INDUSTRIAL AND ENGINEERING CHEMISTRY
November 1948 TABLEVII.
cut Isohexane Isoheptane Iso-octane I8on on a n e Isodecane
OCTANENUMBERS OF Isocu’rs BARBARA NAPHTHA F-3 O.N. (4 ml. TEL/gal.) Detd. Calod. 98.5 91.3 91.2 90.9 87.3 95.3 92.5. 81.5 84.7
...
F-2 O.N. (Clear) Detd. Calad. 75.8 75.7 75.3 70.7 76.6 77.2 78.5 71.0 62.1 61.1
FROM
SANTA
Deviation F-3 F-2 ... -0.1 -0.1 -4.6 -3.6 CO.6 -2.8 -7.5 4-3.2 -1.0
each isocut as listed in Table VI. These percentages were used with the blending octane numbers given in Table IV to check on the reliability of the blending octane numbers. These calculated values are compared with the determined values in Table VII. The deviation of the calculated and the determined octane numbers of cuts from this crude are much higher than for similar cuts from three other naphthas for which data are available (6). It is believed that the higher deviations may result from the unusually high aromatic content of this naphtha. The distillation curve and the octane numbers of each fraction plotted against volume per cent are shown in Figure 1. The bar graph of fraction composition a t the bottom of this figure shows clearly the high concentration of aromatics in the naphtha. The heavy line shows the F-3 octane numbers of the fraction with 4 ml. of tetraethyllead per gallon. The dotted lines show F-3 octane numbers obtained for the fraction by blending the fractions of low octane number with a high rating stock to bring them within the satisfactory rating level for the F-3 method. The deviations in the spread between the F-2 (clear) and F-3 (+4 ml. of tetraethyllead per gallon) are probably due to a differ-
2169
ence in lead susceptibility a t low and high rating levels rather than to any change in composition of the fractions. The ranges covered by the several isocuts are marked off for convenience in comparing the data with those presented in the tables. For example, the high concentrations of aromatics and naphthenes in the isononane cut are veryevident when the bar graph is examined. ACKNOWLEDGMENTS
This study required the services of many members of the chemistry and refining section of the Petroleum Experiment Station; their helpful cooperation made this work possible, Grateful acknowledgment is extended to the Sinclair Refining Co. and especially to G. H. Taber, Jr., of that company for supplying the naphtha. LITERATURE ClTED
(1) American Petroleum Institute, Research Project 45, 8th Annual
Report, July 1, 1945,to June 30,1946. (2) Aviation Gasoline Advisory Committee, Subcommittee on Blending Octane Numbers, Rept. 5, June 1, 1944. (3) Forziati, A. F., Willingham, C. B., Mair, B. J., and Rossini F. D., J. Research Natl. Bur. Standards, 32, 11 (1944): Research Paper RP 1571.
Gooding, R. M., Adams, N. G., and Rall, H. T., IND.ENO. CHEM.,ANAL.ED.,18, 2-13 (1946). (5) Thorne, H.M., Murphy, Walter, and Ball, J. S., Zbid., 17, 481-6 (1945). ( 6 ) Ward, C. C., Gooding, R. M., and Eccleston, B.H., IND.ENO. CHEM.,39, 105-9 (1947). (4)
RECEIVED July 1, 1947. Presented before the Petroleum Chemistry Group a t the Midwest Regional Meeting, Kansas City, Mo., June 23 to 25, 1947. Published by permission of the director, Bureau of Mines, U. 8.Department of the Interior.
Extraction of Polar Constituents from Hydrocarbon Solutions FIELD APPLICATION IN NATURAL GAS CONDENSATE WELL§ D. A. SHOCK AND NORMAN HACKERMAN University of Texas, Austin, Tex. Removal of acidic organic constituents from the liquid hydrocarbon phase of gas condensate wells by alkali extraction was too inefficient when done with inexpensive equipment. A n attempt was made to remove the dissolved substances by adsorption. The adsorption and recovery of compounds typically found in the system under investigation were studied in the laboratory, using commercial silica gel as the adsorbent. The adsorbates chosen were acetic acid, naphthenic acid, and resorcinol. Laboratory results showed that a sufficient quantity of each could be retained on the gel and then removed by proper treatment to make this procedure feasible as a field method. A simple apparatus is described for field experiments. Extraction by this method is effective, provided the rate of flow is low and the course of the extraction is followed so that fresh adsorbent can be provided periodically.
