INDUSTRIAL A N D ENGINEERING CHEMISTRY
165
Vol. 17, No. 2
Influence of Operating Practice on Composition of Carbureted Water-Gas Light Oil’32 By R. L. Brown and H. G. Berger PITTSBURGH EXPERIMENT STATION, BUREAU OF MINES,PITTSBURGH, PA.
I
N PREVIOUS publica-
light oils and of the amount The composition of the light oils, including their total t i o n ~bearing ~ on the of gum-forming constituunsaturation figures and their styrene and indene conproblem of gummy deents therein. These light tents from five present-day carbureted water-gas plants, oils were recovered from posits in gas pipes and has been presented. The relation of this light oil comthe gases produced i n a meters it was pointed out position to the operation of the plant and to the meter and number of carbureted waterthat these deposits are demain conditions of the distributing system has been estabpendent basicly on the comgas plants in actual daily lished. Thus, from data secured from plant sources it has operation. The plants were position of the gas, and been found that thoroughcracking of the oil in the carbuselected and studied because more specifically on the reted water-gas plant gives a light oil in thegas of such comcomposition of the light oil of the known variation in position that there is little condensate (drip oils) and gummeter and pipe conditions of the gas. Their formation my material in the distributing system. That which is considered from the standwas discussed from both deposited is found near the plant because the light oil is point of gummy deposits, physical a n d chemical high in benzene and toluene and low in the high-boiling points of view. and because of their known and gum-forming constituents. These latter are the ones or suspected divergence in Indene and styrene were which give rise to the greater part of condensates and to those operating factors (in found to be the principal the meter deposits. The gum-forming constituents as the carbureting units) which gum-forming constituents, measured by the total unsaturation and the indene and are most important and a t and the properties of those styrene contents of the light oil have been found to dethe same time flexible and constituents were described. crease sharply with increase of temperature, with desubject to the operator’s I n two reportsga~” there was crease of rate of oil input, and with increase of closeness control-that is, temperaemphasized the co-existence in checker-brick spacing,. keeping within the respective ture, rates of oil input, and of gummy deposits with ranges of practical operation. contact in the checkerwork. the presence of natural conThese findings fully confirm previous work on a laboraIt is the purpose of this densate (drip oil) in meters tory scale and the statements of earlier reports. The data paper to show from these and in the outer parts of of this report furnish standards for recognizing desirable data, secured wholly from distributing systems genand undesirable light oil compositions from the standplant sources, the complete erally. Equally true was point of gummy meter deposits. interrelationships existing the physical association or generallv between the comabsence of such deposits and conditions in the of drip oil in the same parts of the systems. Similarly, the position of the light oils, the o:erating association of gummy deposits in meters with low-cracking carbureting units, and meter and main conditions considtemperatures, very high rates of oil input, and a low amount ered from the gummy deposit standpoint. The influences of contact was rendered evident. The reasons for these ob- exerted by the sequence of units and the size of units in the served facts were given and discussed.3b The influence of generating system (other than the carbureting units), and by the character of the gas oil used on the composition of the the time-temperature influences of those units, on the depolight oil and on its gum-forming constituents was consid- sition of gummy material are not to be considered here. ered.*cld The relationships of operating practice in the General Plan of Investigation carbureting units to the general composition of the light oil and to the amount of gum-forming constituents of the gas The plan is threefold in character-namely, a study of (1) produced was demonstrated e~perimentally.3~ With this meter and main conditions, ( 2 ) operating practice, and (3) was correlated the observed meter conditions, the natural light oil sampling and composition. condensate (drip oil) conditions, and practice in plant operation obtained in the survey of 1922. (1) I n brief, the condition of the meters was first observed with Scope and Purpose of Present Study
This present paper presents the results of a study made by the United States Bureau of Mines in cooperation with the American Gas Association, of the general composition of the 1 Presented under the title “Light Oil Compositions in Carbureted Water Gases: Relation to Operating Practice and Gummy Deposits in Pipes and Meters (Plant Data)” before the Section of Gas and Fuel Chemistry at the 67th Meeting of the American Chemical Society, Washington, D. C., April 21 to 26, 1924. 2 Published by permission of the Director, U. S. Bureau of Mines. * ( a ) Brown, Pvoc. A m . Gas Assoc.,Tech. Sect., 4,280 (1922), Gus AgeRecovd, 60, 571 (1922). ( b ) Brown, BUY.Mines, Repts. of Investigations 2503 (1923); A n . Gas Assoc. Monthly, 5 , 309 (1923); Gas Age-Record, 62, 273 (1923). (c) Brown, Proc. A m . Gas Assoc., Tech. Sect., 6, 1177 (1923); A m . Gus J.. 119, 530 (1923). ( d ) Brown, Pohlmann, and Berger, BUY.Mines, Repts. of Inwestigations 2637 (1923); Gas-Age Record, 62, 645 (1923).
