When the reaction again drops to 25” C., add the reaction liquor to 400 cc. of vigorously stirred petroleum ether. A heavy oil separates, which on further stirring hardens to a light cream-colored solid. Filter this solid in a dry box through a filtered glass funnel, to recover 42.7 grams of product. This product is further purified by redissolution in fresh dichloroethane and reprecipitation with fresh petroleum ether. The purified material is dried in the dry box and is extremely hygroscopic. I n fact, if a very well dried stream of nitrogen is not employed in the dry box, it turns into a puddle in a matter of hours. The product was analyzed for C, H, and S and checked almost exactly for caprimidyl sulfate. This solid has a melting point of 122” C. as compared with 88”C. for cyclohexanone oxime, and 67-C. for caprolactam. I t can be converted quantitatively to caprolactam by treatment with sodium hydroxide. To demonstrate that this solid was not caprolactam sulfate, the sulfate was prepared from known caprolactam. The physical properties and behavior of the two products were entirely different. If desired, 36.6 grams of triethyl phosphate (0.2 mole) can be used in place of the 17.6 grams of dioxane. This gives a liquid rather than a solid complex specie.
has been isolated as a new composition of matter. I t is a very hygroscopic but stable white solid, melting a t 122” C. Other ketoximes can be used but aldoximes undergo dehydration rather than rearrangement. Aromatic ketoximes can be rearranged by phosphate complexed SO3 without ring sulfonation. literature Cited
Blatt, A. H., Ed., “Organic Syntheses,” Coll. Vol. 2, p. 77, Wiley, New York, 1950. Donaruma, L. G., Heldt, W. Z . , “Organic Reactions,” Vol. 11, pp. 1-156, Wiley, New York, 1960. Gilbert, E . E., “Sulfonation and Related Reactions,” Wiley, New York, 1966. Turbak, A. F., IND. ENG. CHEM.PROD.RES. DEVELOP. 1, 275 (1962). Turbak, A. F., Livingston, J. R., Jr., IND. ENG. CHEM. PROD. RES. DEVELOP. 2, 229 (1963).
Conclusions
Caprolactam can now be prepared in good yields under mild and safe conditions a t low temperatures by use of phosphate or dioxane complexed SOa. Caprimidyl sulfate
RECEIVED for review August 7, 1967 ACCEPTED May 13, 1968 Division of Organic Chemistry, 153rd Meeting, ACS, Miami, Fla., April 1967.
RAPID AND COMPREHENSIVE CRUDE OIL EVALUATION WITHOUT DISTILLATION V .
FRANCES
GAYLOR
AND
CAROLINE
N .
JONES
The Standard Oil Company (Ohio), 4440 Warrensville Center R d . , Cleveland, Ohio 44128
A rapid method for evaluating chemical and physical properties of crude oil has been developed. Yield and quality of all fractions through asphalt are obtained in two to three hours. The results provide much, and often sufficient, information to judge the refining value of a crude; and analysis time and costs are considerably less than required for a conventional distillation assay of comparable scope. The method consists of six tests made on whole crude oil. The tests are gravity, per cent sulfur, nitrogen, pentane insolubles, polarographically measured condensed aromatics, and gas chromatography. Composition parameters obtained by applying these tests to 23 crude oils were correlated w i t h quality tests of the 10 fractions per crude obtained from a n 8-gallon batch still. Correlations ranged from acceptable to excellent and included such important quality parameters as gravity of all fractions, smoke point and freezing point of kerosine, w a x content and viscosity index of lubricating oils, U.O.P. ”K” factor and carbon residue of cracking stocks, and retained ductility and petrolene VGC of asphalt.