D
URING an investigation of corrosion in high pressure con-
densafe natural gas wells (4, it became apparent that that phenomenon is related, in part at least, to the presence of polar organic compounds in the liquid hydrocarbon phase. Inasmuch as the presence of fatty acids in the liquid water phase has been demonstrated (8), and it has been shown that there is a measurable concentration of the acids, at least up through heptanoic (6), it is not unreasonable t o expect to find polar organic compounds in the liquid hydrocarbon phase. Other evidence (10) indicated that these polar compounds included not only some straight chain carboxylic acids, but other types also. A project was set up to identify these compounds; this in turn required a method for concentrating the substances in question. The first method tried was extraction of the liquid hydrocarbon phase with aqueous sodium hydroxide. While it seemed plausible that a system could be designed capable of effecting efficient extraction, it was evident that such equipment would be too complex and costly for t h e purpose a t hand. Therefore, a relatively simple apparatus, consisting of a conical mixing tank,
2170
INDUSTRIAL AND ENGINEERING CHEMISTRY 60
LITERS, 5 P.P.M. C O N C E N T R A T I O N 80 1 PO 160 zoo
240
1.50
1.40
i
-
Id
1.30
8
2
2
z
1.20
z
4 8 12 16 LITERS, 100 P.P.M. C O N C E N T R A T I O N
100
Figure 1. Laboratory Adsorption Curves Intercept on ordinate gives blank acid value for isopropanol A . 100 p.p.m. acetic acid B . 100 p.p.m. naphthenic acid C. 100 p.p.m. naphthenic acid o n reactivated gel D . 5 p.p.m. naphthcnic acid E . 100 p.p.m. resorcinol
in which the sodium hydroxide solution \%as mixed with the hydrocarbon by a high speed centrifugal pump, was used. This procedure was less effective than desired. It was desirable to keep the volume of sodium hydroxide solution low and, in doing BO, the over-all recovery was inefficient; emulsions formed which, although unstable, required enough time in breaking to make the period of time of operation too lengthy; and the apparatus required continuous attention by the operator. Concentration of the polar organic molecules by adsorption on a solid appeared possible. The literature offers considerable evidence of adsorption of fatty acids (1,S, Q),naphthenes, aiid aromatics ( 7 ) on solids. It was evident that the adsorbent should be polar in character and it was thought that neither the relatively small quantities of inorganic substances present-e g., water, carbon dioxide-nor the large amounts of hydrocarbon would interfere to too great an extent. Silica gel was chosen for the sorbent to be used in this work by virtue of its general effectiveness as an adsorbent for polar compounds. Furthermore, it is readily available on the commercial market in sufficient quantity for field use and i reasonably uniform in quality. LABORATORY I h Y ESTIGATIONS
Three materials were chosen as representative types of compounds that occur in the liquid hydrocarbon phase of condensate gas wells. The lower fatty acids, through heptanoic, have been proved to be present in the water phase (6). It is reasonable to assume that they will also be present to some extent in the liquid hydrocarbon. Of the total fatty acid content in the mater phase of most of t h v e wrlls, roughly 60% is known to be acetic acid The remainder is made up of decreasing percentages of
Vol. 40, No. 11
the higher acids through heptanoic. No evidence of the presence of formic acid has been found. It was decided, therefore, to use acetic acid as one of the adsorbates. As presence of phenolic compounds has been reported by Griffin and Greco ( 2 ) and confirmed by Lochte (6),it was deemed advisable to include a phenolic compound in the adsorbates to be studied. On the basis of recent investigation by the authors (IO),a commercial naphthenic acid was also included. A number of solvents were used i n attempts to elute the adsorbed material from the gel. Among these were carbon tetrachloride, benzene, amyl alcohol, isopropyl alcohol, and inethyl ethyl ket,one. METHODOF EXTRACTION.The method consisted in packing 300 grams of silica gel into a glass tube approximately 60 em. long and 