November 15,1934
I N D U S T R I A L A N D E NG I NE E R I N G C H E M I S T R Y
APPARATUS The apparatus, assembled as shown in Figure 1, consists of an absorption apparatus in connection with a magnetic gas balance. The buret, A , has a capacity of 200 cc. and is graduat,ed in 0.1 cc. It is in connection wit,h the absorption apparatus and the gas balance through the Karlsruhe sto cock, 5 . The auto-bubbler, B , contains 33 per cent potassium Rydroxide for the removal of carbon dioxide and hydrogen sulfide. Pipet C, filled with 6-mm. glass beads, contains 87 per cent sulfuric acid for the absorption of propene and butene. Two other pipets similar to C, containing 63 per cent sulfuric acid and activated sulfuric acid, sre used for determining isobutene and ethylene. The reservoir, D, and the leveling bottle, E , contain salt water, the solution in D being used for flushing the manifold through stopcocks 1,2, 3, and 4. Tra,ps F and G, containing the same solutions as the pipets, protect the solutions in the pipets from the atmosphere. The drying tube (8, Figure 2) contains anhydrone for drying all gases that pass into the gas balance. The precision manometer and the magnetic gas balance have been described in an earlier paper ( 2 ) .
PROCEDURE The calibration of the gas balance is made as previously outlined (9). At the beginning of the analysis, all gas is removed from the buret and the manifold by displacement with salt solution. A sample of approximately 200 cc. is taken into buret A t,hrough stopcock 2, and passed into the potassium hydroxide solution to remove carbon dioxide and hydrogen sulfide. The volumes are recorded and approximately 100 cc. of the gas passed into the evacuated balance. The volume of the remaining gas is recorded and passed into pipet C, containing 87 per cent sulfuric acid, for 3 minutes. While this absorption is taking place, the molecular %-eightof the portion of the sample in the balance is determined. The milliammeter reading, the temperature of the balance, the manometer reading, and the temperature of the manometer are recorded. The balance is evacuated, and at the end of the 3minute period the gas is passed from pipet C back into the buret for the volume reading, then into pipet C for a second 3-minute period if the first, absorption shows an appreciable quantity of propene or butene. After the second absorption by 87 per cent acid, the gas is passed into the potassium hydroxide pipet, B, the ma,nifold flushed with salt solution, and the volume recorded. The gas is then passed into the evacuated balance for the molecular weight determination of the propene- plus butene-free gas, recording the same data as for the first molecular weight determination. The analysis requires about 10 to 15 minutes.
CALCULATIONS The volume per cent of propene plus butene in the original sample is calculated from the volume absorbed by 87 per cent
405
acid and the per cent of carbon dioxide and hydrogen sulfide removed by potassium hydroxide solution. The apparent molecular weights, M , of the original and of the propene- plus butene-free gas, as read from the current molecular weight relation, are corrected to the molecular weight at 0" C. and 760 mm. pressure, MI and M z , by means of the formula M1orMs
=
M X
760 t + 273 X P 273
The following relations are used for calculating the volume per cent of propene and of butene in the original sample: Per cent C3H6= 4U
+ Mz
- Mi
14.02
MJ
(2)
Per cent C4H8 = U - per cent C3Hs (3) where MI = molecular weight of original gas M a = molecular weight of gas after removal of propene and butene U = volume per cent of propene plus butene in original sample
ANALYSIS OF SYNTHETIC MIXTURES Representative analyses made of a synthetic mixture and the composition of the synthetic mixture are shown in Tables I and 11.
TABLE I. DETERMINATION OF OLEFINS BY MOLECULAR WEIGHT METHOD VOLUME
PER CEINTOF
VOLUM~ PERCENTOF n-BUTENE
26.0 24.5 24.8 24.8 24.5 24.3 24.6 26.0
9.4 9.9 9.4 9.6 9.8 9.9 9.7 9.3
PROPBNB
DETBRMINATION 1 2 3 4 6 6
Av.
Calod.