A COMPLETE crude oil assay should define both quantity and quality of all major fractions contributing to value of this raw material. Quantity of each fraction is conventionally determined by distillation of a sample of the crude. Quality is then determined by separate testing of each major fraction recovered. Time and costs thus multiply in direct proportion to the number of fractions examined during any type of distillation assay. Significant reductions in time and costs of evaluating crude oils probably can be achieved only by eliminating
the distillation step. Analytical tests applicable to whole crude oil, but which will adequately characterize all distillation fractions of the crude, are then needed. The present work used six such tests (Table I). Quantity and quality of all crude oil fractions, through asphalt, are calculated from information derived from these simple tests. The gas chromatographic analysis provides experimental boiling range distribution data up to 725” F. Yields of higher boiling fractions are predictable from knowledge of quantity of 700”F.+ bottoms, combined with VOL. 7 NO. 3 S E P T E M B E R 1 9 6 8
191
Table 1. Crude Oil Evaluation Tests
Test
Elapsed Time, Hours
Gas liquid chromatography Polarographic condensed aromatics Pentane insolubles Specific gravity c70 sulfur $C nitrogen
1-3 0.75 2 0.20 0.20 2-3
composition information provided by the rapid assay scheme. The gas chromatographic analysis also yields hydrocarbon-type composition analysis of fractions boiling t o 390" F. Composition parameters derived in this manner were empirically correlated with physical properties of fractions obtained in a conventional distillation assay. The multiple regression equations so derived thus allow direct computation of quality, in conventional terms, from rapid assay data. Experimental Rapid Assay Tests. The gas chromatographic analysis procedure has been previously described (Gaylor et al., 1964). Yields of fractions boiling to 725.F. were calculated from peak area measurements, based on n-paraffin calibrations. Hydrocarbon-type distribution analyses (POXA) of naphthas boiling to 390' F. were calculated by a peak height ratio (P.H.R.) method described in the above reference. The same peak height ratio method was applied to estimate n-paraffin contents of fractions boiling 390" to 725" F . Distribution of Cg through C1,, alkyl benzenes was also obtained by the peak height ratio method and required more than one experimental gas-liquid chromatographic (GLC) analysis for some crudes. Retention times of single aromatics varied with amounts injected, owing to adsorption effects. The required relative retention times were obtained by appropriate adjustments in sample sizes. Condensed aromatics in crude oil were measured polarographically, using a Sargent Model XXI recording polarograph. One milliliter of a benzene solution, containing 0.2 gram of crude oil in 10.0 ml. of benzene, was added to 9.0 ml. of freshly distilled methyl cellosolve (ethylene glycol monomethyl ether), 0.10M in tetra-n-butyl ammonium iodide. After de-aeration with oxygenfree nitrogen, the solution was polarographed from -1.0 to -2.5 volts with a dropping mercury indicator electrode us. a mercury pool reference. The equipment was calibrated with phenanthrene, recrystallized from ethanol, and crude oil results were calculated as micromoles of condensed aromatics per gram of sample. Pentane insolubles were determined by weighing the insoluble matter which separated from a solution of crude oil in n-pentane. The experimental procedure was identical to that described in ASTM D893 except that the procedure was applied to 4 to 5 grams of crude oil, sampled a t ambient temperature. Specific gravity was determined by hydrometer (ASTM D287); s u l h r content was measured by x-ray fluorescence; and nitrogen content was measured by a Kjeldahl procedure. Distillation Assay. Test crudes were distilled in an %gallon, batchtype laboratory still operated a t atmospheric pressure to an overhead temperature of 400" F. Reduced pressure distillation was continued in the same apparatus to an overhead temperature equivalent to about 800° F. a t atmospheric pressure. The bottoms were then transferred to a molecular still, and fractionation was continued until a target penetration for the asphalt bottoms was achieved or until two oil fractions boiling about 800" to 900°C. and 900" to 1000"F. were removed. In the latter case, the asphalt, if still soft, was oxidized to the target penetration; and quality testing was done on only the oxidized bottoms. Physical properties of all fractions were measured by standard petroleum testing techniques. Composition, in terms of hydrocarbon-type distribution, was determined only on naphtha fractions, boiling up to 400" F. Heavy Ends Distillation Yields Correlation. Distillation-type yields of crude oil fractions boiling up to 700°F. were determined experimentally by GLC analysis of unfractionated crudes. Methods 192