5 cm. in diameter. The ends of the tube were packed with glass wool: on the top, to aid in providing a proper distribution of liquids through the gel, and at the bottom, to aid in providing for adequate drainage. A stopcock at the bott’om of the tube regulated the flow, which was maintained a t approximately 300 ml. per minute. ANALYTIC~L METHOD. Concentration changes were followed by titration of the effluent solution for acidity. A 100-ml. sample of the solution was mixed with 100 ml. of isopropyl alcohol and 2 drops of phenolphthalein. This solution was titrated to EL pink end point with 0.02 N sodium hydroxide. I n general, a 5- or 10ml. buret graduated to 0.05 or 0.10 ml. was used. For very low acid concentrations it was found desirable t o use a 5-ml. buret graduated to 0.01 ml. The isopropyl alcohol exhibits an appreciable acidity as coinpared to the solutions used in this work, and it was necessary t,o be sure that its acid value mas not so large as to obficure the results. Accordingly, a blank was run on each batch of the alcohol supplied by adding 2 drops of phenolphthalein to 100 ml. of the alcohol and tit’ratingto a permanent pink with the 0.02 N sodium hydroxide solution. Only those samples of the alcohol were used which required approximately 1 ml. of the standard base per 100 ml. of the alcohol. PROCEDURE.. The saniple of distillate was blown out with dry air until the carbon dioxide was removed. I n a quart sample with air bubbling rapidly through the solution, t,his generally took about 15 minut,es. Removal was considercd complete when the solution exhibited a constant minimum acid value. The method used for elution consisted simply of shaking the saturated gel wit’h one of the solvents listed above. The 300 grams of gel were shaken with 200-ml. batches of fresh solvent until the amount extracted (per batch) mas negligible as determined by the tit,ration method. Generally ten washes or approximately 2000 ml. of the solvent were required.
RESULTS.Samples were taken a t regular intervals in terms of liters of solution put through the gel. Results are shown in Figure 1 and Table I. Acid values of the isopropyl alcohol were 0.78 for curves A , B , C, and E and 1.28 for curve D. A sharp change in the slope of the curve was taken to indicate that the gel was becoming saturated with the solute in question. As can be seen from Figure 1, A , the 300 grams of gel show no indication of approaching saturation even though 10 grams of acetic acid had already been adsorbed. Each of the other adsorbates showed definite saturation values when 0.7 to 1.0 gram of solute had been adsorbed. TABLEI. WEIGHTAD~ORBED ON 300 GRAMSOF SILICA GEL Hydrocarbon Solution, P.p.m. Weight Adsorbed, G r a m 10.0 Acetic acid 100 0.65Q Kaphthenio acid 100 0.85b Saphthenic acid 100 0.83 Naphthenic acid 5 0.85 Resorcinol 100 5 On pel as received b This gel had been saturated, extracted, and reactivated.
I t was known that the concentration of acids in the liquid hydrocarbon produced from many wells is low, and it was clear that an adsorption experiment had to be made using very dilute solutions. Figure 1, D , shows that the “saturation value” for the 300 grams of gel is not appreciably affected by the use of solutions as dilute as 5 p.p.m. of acids.
INDUSTRIAL AND ENGINEERING CHEMISTRY
November 1948
2111
TABLH 11. COMPARATIVE EXTRACTION EFFECTIVENESS (1st wash, gel saturated with naphthenic acid) Solvent % Extraated 10.3 9.2
0.3 2.6 0.6
TABLE 111. EXTRACTION RECOVERY Solvent Isopropyl alaohol Isopropyl alcohol Methyl ethyl ketone
Gel
% Recovery
Saturated with naphthenic acid Saturatcd with resorcinol Saturated wirii naphthenic acid
86 70 93
SE PA RAT 0 R c -
Figure 2. Silica Gel Extractor
The question might be raised as to the lesser effectiveness of the gel used for the experiment described by Figure 1, B . This gel had been standing in the laboratory for some time and its adsorptive capacity had decreased. The adsotbed naphthenic acids were removed from the gel with isopropyl alcohol and the gel was reactivated by heating in air at 150' C. The adsorption on the reactivated gel is shown in Figure 1, C. This clearly shows that it is possible t o re-use the gel.