TABLE 11. COMPOSITION OF SYNTHETIC MIXTURE Methane Ethylene
% 63.8 11.9
Propene n-Butene
% 25.0 9.3
LITERATURE CITED (1) Tropsch and Dittrich, Bransto.f-Chem., 6, 169 (1925). ( 2 ) Tropsch and Mattox, IND.ENQ.CHEM.,Anal. Ed., 6, 236-41
(1934). (3) Tropsch and Mattox, Universal Oil Products Booklet 138 (1934). RBC~IVFID June 22,1934.
Determination of the Gasoline Content of Gases HANSTROPSCH AND W. J. MATTOX, Universal Oil Products Company, Riverside, Ill.
A ''
S U M M A R Y of t h e A low-temperature fractional condensation carbons CKand higher are then method for &termining the gasoline content of vaporized into a known VOhme methods used previously where the pressure and temperadetermining t h e gases has been developed and its application to ture Thegasogasoline content of gases has been the analysis of a wide variety of refinery gases line vapors are condensed by given in an earlier review ( 2 ) . The m e t h o d described here has shown it to be suitable for the rapid, accurate, liquid nitrogen in a bulb which is removed from the apparatus and for the d e t e r m i n a t i o n of the routine analysis of these gases. gasoline c o n t e n t of gases conaccurately weighed. From the weight of the gasoline and the sists in condensing the hydrocarbons at the temperature of liquid nitrogen (- 195' C.) and pressure it exerts when vaporized in a known volume at a defiatmospheric pressure, removing methane and other difficultly nite temperature, the average molecular weight of the gasoline condensable gases, especially dissolved nitrogen, under re- is calculated. From a curve relating the molecular weights to duced pressure, and at low pressures fractionating the hydro- the specific gravities of the pure hydrocarbons, the specific carbons in a U-tube cooled t o -105" C., a temperature at gravity of the gasoline is read. With the weight of the gasoline which the hydrocarbons above C4are condensed. The hydro- from a known volume of gas and the specific gravity of the
ANALYTICAL EDITION
406
Vol. 6, No. 6
to the height of the salt water above the lower outlet tube of the gas holder. Stopcocks K, L, and R are closed and U-tubes E and G cooled by immersing in liquid nitrogen contained in Dewar flasks F and H . Screwclamp B is closed and stopcock A slowly opened to fill the gas velocity indicator (containing a small amount of mercur ) with gas. After some experience in condensing the sam les, t[e velocity indicator may be omitted and the rate judged Prom the pressure increase in the apparatus. The flow of gas into the condensation tubes is regulated by screwclamp B to a rate of about 100 cc. per minute. The pressure in the apparatus is gradually increased by methane and gases other than hydrocarbons which are not completely condensed, and if the pressure in the apparatus reaches atmospheric, stopcock Q is opened to allow these uncondensed gases to pass from the apparatus. A / L L /VOLTMETER When the gas sample is displaced from the gas holder and the tubes to A, stopcocks A and Q are closed and the apparatus is slowly evacuated through R. If the evacuation takes place too rapidly, a part of the hydrocarbons is sometimes swept out of the U-tubes. While the uncondensed gases are being removed, the evacuated bulb N is removed and weighed to 0.1 mg., replaced on the apparatus, and L opened. When the pressure in the apparatus has been reduced to about 0.1 mm., D is closed and U-tube E allowed to warm slowly to distill the hydrocarbons into the second U-tube, G. Stopcock I is then closed and the hydrocarbons distilled back into E,removing through R any gases not condensed. These distillations are necessary to remove dissolved methane and nitrogen. During these distillations, the aluminum block, P, is cooled to OR/D€ L/M& VELOC/TY a temperature a few degrees below -105" C. The block is then /NDICATOR laced around U-tube G, keeping the temperature a t -105' C. /GAS HOLDER (SALT WRT&& gy the occasional addition of small amounts of liquid nitrogen at. S. Sto cock I is opened and the hydrocarbons in E are allowed FIQURE 1. APPARATUS FOR DETERMINING GASOLINECONTENT to distil~slowlya t about 1 mm. pressure or less through G where OF GASES the gasoline fraction is condensed. With gases containing appreciable quantities of butanes, a small amount of the C4 hydrogasoline the gallons of gasoline per 1000 cubic feet of wet gas carbons is condensed tvith the gasoline fraction but is completely at 60' F. and 30 inches of mercury are calculated by substi- removed by a second condensation. With all the hydrocarbons removed from E and the pressure reduced to less than 0.2 mm., tuting the values in a simple formula. R is closed, K and M are opened, and U-tube G is allowed to come to room tem erature. The pressure of the vaporized gasoline, as indicated y! the manometer, is recorded, together with the APPARATUS room temperature and the height of the mercury in the manometer tube. The apparatus is shown diagrammatically in Figure 1.