l&EC PRODUCT RESEARCH A N D DEVELOPMENT
for predicting yields of fractions boiling above 700° F. were based on composition tests and specific gravity calculations. Specific gravity of a hydrocarbon mixture reflects both boiling range distribution and hydrocarbon-type composition. Boiling range distribution can obviously be calculated if both gravity and composition are known. Specific gravity of the 700" F.+ bottoms can be computed. Yield of the 700'F.+ bottoms is known, by difference, from the GLC analysis. Specific gravity of each of the fractions boiling to 700" F. can be calculated from rapid assay data, as will be shown later. Specific gravity of the 700"F.+ bottoms can then be calculated by correcting the experimentally determined crude oil specific gravity for gravity contributions of the volatile fractions. Two of the rapid assay tests selectively estimate composition characteristics of the heavy ends of crude oil. The pentane insolubles test provides a measure of asphaltene content. This value may roughly suggest asphalt content. More important, the asphaltenes likely reflect level of lower molecular weight, highly condensed ring systems. Condensed aromatics are also measured by the polarographic test. However, the polarographic method is essentially insensitive to the highly condensed systems such as are found in asphaltenes. Rather, polarography is sensitive to the simpler substituted polynuclear aromatics-such as naphthalenes, anthracenes, phenanthrenes, etc. Thus the two tests together, pentane insolubles and polarography, provide an empirical means of characterizing distribution of condensed aromatics. Initial attempts to evaluate paraffinic characteristics of the heavy ends of crude oil used a quantitative urea adduction technique. We later found that n-paraffin contents estimated from the GLC analysis were equally effective and experimentally more repeatable. Yields Correlations. Five independent variables were correlated with heavy ends yields of 30 crudes, all of which had been assayed by the distillation technique. Specific gravity of the 700"F.+ bottoms, pentane insolubles, polarographically measured condensed aromatics, and a n-paraffin value from GLC analysis constituted four of the independent variables. The fifth was yield of a 700" to 725" F. fraction, measurable from the GLC analysis. This was a convenience variable. The distillation assay measured yields at exactly 700" F. Computations were slightly simplified by using the extra 25" F. yield from rapid assay as an independent variable. All but one of the independent variables were related to amount of 700" F.+ bottoms-Le., grams of asphaltenes per 100 grams of bottoms, etc. The n-paraffin peak height ratio was an estimate of n-paraffin concentration in the boiling range composed of chromatographic peaks eluting after n-Cli and through n-Clj. Higher carbon number peaks would be preferred. Limitations in data available, however, confined the choice to n-CLrand n-Cl; Experimental distillation yields of crude oil heavy ends were expressed as three dependent variables-a 700" to 800" F. fraction, an 800O to 900c F. fraction, and a 900" F.+ bottoms. Each of these fraction yields was expressed as volume per cent of 700" F.+ bottoms. Regression analysis of the eight variables on 30 crudes yielded reasonable multiple regression coefficients (Table 11). Correlation
Table II. Calculation Constants for Yield Prediction
Multiple Regression Coefficients Y, Y?
YI XI
X, Xi
X, X,
z Y
R
XI
3.188 6.580 -14.25 0.0341 0.000570 15.15 27.55 0.976
-1.244 43.76 -206.8 0.3341 0.0146 213.82 25.69 0.703
-1.943 -50.34 221.0 0.3682 0.0141 -128.97 46.75 0.895
x 6.95 0.517 0.9519 4.69 469.0
... ...
...
= 7 700"725".
X 2 = n-paraffin P.H.R. Xi = 700" + specific gravity.
X, X, Z
= pentane insolubles. = polarographic condensed aromatics.
= intercept.
Y = mean. R = correlation coefficient. Yi = % 700"-800" F Y2 = R 800"-900" F . Y3 = 7i 900°+ bottom (F.).
Figure 1. Yield correlations
10 Exptl
Vol.
% of
Crude
20
Exptl Vol.