The accumulated condensate flowed into a 4-inch pipe A , which was packed with fiber glass so that it filled the tube a6out one half full from the exit end. The emulsion broke in passing over the packing. The two phases were separated in a 3-liter glass separator, B. Water was drained from the bottom valve continuously and hydrocarbon was allowed to pass into the adsorbing unit, C, which contained the silica gel. I n order t o minimize the frequency of replacement of gel in the adsorption unit, the adsorption chamber was made sufficiently large for approximatelv 10 hours of continuous operation. Figure 3 shows one of the field units in operation. Two factors might have prevented efficient operation of the field extractor unit: water and carbon dioxide dissolved in the hydrocarbon. By titrating the amount of acid in the entering and exiting hydrocarbon, and knowing the total throughput, it was found that the field unit approximately duplicated laboratory experience-that is, the field extraction was 95% as eficient as the laboratory extraction in terms of acid adsorbed per unit weight of gel. From this fact it is apparent that the effect of water and carbon dioxide in these concentration ranges is not appreciable. The field adsorption results are shown in Figure 4. The percentage of acidic constituents removed by adsorption are plotted against gallons of hydrocarbon condensate put through for three different rates of flow. The ordinate values were calculated from the differences in acid titration (as described for the laboratory experiments) of aliquots of the liquid hydrocarbon before and after adsorption, It is clear that the largest percentage take-up is obtained in the initial period of extraction. Furthermore, as is to be expected, the gel was most completely
Figure 3. Field Extractor i n Use
Of the solvents tried, only isopropyl alcohol and methyl ethyl ketone proved effective in removing the adsorbed material from the gel. Table I1 gives the comparative effectiveness of the solvents used, and Table I11 gives the amount of adsorbate recovered for the two most efficient eluting agents. The methyl ethyl ketone was chosen as the most feasible for practical use, because it will not esterify with the adsorbed acids. Use of inorganic eluting agents, such as water or aqueous salt solutions, was not attempted in this work; however, it is evident that it should be investigated. FIELD EXTRACTOR
The laboratory work indicated that the method would be feasible for field use. As liquid water is produced along with the liquid hydrocarbon, it was necessary to separate these constituents before extraction. Furthermore, the two liquids generally appear in thc form of an cmulsion and, therefore, some mechanism for breaking the emulsion was required. The field apparatus was designed as shown diagrammatically in Figure 2.
J
0
100 BOO 300 400 GALLONS PUT THROUGH GEL
Figure 4.
Field Adsorption Curves
500
2172
.
INDUSTRIAL A N D E N G I N E E R I N G CHEMISTRY
saturated by a given volume of hydrocarbon flowing at the lowest rate. I n all cases which were run for sufficiently long periods of time, up to 20 hours, small amounts of acid continued to be adsorbed. The results show definitely that a large proportion of polar constituents may be removed from very dilute solution in liquid hydrocarbon by adsorption methods. Optimum conditions call for relatively short periods of exposure at slow rates of flow. As pointed out by Mair and Forziati ( 7 ) among others, olefins and aromatic hydrocarbons are also adsorbed on silica gel. This would account in part for the inability to remove the dissolved polar constituents completely. There is, in addition, the question of distribution of types of adsorbed materials a5 contrasted with the normal distribution in the liquid hydrocarbon. Thus, there remains work in determining the types of cornpounds adsorbed, the distribution of these types, the variation in distribution in the early stages of adsorption and the later periods, and whether or not those substances which are adsorbed initially are displaced by other materials as the flow continues. ACKNOWLEDGMENT
The authors wish to express their appreciation to the Corrosion
Research Projects Committee of the Natural Gasoline Asaociation of America, Tulsa, Okla., for its financial and technical support, in carrying on this work. LITERATURE CITED
Gapon, E. N., J . Phus. Chem. ( U . S . S . R ) ,11, 651 11938). Griffin, H. T., and Greco, E. C., Corrosion, 2, 138 (1946). Gyani, B. P., and Ganguly, P. B., J . Indian Chem. SOC.,20, 331
4ND
-
7 (1943).