H
A gas holder of 3- to 5-liter capacity containing saturated salt solution is used for measuring the gas sample. Absorption towers containing moist soda lime, calcium chloride, and anhydrone, or dehydrite, remove carbon dioxide and water vapor from the gases entering the condensation tubes. The Dewar flasks, F and H , are used as containers for liquid nitrogen to cool the U-tubes, E and G, during the condensation of the hydrocarbons. The aluminum block P (1, Figure 3), fitting inside Dewar flask H and around U-tube G, is provided with hole S for the introduction of the liquid nitrogen used to cool the block. An iron-constantan thermocouple is used for the temperature measurements. The junctions of the iron and the constantan wires with the copper lead wires from the millivoltmeter are maintained a t a constant tem erature of 100' C. by immersion in steam. ' d e volume of J, about 500 cc. (accurately determined), and the apparatus from D to R provide a volume sufficiently large for the complete vaporization of the gasoline fraction. The volume of the apparatus, including U-tubes E and G and bulbs J and N , is determined from the known volume of J . A connection is made at R to a vacuum pump which is capable of reducing the pressure in the apparatus to less than 0.001 mm. of mercury. A manometer connected directly to the a paratus is used to indicate the pressures. The weighing bug, N , in connection with the apparatus through a ground-glass connection, is used to determine the weight of the gasoline fraction.
PROCEDURE While the gas sample is being measured, the apparatus is evacuated through R until the mercury in the manometer tube stands a t the same level as the barometer. The volume of the sample for the determination should be from 0.3 to 1 liter for gases containing more than 0.3 to 0.5 gallon of gasoline per thousand cubic feet (40 to 66 cc. per cubic meter) of gas, while for very lean gases (0.01 to 0.2 gallon of gasoline per 1000 cubic feet (1.3 to 26 cc. per cubic meter) as much as 1.5 to 3 liters is required. The sample volume is taken as the volume of the salt water displaced from the gas holder. The temperature of the sample is taken as the temperature of the salt water a t the time the sample is measured. Pressure corrections are made for the vapor tension of the salt water and the decrease in pressure due
0.710
I
I
I
I
N-HIPTANE
I
I
I
70
80
SO
0.6/0
I too MOL€CUIAR
I
-100
I //O WE/GH7
I
o
I
I
le0
/30
FIGURE 2. MOLECULAR WEIGHT-SPECIFIC GRAVITYCURVE FOR HYDROCARBONS Cs TO CS Bulb N is thencooledinliquid nitrogen contained in the Dewar flask, 0. The gasoline is condensed in N in a short time, as is indicated by the decrease in the pressure in the apparatus to 0.6 mm. or less. Small amounts of air in the apparatus make the complete condensation difficult, and for this reason air should not be admitted to bulb J or to other parts of the apparatus exce t when it is necessary to grease the stopcocks, etc. Stopcock is closed and weighing bulb N removed and weighed to 0.1 mg. as soon as it reaches room tem erature. The weight of the gasoline is obtained from this weiglt and the weight of the evacuated bulb.
!&
The analysis requires about 30 t o 45 minutes and approximately 150 to 200 cc. of liquid nitrogen.
