium data for the system n-decane-transDecalin were found to give relative volatilities substantially independent of composition, averaging 1.145 and 1.11 a t an absolute pressure of 20 and 10 mm., respectively. The change of relative volatility with pressure was found to be an exponential function from 10 to 100 mm. Total holdup of the column was found t o vary linearly with distillation rate for a given head pressure drop, and to be higher at lower head pressures for constant distillation rate. Pressure drop was found to be a function of distillation rate and to give straight lines on log-log graphs. Efficiencies increased with decreasing flow rates at a given pressure and a plot of log height of transfer unit versus log distillation rate gave a straight line for each packing. The maximum efficiency for all packings occurred in the region of 50 t o 100 mm. of pressure, with slightly lower efficiencies resulting a t 10, 20, and 730 mm. The assumption that greatly decreased efficiencies would result from operation at reduced pressures has been shown t o be incorrect, probably due, in part a t least, to greater vapor

LITERATURE

crrm

Beebe, ,4.H., J r . , Coulter, K. E., Lindsay, R. A , , and Bakel. E , A[., IND. E N G . CHEM., 34, 1501 (1942). Berg, C., and James, I. J., Jr.. Chem. Eng. Progress, 44, 307 ( 1948) Berg, L . , and Popovac, D . O., paper presented before the Am. Inst. Chem. Engrs. a t New York City, Kovember 1948. P h . D . thesis, University of Pittsburgh, 1942. Bishop, C. d., E X G . CHEY.,34, 1088 Bragg, L . B . , and Richards. A. R., IND. (1942). Cannon, M. R., private communication (Aug. 8 , 1949). and Lewis, W.K., IND. E N G .CHmf., 24, 882 (1932). Carey, J. S., Chilton, H. T . , and Colburn, A. P., Ibid., 27,255 (1935). Collins, F. C., and Lantz, V., A n a l . Chem., 18, 673 (1948). Coulter, K . E., Lindsay, R . A , , and Baker, E. M., Isn. J ~ s G , CHEM.,33, 1251 (1941). Elgin, J. C., and Weiss, F. B., I b i d . , 31,435 (1939). Feldman, J., Myles, M . , Werider, I., and Orchin, M., Ibid., 41, 1032 (1949). Fenske, M . R . , “ T h e Science of Petroleum,” Vol. 11. p. 1629, London, Oxford University Press, 1938. Fenske, M . R.. Lawroski, S..and Tongberg, C. 0.. I s n . ENG. CHEM..30, 297 (1935). Fenske, M. Ii., Myers, H. S., and Quiggle, D., Ibid., t o be published. Fenske, M.It., Quiggle, D.. and Tongberg, C. O., Ibid.. 24, 408 (1932). Fenske, M . It., Tongberg. C . O., and Quiggle, D . , Ibid., 26, 1169 (1934). Gilmont, R., and Othmer, D. F., Ibid., 36,1061 (1944). H a g y , J. D . , M.S. thesis, T h e Pennsylvania State College, 1941. Hinkel, 11. D.. P h . D . thesis, T h e Pennsylvania S t a t e College, 1946. Langdon, TT’. M.,and Keyes, D. B . , ISD.E N G . CHEM.,34, 938 (1942). Molstad, M. C . , M c K i n n e y , J. li., and Abbey, 11. G., T r a n s . Am. I n s t . Chem. E n g r s . , 39,605 (1943). Myers, H. S . , M.S. thesis, T h e Pennsylvania State College, 1948. (24) Othmer, D. F., IND.ENG.CHEM.,20, 743 (1928). (25) Othmer, D . F., and Benenatti. R. F., I b i d . , 37,299 (1945) (26) Othmer, D. F . ,and Josefowitz, S., I b i d . , 39, 1175 (1947). (27) Ot,hmer, D. F., Schlechter, N., and Koszalka, TV. A,, Ibid., 37, 895 (1945). (28) Perry, E. S., and F u g u i t t , l i . E., I b i d . , 39,782 (1947). (29) Richards, R. B., M.S. thesis. The Pennsylvania State College, 1941. (30) Struck, R . T.. P h . D . thesis, T h e Pennsylvania S t a t e College, 1949. (31) St,ull, D . It.,IsD. E N G .CHEM., 39,517, 1684 (1947). (32) Williams, F. E., I b i d . , 39,779 (1947). (33) Willingham, C . B . , and Rossini, F. D., J . Research S a t l . B u r . S t a n d a r d s , 37, 15 (1946).

.

RECEIVED June 17, 1949. This paper is based o n a thesis presented by R. T. Struck to the faculty of The Pennsylvania State College in partial fulfillment of the requirements f o r the Ph.D. degree.