Hackerman, Norman, and Shock, D. A., IND. ESG. CHEM.,39, 807 (1947).
Lochte, H. L., and Black, Harry, Na% Gasoline Sssoc. of America, Corrosion Research Projects, Committee Minutes, 1‘01. 3, p. 271 (Januaiy 1947).
Lochte, H. L., and Roberts, G. B., Ibid., Vol. 3 , p. 41 (June 1947).
Mair, B. J., and Forziati, A. F., J. Re,sear.ch iVat2. Bur. Standards, 32, 165-83 (1944). Menaul, P.L., Oil Gas J.,43, No. 27, 80-1 (1944). Rykhlikov, G. P., Chem. Zentr., 1940, 11, 3595. Shock, D. A,. and Hackerman, Norman, IND. Ex-c;.CHEM.,39, 1253 (1947). RECEIVED August 9, 1947. Presented before the Division of Industrial and Engineering Chemistry at the 112th Meeting of the AMERICAN CHEMICAL SOCIETY,Xerv York, S . Y .
dulus”
Some “Temperature-Yo Relations hi H. W. MOLL
Vol. 40, No. 11
W. J. LEFEVRE
The Dow Chemical C o m p a n y , Midland, M i c h . T h e variation of Young’s modulus with temperature for any given plastic material is useful in evaluating the suitability of that matei-ial for certain applications. A laboratory apparatus is described, which makes possible the accurate and efficient determination of the relationship between temperature and Young’s modulus for a wide variety of plastic materials and sample sizes. Data for a number of plastics have been compared with impact and service data and are reported in this paper.
ANY tests have been developed for measuring the effect of temperature on the physical properties of rubbers and plastics. Several of the better known tests are based on elongation and recovery t o original dimensions ( I ) , permanent set and vulcanization ( d ) , and dead load compression characteristics ( 2 ) of elastomers. For this reason the machines described for quch tests are not applicable t o a number of plastics. Stifler and Mitchell (16) reported in 1931 a method of measuring and interpreting the properties of rubber by measuring the coefficient of rigidity and Young’s modulus at various temperatures, but did not decribe their apparatus. Koch ( I S ) described an apparatus for measuring Young’s modulus of rubber samples at various temperatures. Kish (12) and Green, Cholar, and Wilson ( 1 1 ) have also described apparatus and published data on the variation of Young’s modulus with temperature obtained by beam-bending methods similar to Koch’s met,hod (15). A fairly complete discussion of the effect of low temperatures on Young’s modulus of elastomers is given by Liska (6,l.d). Forman (10) reports similar type data obtained by hardness tests on elastomers at low temperatures. Clash and Berg (8) discuss an apparatus and report results on vinyl polymers using a torsion method, and a beam-bending method for the modulus of rigidity of specimens at temperatures to -75” C. Carswell and Nason (7) discuss in a general manner the effects of temperature on plastics as tested by several methods, including measurements of modulus of elasticity. The last three papers mentioned con-
tain excellent bibliographies on the subject of the effects of temperature on the mechanical properties of plastics. Of the large number of tests and apparatus described in the literature, none seems to be so adaptable t o the wide selection of samples available in the average plastics research laboratory as the test and apparatus described in this paper. The present test permits a rapid and reliable measurement of the relationship of temperature to Young’s modulus, besides allowing the use of a large variety of sample sizes and shapes. METHOD OF MEASUREMENT
Young’s modulus is defined as the ratio of the longitudinal stress to the longitudinal strain and may be obtained conveniently by beam-bending methods as described in any standard physics text. A bar or rod of the material to be tested i s supported at each end and bent or deflected, by application of a load a t the middle at a specified temperature, a n amount that does not exceed the proportional limit. The formula for calculation of Young’s modulus, M , from such a test is given as:
where W is the load, g is the gravitational factor, L is the length of the specimen between the supports, d is the deflection, a is the specimen thickness, and b is the specimen width. If a round rod is used in place of a rectangular bar, 3 r r 4 is substituted for a3b, in which r is the radius of the rod. I n engineering practice it is customary t o express Young’s modulus in pounds per square inch, Wg being measured in pounds and the sample dimensions in inches. The effective specimen dimensions are obtained by measure