reference to the amount, if any, of gummy deposits present, and
to the amount and character of any meter condensate that was present. The general conditions as to amount and character of condensate (drip oil) in the outer parts of the distributing system were likewise sought. I n most of the cases under examination careful analyses had been made of the drip oils of those systems. The general distribution of condensate in the distributing system of three of the plants had been followed for about a year previous to the time of the present investigation. ( 2 ) Those operating practices, particularly in the carbureting units, considered most influential in fixing the composition of the light oil in the gas being produced, were followed as closely as proved practicable throughout the several days or the week over which the light oil was being secured. (3) I n order t o ascertain the chemical character of the light oil4 in the gases of the five plants in which quite different meter 4 The term “light oil” is used to mean those oils boiling between 70° and ZOOo C., which appear in the gas. They are almost all members of the aromatic series of hydrocarbons.
February, 1925
INDUSTRIAL A N D ENGINEERING CHEMISTRY
and main conditions obtained, samples of that light oil from the gas being produced in those plants were taken over a period of at least 30 hours, either continuously or, more usually, throughout a week. The gas scrubbed for its light oil was in every case that coming from the storage holder and going to the city for consumption-that is, the gas being passed into the distributing system. The manner in which the scrubbing of the gas for the recovery of its light oil content was effected has been described in an earlier ~ a p e r . 3 ~ The apparatus used in the present work is shown in Figure 1. A supply of wash oil (pale paraffin or similar oil) from a was kept flowing through line b regulated by valves c by means of sight glasses g, through twin towers t ( 2 by 60 inches) filled with carefully sized * / p - to 1/2-inch pieces of coal, f, into and through the self-established and self-maintained oil seals, s, and the tubes, d, into a storage receiver, r. The surface-giving material f was supported above the oil outlet tubes, e, and gas inlet tubes, j , by plates of wire screening, $, of l/r-inch mesh. The gas to be scrubbed for its light oil was introduced into the system at i, metered a t m, and led through tubesj into the towers t , where, as it passed upward and countercurrent to the descending wash oil, its light oil was taken into solution by the wash oil. The resulting scrubbed gas passed out of the system a t 0 . The benzenized (light oil-bearing) wash oil was preserved for the subsequent recovery and examination of the light oil it contained. The total gas flow was a t the rate bf 12 to 16 cubic feet per hour, with this flow equally divided between the two towers of the system. No differential pressure in either tower developed a t any time. Such a pressure would have quickly manifested itself by the oil level in the tube e, as early special tests of the apparatus had shown. The downward flow of fresh wash oil was carefully regulated so as to give an enrichment of the wash oil of about 2.5 per cent, which insured complete removal of the light oil in the gas. (Benzene scrubbing in coke plants is conducted with 3 per cent enrichment, and high efficiencies a r e secured.) In the case of Plant 11, discussed below, however, the recovery of the light oil was effected by the use of pply a double coil scrubber.& The enrichment here was 2.9 Der cent: the rate of gas flow was.l.8 cubic feet per hour,, and the scrubbing efficiency was 95 per cent. Meter and Main Conditions
PLANT I-The meters of this system are free from deposits and have been for years. There is relatively little oily condensate in the mains, only 0.04 gallon of drip oil per 1000 cubic feet, and that is found close to the plant. PLANT 11-The meters of this d i s t r i b u t i n g system are likewise free f r o m d e p o s i t s . The amount of condensate (drip oil) is low, only 0.03 g a l l o n p e r 1000 0 cubic feet, despite the Figure 1-Light Oil Scrubber fact that part of the gas is distributed under 15 to 25 pounds pressure. The drip oil, as in Plant I, is principally in the drains close to the plant. PLAST111-The consumers’ meters of this system are decidedly gummy, and have been so for more than a year. No trustworthy record of the amount of condensate through6
Bird, Chem. Met. Eng., 22, 705 (1920)
169
out the distributing system was available, but its presence there is evidenced by the condition of the meters. PLANT IV-The meters here were in very bad condition 18 months ago, the gummy deposits being especially heavy, but the condition is rapidly improving. The total quantity of main condensate is now less, amounting to roughly 0.07 gallon per 1000 cubic feet.