TABLE11. SPECIFICGRAVITYOF HEXANEAND HEPTANEBY MOLECULAR WEIGHTRELATION
CALCULATIONS In this method the hydrocarbons C5 and higher are determined as the gasoline constituents in grams per liter of dry gas a t 0" C. and 760 mm. pressure. It is usually desirable to express the gasoline content in gallons per 1000 cubic feet of wet gas a t 60" F. and 30 inches of mercury. To make this conversion, it is necessary to know the specific gravity of the gasoline,, The average molecular weight of the gasoline is calculated from the weight of the gasoline and the pressure in a known volume a t temperature tz- Then from a curve, relating the molecular weights to the specific gravities of the pure hydrocarbons (Figure 2), the specific gravity of the gasoline is read. The following relations have been derived by which the calculations are easily made:
+
==
G =
407
I N DU STR I A L AN D EN G I N EER I N G CH E M I STR Y
November 15,1934
22410 X g X 760(tz 273) Pz(Vz- rr%) 273 273) 19.39 X g ( t l
HYDROCARBON Hexane Hexane Hexane Hexane Heptane
1 2 3 4 5
SPBICIFIC GRAVITY From By Westphal curve balance 0.671 0.662 0.662 0.660 0.662 0.666 0.662 0.664 0.684 0.687
The anaIysis of synthetic mixtures of nitrogen, butane, and pentane containing approximately 5 per cent of butane and 3 per cent of pentane has served as a further check on the ac-
(1)
+
s x PI x
SAMPLE No.
(2)
Yi
Where = average molecular weight of gasoline G = gallons of gasoline per thousand cubic feet of wet gas at 60" F. and 30 inches of mercury g = weight of gasoline, grams X = specific gravity of gasoline at 20" C. VI = volume of gas sample, liters Ye = volume of apparatus PI = pressure of gas sample, mm. Pz = pressure of gasoline vapor, y m . tl = temperature of gas sam le, C. ta = room tempe;ature wlen gasoline fraction is vaporized, C. h = height of mercury in manometer, cm. T = radius of manometer tube, om.
1l.I
FIGURE3. GAS EDUCTION CHART
TABLE111. DETERMINATION OF PENTANEIN SYNTHETIC NITROGEN-BUTANE-PENTANEI MIXTURES
ACCURACY OF METHOD The accuracy of the above method for determining specific gravity was checked by determinations on samples of commercial pentane, hexane, and heptane. The procedure was to introduce from 0.05 to 0.30 gram of the hydrocarbon into the apparatus, removing dissolved gases, making the molecular weight determination as outlined, and reading the specific gravity from the specific gravity-molecular weight curve, Figure 2. The data in Table I were obtained for a sample of pentane.
SAMPLBI
No.
Calculated
PENTANE
Ga1./1000 CU. ft.
1 2 3 4
1.02 1.08 1.09 1.15
Found Gal./lOOO CU. ft. 1.02 1.07 1.06 1.15
Analyses made on plant refinery gases by the low-temperature condensation method are compared, in Table IV, with analyses made on the same samples by the Podbielniak method.
TABLEI. SPECIFIC GRAVITYOF PENTANEBY MOLECULAR TABLEIv. ANALYSIS OF REFINERY GASES. COMPARISON WITH WEIGHTRELATION PODBIELNIAK METHOD SAMPLE
No.
PENTANE
SPECIFIC MOLECULAR GRAVITY WEIQHT (PROM CURVI)
DEVIATION MEAN SP. GR.
PROM
SAMPLE
No.
@am
1 2
3
4 5 6
7
8 9 10 11 12 13 14 16 16 17
0.0554 0.0736 0.0799 0.1202 0.1215 0,1262 0.1286 0.1412 0.1522 0.1574 0.2776 0,2755 0.2806 0.2837 0.2903 0.2950 0.3077
71.8 73.1 71.7 72.8 72.9 73.4 72.1 73.4 73.1 72.0 72.6 73.4 73.2 72.7 72.8 72.6 73.3
%E XvKiion from av. Sp. gr. by Westphal balance
-0.003 0.000 -0,003 0.000 0.000 +0.002 -0.003 $0.002 0.000 -0.003 -0.001 f0.002 0.000 -0 * 001 0.000 -0.001 +0.002
0.624 0.627 0.624 0.627 0.627 0.629 0.624 0.629 0.627 0.624 0.626 0.629 0.627 0.626 0.627 0.626 0.629 0.6265 0.0013 0.626
The results of similar determinations on hexane and heptane are summarized in Table 11.
Fractional condensation Ga1./1000 cu. ft.
METHOD Podbielniak Ga1./1000 cu. ft.