m
L
FRACTIONS
Figure 2A-Light
Oil Composition
PLANT V-The meters of this system show slight deposits now, but considerably heavier deposits have been experienced in recent years. The amount of main condensate is heavy, and it contained (when examined) normally about 20 per cent of resinous material. The writers were unable to obtain exact figures, but two days were spent by one of them in investigating the condition of these mains and collecting samples, both in company with the “pumper” for this system and with special assistance. A detailed laboratory study of six samples of “drip oil” from the system has been made. The fact that a large portion of the gas from the plant is distributed under pressures up to 25 pounds, and is consumed at comparatively long distances therefrom, shows why the meter deposits are not heavy. Operating Conditions
In the present study consideration is given to those factors bearing on the generation of gas, especially those pertaining to the cracking operation. From the standpoint of accumulation of meter condensate and gummy deposits, the condensation processes in the plant, though important, are strictly secondary factors. This is true because the composition of the light oil in carbureted water gas when boor cracking obtains-that is, faulty operation of the carbureting units-
Figure 2B-Light
Oil Composition
is such that, although the gas is perfectly condensed to 25’ C. a t the plant, it will give rise to condensate and gummy deposits in meters that are substantially cooler (5” C.) than the
IlVD USTRIAL AND ENGINEERING CHEMISTRY
170
Table I-Details Plant I
Cracking temperature+ carburetor b o t t o d O
c.
(a ) 750 to 770
I1 I11 IV
(see text)
V
Night 700 to 760 ( b ) 690
Gemperature control Excellent Excellent Verygood Poor Fair Poor Good Good
Vol. 17, No. 2
of Carburetor Operation at Five Plants OPERATING CONDITIONS CHECKER-BRXCK -Rate of oil inputOil qualit; INCFIES SPACIN~ (c) (d) O B6. Carburetor Superheater 2 to 3 0 . 9 t o 1.0 0.4 28 to 29 3 (4 O . 9 t o 1.0 0.4 28 3 4 1.0 0.50 32 to 36 2.5 1.0 to 1.85 0.7 to 1.3 32 to 36 2.5 1.4 0.70 32 to 36 3 2.5 2.4 to 3.1 1.1 to 1.5 26 5 5 2.7 1.0 28.5 3 to 5 3
if3
2.7
1.0
28.5
3 to 5
GENERATOR
FUEL
Coke Coke Coke Coke Bituminous coal Coke Coke
3
Coke
(a) Data secured during 1923.
h
b ) Data secured during 1922. G allons per minute divided by diameter of the set in feet, ( d ) Gallons per minute divided by the radius squared. (e) Top courses spaced at 6 inches.
Table 11-Data
on LWht Oil Recovery
I I1 I11
2.95 533 100 450.0 37.0 3 12.2 16.38 384 2.34 0.2235 0.04 3.12 570 108 431.2 239.5 14 1.8 17.17 000 2.90 0.3065 0.03 3.06 528 95 477.0 30.0 3 16.9 18.25 432.5 2.37 0.2396 ... IV 3.12 530 95 463.0 28.6 4 16.2 16.45 407 2.37 0.2323 0.07 V 2.96 525 97 462.5 35.5 4 13.0 18.85 444 2.36 0.2536 High This term is used here as it is used by practical operators, and means the B. t . u. furnished p e r foot to 1000 cubic feet of finished gas by 1 gallon of oil calculated on the basis of 70 cubic feet of oil gas per gallon of oil and 300 B. t . u. per cubic feet for the water gas.
gas. On the other hand, gas with a favorable light oil composition can be relatively poorly condensed and yet never give rise to deposits and condensate in meters. The factors in carburetor operation are manifold, but the three most important in this study are (1) temperature, (2) rate of oil input, which determines the length of the cracking period, and (3) contact dependent upon the checker-brick spacing and interdependent with both (1) and (2). The information collected for the five plants studied has been reduced so far as possible to numerical form, and for easy comparison is given in Table I, together with information as to the quality of oil, generator fuel, and control on operation, particularly temperature, which is most subject to variations. I n the case of Plant I11 the variation of temperature is perhaps greater than is normal a t that plant, since a change of sets took place during the test. However, the conditions which obtained during the period of study showed temperatures of 640" to 740" C. a t the superheater top (700" C. was desired) a t different periods of the day and night. Carburetor-bottom temperatures varied from 650° to 790" C. During a previous period (not long before) of 190 hours a recording pyrometer showed an average carburetor-bottom tempera-
fortunately, the pyrometers had been found faulty and had been removed for reconditioning.
Results of Light Oil Study The facts and data pertinent to a consideration of the recovery of light oil from the gases from the five plants are given in Table 11. Composition (Volume) of Light Oil
I
The light oils collected a t the five plants were isolated by steam distillation from the wash oils containing them, the vapors being subjected to anespecially thorough condensation. The analyses of light oils were based on the light oil as previously defined, and on empirical fractionations and determinations of constituents previously described.3d The results of the analyses are strictly comparative and relative, and are presented in the various tables and graphic representations which follow. Table I11 shows the composition of distribution in per cent of the light oils, among the five arbitrarily chosen fractions.ad These are shown graphically in Figures 2A and 2B, the latter making comparison easy. The same data are shown in different form in Table IV, which has been prepared from Figure 3 for purposes of discussion. For the same reason Table V and Figure 4 have been
200
180
y '€4
d
$ L E
140
120
1W
Figure 3-Light
Oil Composition
ture of 710' C. (extremes 650" t o 760" C.) for the beginning of the cracking period. The average range was 680" to 710" C. I n Plant IV, during 1923, the temperature sought was,760' C., but was judged to be somewhat under that figure. Un-
HT OIL. PER CENT Figure &Light Oil Composition Experimental Apparatus. Cracking Temperature:, 7 0 4 O to 78S0 C.
introduced, being taken from a previous paper.sd Curve VI, which appears on Figures 5 and 6, is for a typical coal-gas light oil from coke ovens, based on data secured by the writers, and is strictly comparable. Curve VI1 is based on data for coal-gas light oil from a different plant.
Table 111-Percentage Composition of Light Oil FRACTION, O C .------D E B C 140 to 160 to 120 to 100 to 160 200 Total 140 120 Plant 100 2.2 100.0 1.4 I 62.4 9.4 24.6 100.0 4.7 2.4 5.4 I1 66.7 20.8 5.9 4.6 15.8 100.0 27.0 I11 46.7 4.9 7.4 3.1 100.0 19.7 IV 64.9 11.0 8.6 13.5 100.0 22.6 v 44.3 6.3 5.1 5.4 14.4 VI 68.8 5.0 3.9 5.3 10.9 VI1 74.9 A 75 to
Plant I I1 I11 IV V
Table IV-Percentage Composition of Light Oil ITEMPERATURE, O C. 160 120 140 100 96.4 98.6 87.0 62.4 87.5 92.9 97.6 66.7 89.5 95.4 73.7 46.7 87.7 92.6 84.6 64.9 89.0 66.9 80.4 44.3
Table V-Percentage
704 732 760 788
Discussion of Results General Correlation C + D 11.6 10.1 21.7 8.0
22.1 11.7 9.2
200 100.0
The physical facts and data have all been recorded. Bearing in mind that Plants I and I1 have no gummy meter deposits, Plant I11 has deposits, Plant IV with bad meter and main conditions 18 months previous is considered to be rapidly improving, and Plant V, although its meter deposits a t present are slight, has large amounts of gummy material and condensate in its mains, let us first examine the operating conditions of these plants and then the composition of the light oils.
100.0 100.0 100.0 100.0
Composition of Light Oil in Gas from Experimental Apparatus
Cracking temperature
c.