1 2 2 3
4 5
This method gives results in good agreement with those of the Podbielniak apparatus on gases of high gasoline content, approximately 2 gallons per 1000 cubic feet or over. With gases of low gasoline content, particularly if nitrogen is present, it is much more reliable than the Podbielniak apparatus. Gases which could not be analyzed by the Podbielniak method on account of the high nitrogen content, over 70 per cent, have been analyzed satisfactorily and without difficulty by the fractional condensation method. The method has been applied in the analysis of a large number of refinery gases. Typical exampIes are given in Table V.
ANALYTICAL EDITION
408 TABLE
v.
ANALYSIS OF GASES ENTERING AND
ABSORPTION PLANT
LEAVING
AN
(Gallons of gasoline per 1000 cubic feet of gas) GASENTBRINQ SAMPLE ABSORPTION EXHAUST APPROXIMAT. No. TOWER GAS RZCOVERY 1 0.26 0.08 0.2 2 0.34 0.16 0.2 2 0.35 0.16 0.2 3 0.43 0.12 0.3 4 0.92 0.21 0.7
samples were taken of the plant gases entering the absorption system and also of the exhaust gases leaving the plant.
Vol. 6, No. 6
The difference between the gasoline content of the exhaust gases and that of the gases entering the absorption system was in close agreement with the quantity of gasoline collected from the absorption. LITERATURE CITED (1) Tropsoh and Mattox, IND.ENQ. CHEM.,Anal. Ed., 6, 235-41 (1934). ( 2 ) Tropsoh and Mattox, Universal oil Products Booklet 138 (1934). R ~ C ~ I VJune E D 22, 1934.
Determination of Tie Lines in Ternary Systems Without Analyses for the Components THEODORE W. EVANS,Shell Development Company, Emeryville, Calif,
T
ERNARY liquid systems are generally characterized
by plotting on a triangular field the curves separating the regions of homogeneity and heterogeneity, together with the tie lines connecting the points representing conjugate phases. The boundary curves may usually be easily obtained by titrating a homogeneous mixture of two of the components with the third until turbidity appears, thereby securing one point. To determine the tie lines, however, no simple method, except that of Miller and McPherson ( I ) , seems to have been developed that does not require a direct
two layers, the composition of the lower layer being represented by a point on, say, the left-hand portion of the curve, the upper layer on the right-hand side. The line joining these two points is a tie line and passes through P. Through P draw the lines IPJ, KPL, and MPN. Ordinarily the composition of lower and upper layers can be estimated roughly and the lines drawn so that these compositions lie somewhere between I and M , and J and N . Assume this is the case here; in case it is not, the discrepancy will appear later and if necessary one of the lines can be changed accordingly. Let ml, m2 be the weights of the upper and lower layers obtained, mA, etc., the weights of the components taken, and L A , etc., the percentage of component A corresponding to the point L. In case KPL is the true tie line we have o.Ol(mlL~ f m2KB) = ma. In general KPL will not have been so chosen as to be the exact line, and hence O.Ol(rnlL~ ~ K Bwill : not equal me. However, by plotting the value of this expression for the three lines drawn against the percentage B of the points on the right-hand branch of the curve or the percentage A on the left branch, a value of these percentages will be obtained for which the e x p r e s s i o n does equal mB. By now e5 locating this point on the curve and joining it to P the tie line is secured. In the s a m e w a y t h e percentage A could have been used. The essential point to the process is that as we proceed I from I to M the-percentage B does not change 65 70 % 5 CORRFSPONDiNG TO J, L, ETC. greatly, and what change there is is in a positive FIGURE 2 direction, while from J to N the change is very rapid and positive, thus making the method relatively sensitive. It is evident that the C component would not be suitable, since it increases from I to M and decreases about equally from J to N , the two effects offsetting each other and rendering the variation in 0.01 (ml& m2Kc)too small to be useful. The following example will illustrate the met hod.
+
2
FIGURE 1 analysis for at least one of the components (8). Inasmuch as systems are often investigated in which none of the components can be accurately or easily determined it is desirable to have some rapid method of determining the tie lines which is independent of chemical analyses. The following method seems new, is applicable in the usual cases, and compares favorably with that of Miller and McPherson in ease of manipulation. In Figure 1 is shown a typical ternary diagram for the system of three components A , B, and C. Suppose weighed amounts of the three components are now taken in such amounts that the over-all composition is represented by the point P. Thi8 mixture upon shaking and settling will give
75L
+
.