171
IIVD USTRIAL Ail-D ENGINEERING CHEMISTRY
February, 1925
--
TEMPBRATORB, ' C. 140 61.8 75.3 74.4 80.9 82.1 87.7 81.4 87.7
120
100 35.4 47.2 62.4 62.4
160 87.7 91.5 94.1 94.8
200 100 100 100 100
Composition of Light Oil-Unsaturation-Gum-Forming Constituents Gummy deposits are due fundamentally to the unsaturated hydrocarbons in the light oils, and particularly those that collect as or in the condensate of meters and mains. Indene and styrene, and to a lesser degree those of lower boiling point, have been found to be the important hydrocarbons. Table VI gives the unsaturation figures expressed in percentages of total light oil for all the fractions and for their totals. The final column gives the percentage concentration of unsaturation, largely indene, in Fraction E. The samples are those of the five plants under study, and similar data for the two coke-oven light oils, VI and VII. Table VI-Percentage
Unsaturation of Light-Oil Composition
---
TTnsit.
_ . I _
uration of E D E Per Plant A B C (Styrene) (Indene) cent I 1.00 0.97 1.73 0.77 0.72 51.4 I1 6.50 2.70 2.00 2.05 0.75 31.2 I11 5.40 2.50 1.58 2.52 1.97 42.8 IV 0.76 2.06 0.55 1.62 2.46a 33.3O V 8.86 3.00 3.80 3.40 4.30 39.1 VI 8.8 0.76 0.65 1.75 2.30 2.67a 52.0a VI1 5.8 ... ... ... 0.94 1.74 35.0a a Phenols were removed before the unsaturation figures were obtained. Total unsaturated 5.2 14.0 14.0 7.5 23.4
FRACTION----
. Figure 5-Coal-Gas
.
Light Oil
Certain of the factors listed can be eliminated almost entirely. In all these plants the quality of the gas oils used is comparable, and may be left out of consideration. (See Figure 7 and Reference 3d) The checker-brick spacing and contact surface area in all plants a t the time of the tests in 1923 were approximately equal, and the spacing was close. The generator fuel has been included, because bituminous coal was being used a t Plant IV. Arrangements for the study of this plant were made before this fact was learned. The very apparent effect on the light oil composition will be given later. I n Plants I and I1 the average cracking temperatures were 760" and 730" C., and were rigorously maintained close to those figures. There were no lapses. Relatively, the rates of oil input were low. The cracking efficiency (Table 11) in both plants was high and the drip oil produced was low.
Vapor Pressures For purposes of subsequent discussion, the vapor pressures of the hydrocarbons of the aromatic series a t 5" C. as the average minimum (winter) temperatures of distributed gas and a t 25" C. as the average maximum (summer) main temperature are shown in Figures 8 and 9, respectively. I n Table VI1 are given the relative partial pressures expressed as percentage of saturation (vapor pressure) for benzene, toluene, a hypothetical solvent naphtha representing the combined hydrocarbons other than those of benzene snd toluene, and lastly indene. The first three columns of values are strictly relative and comparable; and the values of Column 4, indene, are more nearly absolute as well as relative. They are somewhat low however because there is always some loss due to resinification during fractionation. I n Column 5 are given the sums of the values of Fractions C and D in Table 111,and they express relative volumes of these two fractions in per cent of total light oil. These fractions enter largely in the collection of condensate in mains.
0
Figure 6-Light
Oil Composition, Unsaturation
With these conditions obtaining, and with mains and meters free from deposits, it is well to study the compositions of the light oil in the gas from these two plants. I n general, the light oils contain high percentages of benzene and toluene, the more volatile constituents, and little of Fractions D and
INDUSTRIAL AND ENGINEERING CHEMISTRY
172
E, the less volatile, which are rich in gum-forming constituents. (Table 111, Figures 2A and B, and 3) The relative volatilities of the constituents of these fractions are shown in Figures 8 and 9.
Figure 7-Light
Oil Unsaturation
On the other hand, in Plants I11 and V the cracking temperatures were low, a t least low for long periods (Table I), and the rate of oil input was relatively high, which means a shortened period of cracking. The control of temperature and oil rate in Plant I11 was markedly variable. The cracking efficiency relative to I and I1 was low (Table 11). The amount of drip oil from Plant V is known to be large. The light oil in the gas from Plants I11 and V is relatively low in benzene and high in Fractions C, D,and E, constituting the high-boiling or low-volatile portions and containing the gumforming constituents. (Tables 111, IV, Figures 2A and B, 3,8, and 9) Plant IV is a special case, because of the use of coal as generator fuel and the changes that have been made in operation. The increase in temperature, the decrease in rate of oil input, and decrease in checker-brick spacing are all in line with the findings and recommendation as given in earlier
0
Figure 5-Vapor
5
Vol. 17, No. 2
I and I1 indicate about 65 per cent of the light oil from those plants to have distilled a t 100' C., whereas in the case of Plants I11 and V only about 45 per cent has distilled; a t 120" C., 87.5 per cent in case of Plants I and 11,but only 67 per cent of V and 74 per cent of I11 had been collected. Similarly, a t 140" and 160" C. the remaining portions of the light oil for Plants I and I1 are very small, whereas for V the value is much larger, but for I11 the value is intermediate. Curves I and I1 are typical for the light oil from a plant with good cracking. Curve V is considered as representative from a plant with inadequate cracking. The divergence of Curve I11 from V is considered to be due to a longer period during which a higher degree of cracking was being secured. The major variable factors in the cracking in these cases are temperature and period. The curves of Figure 3, when compared with those of Figure 4,in which temperature was the sole variable, show most strikingly in their close similarity the effect of temperature, not only on the general light oil composition, but on the production of gum-forming and other high-boiling materials which give rise to condensate. This is shown also in Figure 2B, Fractions D and E, and in Figure 6, Parts 1 and 2. The other factor, the rate of oil input, which bears a reciprocal relationship to the length of the cracking period, is likewise important. Compare in the two lines given below the values and sequence of the curves a t 160" C. in Figure 3 or of the graphic representation of Fraction E of Figure 2B, which measure the portion of the light oils destined largely to be deposited as drip oils, with the values for the rate of oil input in Table I. PLANT I Volume, Fraction E (% of light oil) 1 . 4 Rate of oil input, Table I(d) 0.4
I1 2.4 0.5
I11 4.6 0 . 7 t o 1.3
IV
v
7.4 0.7
11.0 1.0
The magnitudes are seen to vary similarly. While temperature and period (rate of oil input) as factors are interdependent and overlapping, the effect of the period is a definite
Vapor pressure, mm. of mercury [Woringer] 10 15 20 25 30
35
Pressures of Aromatic Hydrocarbons a t Average Minimum Main Temperature
papers, and the increasing improvement in meter and main conditions is in evidence. With these general relationships in mind it is proposed to examine more closely the results of the light oil study. Volume Composition
The volume relationships of the light oil of the gases are very evident in Figure 3, where the difference in composition of the light oils from Plants I and I1from Vis notable. Curves
one.6 The work of Downing and Pohlmann' provides the following data as evidence secured under carbureted watergas-making conditions, which show most clearly the increased cracking with increase of "period :" 6
7
(a) Alexander, THISJOURNAG, 7, 484 (19x5). ( b ) Egloff andTwomey, Met. Chem. Eng., 15, 245 (1916). ( c ) Table I, work cited under 3'. Proc. Am. Gas Inst., 11, 688 (1916); A m Gas Light
(1916).
J.,
104, 682
INDUSTRIAL AND ENGINEERING CHEMISTRY
February, 1925 Cracking period, in seconds Illuminants Hydrogen Methane
Cubic feet per gallon of oil 2.6 5.2 10.6 31.6 28.2 26.5 7.8 8.4 10.1 28.5 81.7 38.4
I n Figure 3 it may be seen that Curve IV, representing the light oil composition of Plant IV, takes a somewhat different form from that shown in the other four curves. I n Figure 5, Curves I and V show the extremes in composition for carbureted water-gas light oil, and VI and VI1 are curves of two samples of coal-gas light oil. It is to be noted that Curve IV takes the general form of VI and VII. The form of IV is believed to be due in part to the use of coal as generator fuel and in part to variations in the cracking operation, which could not be followed closely during the actual test a t this plant.
173
terials and gummy deposits in meters have been set forth by the use of data from actual operating plants, and this has completely confirmed the work on a laboratory ~ c a l e 3and ~ other previous reports. It is now proposed to examine briefly the reasons that cause the different light oil compositions to give rise to different meter conditions. The formation of gums and resins from meter condensates has been explained3" and the present concern is the collection of meter condensate and drip oil in general. Condensation from gas in main or meter depends primarily on temperature and composition (in its light oil content) of the gas and on the vapor pressure of these actual oil constituents. First in importance is composition. Carbureted water-gas light oil, like coal-gas light oil, consists mostly of
Vapor pressure, mm. of mercury
Figure 9-Vapor
Pressure of Aromatic Hydrocarbons at Maximum Main Temperature
Unsaturation benzene and smaller proportions of toluene and xylenes. The bases of gummy deposits are the gum-forming constit- The remainder is likewise aromatic in character, and made up uents in the light oil. The amounts of these are given in of the hydrocarbons of that series. Taking a list of such Table VI and Figure 6 . In Part 3 of Figure 6 the total compounds whose vapor pressures are known, in Figures 8 unsaturation is highest for Plant V, in which the cracking and 9 are represented their relative vapor pressures a t 5" C. temperature was low. For I1 and I11 the values are much (winter main temperature) and a t 25" C. (summer main lower, I1 being a case of a well-cracked oil a t a rigidly main- temperature). The advantage of securing a high percentage tained temperature of 730" C . , and I11 one whose cracking of benzene a t the expense of the other homologs, including conditions varied greatly but whose temperature was low indene, a t once becomes apparent (from Figure 8). On the much of the time. I is the product of high cracking tem- basis of the proximate analysis (Table 111) and the vapor perature and compares with similar values for coal-gas pressures, Table VI1 has been compiled. light oil-VI, VII-and with that for IV. These variations Table VII-Light Oil Composition. Comparative Partial Pressures i n Per cent of Saturation with temperature ("period" included) are in agreement with hydrocarAll oth& previous work by the writers (Figure 7) and others,* since bons calcd. Volume under the water-gas-making conditions the total unsaturation as metaxyIndene Per cent Plant Benzene Toluene lene as detd. C +D figures decrease with increase of temperature. (1) (3) (4) (5) However, the more important unsaturation figures of (2) A t S o C. Fractions E for several plants vary widely (Figure 6 , Part 1). These values measure the principal gum-forming constituent, indene, and the progression indicates clearly the quality or completeness of the cracking. Its relation to cracking temperatures and rates of oil input is equally clear as in Figure 7, taken from the writers' earlier work on a laboratory scale. Al 25' C. The same general relation is expressed in Part 2, Figure 6, for styrene, another important gum-forming constituent, in that increase of cracking temperature decreases the amount of that hydrocarbon produced. All clearly point to the advanPlants I and I1 possess light oil of such composition that tages and the necessity of thorough cracking for obtaining low concentrations of gum-forming constituents and con- indene and most other constituents are not near the saturation point, whereas Plants I11 and IV show values for indene sequently freedom from gummy deposits. much higher and the gas from Plant V is more than saturated Vapor Pressure Consideration with indene, which would have begun to condense a t 5" C., Up to this point the relationships between operation, even if no other condensate were present. In all distributing light oil composition, and the presence of gum-forming ma- systems, no matter how good the condensation is a t the plant, * Egloff and Twomey, Met. Chem. Eng.,.ll, 247 (1916). there are small amounts of easily condensable oils and naph-
INDUSTRIAL A N D ENGINEERING CHEMISTRY
174
thalene which initiate the collection of drip oil. Much collecting of drip oils follows through the solution effect. This requires that each light oil constituent go into solution into the drip oil coming down until the partial pressure of each constituent in the gas is in equilibrium with the partial pres200 180
b' 160 w
a
2 140 5 E 120 IO 80
PER CENT
Figure 10-Boiling
Curves of Light and Drip Oils
Vol. 17, No. 2
very similar to those obtaining for coal gas, as indicated under Plants VI and VII. Here, as in Plants I and 11, the coming down of the drip oil is largely completed within a much shorter distance of the plant, and the meters and outer distributing system remain free from condensate and gummy deposits resulting therefrom. Although the exact composition of the high-boiling portion of the light oil is not known, we do have a definite minimum value for indene, which constitutes about one-half of it. With that as an example, it is apparent from Figures 8 and 9 that a gas cooled and condensed perfectly a t 25" C. could carry, in the absence of other condensate, several times as much indene as the gas from Plant V-the richest in indene content within the writers' experience-does carry. The major part of this would condense as the gas cooled during distribution and collect in meters and mains. Other highboiling constituents of the gas play the same role as indene in the condensation process, and the combination is likewise sensitive to temperature change. On the other hand, the light oil composition of Plants I and I1 is such that the drip oil is small in amount and its condensation is quickly effected. In short, operation in the carbureting units such as will give thorough cracking gives a lower concentration of the undesirable high-boiling constituents than can be secured by the usual condensation process in the plant. This does not mean that optimum condensation in the plant should not be maintained; it is quite essential and desirable that it should. The general relationship between a light oil and its condensate is illustrated by the boiling curves of these two oils for both Plant I1 and Plant V, as shown in Figure 10.
sure of that constituent in the drip oil. I n Plants I and I1 there is relatively little indene (16 and 22.6 per cent, respectively) to come down, as shown in Table VII. This is likewise true of other hydrocarbons of low vapor pressure, as evidenced by the values of Column 3, and more especially in Column 5, which expresses the volume of the drip oil boiling between 120" and 160" C. Contrast these figures with those for Plants I11and V, in which the indene is two to six times as great and the volume of the material forming the rest of the drip oil is twice as great as in the case of Plants I and 11. It is a t once apparent why there is little drip oil in the distributing Acknowledgment systems of Plants I and I1 and no meter troubles, and on the other hand, much in Plants I11 and V. The writers gratefully acknowledge their indebtedness to I n the case of Plant IV although the indene fiaturation is J. F. Byrne for experimental aid. They are also greatly moderately high, Columns 4 and 6 show low values, and indebted to the officials and operators of the plants studied indicate little drip oil, since the total solution effect must be for the aid, facilities, and information given during this smaller than in the other four plants. These conditions are investigation.
Estimation of Acetone in Presence of Alcohol b y a Vapor Pressure Method' By E. A. Vuilleumler DICKINSON COLLEGE, CARLIS~E, PA.
NUMBER of samples of liquids recently submitted for A analysis in connection with their alleged use for beverage purposes were found to give a positive test for acetone, and
it seemed desirable to devise a convenient method for estimating the percentage of this denaturant. The difference in vapor pressure between an aqueous alcohol and acetone-containing alcohol of the same specific gravity is very appreciable. Inasmuch as the difference increases to a marked degree with a rise in temperature, and in order t o eliminate the temperature factor, it is advisable to compare the unknown with a solution of known acetone content. Of the four formulas to which manufacturers qualified to withdraw specially denatured alcohol for the preparation of compound rubbing alcohols are limited, the one containing acetone is made up on the basis of 10 gallons of acetone U. S. P. to 100 gallons of pure alcohol, and is designated Formula 23-A. The standard for comparison may ordinarily be prepared by diluting a sample of this composition to the same specific gravity as the unknown sample. I
Received October 28. 1924.
Ordinary alcohol and Formula 23-A have practically the same specific gravity. The following measurements may serve to illustrate tee difference in vapor pressure between the two liquids : A glass tube 800 mm. long, sealed a t one end, was filled with mercury and inverted in a dish of mercury. The level of the mercury fell to the barometric height. A similar tube was filled to within about 5 mm. of its length with mercury, filled to the top with 95 per cent alcohol, and inverted in the same dish of mercury. The level of the mercury in this tube was found to be 47 mm. lower than that in the first tube. I n the same way a sample of 23-A was introduced into a third tube. The level of the mercury in this tube was found to be 67 mm. below the level of that in the first tube, thus showing a difference of 20 mm. in the vapor pressures of the two liquids. The temperature was 22' C .
It is believed that no use has heretofore been made of vapor pressure in the analysis of mixtures of volatile liquids. Because the equipment required is simple and the results can be obtained by direct comparison with known mixtures, the method seems to have cortsiderable possibilities.