coefficients indicated that 95'1 of the variation in 700" to 800" F. yields and 90:; of the variation in yields of 900"F.+ bottoms were explained by the five independent variables. Data fit was quite good over the range of about 5 to 40 vol. 'i, of crude (Figure 1, A). In a further extension of this work, expressions for calculating distillation end point and asphalt yield were also derived. Distillation end point was correlated directly with amount of pentane insolubles ( R = 0.864). This was a reasonable relationship since distillation termination was governed by penetration of the asphalt bottoms. Asphalt yields were correlated ( R = 0.977, Figure 1, A) with four independent variables which included yield of 900° F.+ bottoms, condensed aromatics from polarography, the distillation end point, and an aromatic condensation indicator derived from GLC analysis (defined in a later section). Yield Prediction Test. Primary yield prediction techniques were tested on 23 additional crudes distilled and assayed two to three years after derivation of the above expressions. Prediction accuracy was satisfactory for both the 700' to 800' F. and 900" F.+ bottoms fraction (Figure 1, B ) . Predicted yields of 700" to 800' F. fractions averaged within 0.55; of experimental yields. For the 900"F.+ bottoms fractions, predicted yields averaged within 1.6'; of experimental. Accuracy of predicting 8OOc to 900' F. fraction yields was poorer, with an average difference of 1 . 9 5 between experimental and predicted yields. Heavy Ends Composition Correlations. Composition correlations were aimed at deriving expressions for calculating hydrocarbontype composition of crude oil fractions boiling above 390° F. Crude oil composition prediction was recently considered by Martin et al. (1963b) who concluded that gas oil composition cannot be predicted from component analysis of naphthas. Their data showed that neither differential composition curves nor fraction composition curves were likely building blocks for composition prediction. Trends were evident on comparing crudes, but quantitative generalizations were not possible. The concept of cumulative fraction composition was derived in the present work. Application of this concept to their data indicated that hydrocarbon-type compositions of heavy ends are predictable. Cumulative fraction composition is illustrated best by visualizing a crude oil distillation with a single distillate receiver. Contents of the receiver are allowed to accumulate hut are sampled and analyzed for hydrocarbon-type distribution at the desired vapor temperatures-Le., 200" F., 400" F., etc. The Martin et al. data (1963b) show striking similarities between crudes on comparing cumulative fraction compositions as a function of distillation temperature (Table 111). Cumulative fraction per cent of paraffins decreased with increasing distillation temperature in every case, and cumulative fraction per cent of aromatics increased with increasing temperature in every case. Cumulative fraction composition seems a reasonable and orderly way of comparing crude oil composition trends as a function of molecular weight. This type of comparison integrates possible historical and probability effects. Interdependency between composition of light and heavy ends on an integral basis would be reasonable if crude oil is formed from a single source material; and molecular weight distribution of a single hydrocarbon type may be related, at least in part, to the number of possible single compounds. The Martin et al. data (1963b) were used to derive expressions for calculating cumulative fraction compositions to 800" F. These
30 % of
40
A. 6.
Data fit Prediction accuracy
@
% 700"-800"F.
X
O h
0
O h
A
% asphalt
800"-900"F. 900"+ bottoms
50
Crude
Table Ill. Comparison of Cumulative Fraction (C. F.) Composition (Martin et al. Data)
C.F. %i at Crude Volume % paraffins Kawkawlin Darius Lee Harrison Beaver Lodge Hendricks North Symer South Houston Volume %c aromatics Lee Harrison Beaver Lodge Darius South Houston North Symer Kawkawlin Hendricks
Net Change
600 F.
700 F.
800
F.
520 F.
72 66 53 52 53 50 29
69 65 49 49 46 46 14
66 64 48 48 42 46 11
60 60 46 47 39 45 9
56 57 45 45 36 44 9
-16 -9
24 14 13 9 8 6 6
28 16 17 24 12 10 12
31 17 18 29 14 12 16
34 19 21 32 15 15 18
36 19 23 33 17 18 21
+12 +5 +10 +24 +9 +12 +15
390
O F .
-8 -7 -17 -6 -20
experimental data were supplemented with crude oil composition analyses made by the U. S. Bureau of Mines (Holliman et al., 1950). Use of the latter data required a few additional analyses which were obtained by the rapid assay technique. Paraffin Correlations. Two different correlations were tested on the seven complete sets of data published by Martin et al.(1963b). Volume per cent of paraffins in cumulative fractions boiling up to 800c F. was strongly dependent on cumulative fraction per cent of paraffins a t 390O F. Correlation coefficients were 0.990, 0.982, 0.959, and 0.943 for end points of 520", 600", 700", and 800°F., respectively. Data fit was improved, however, by introducing a second independent variable. Rate of decrease in cumulative fraction per cent paraffins with increasing molecular weight appeared to he partly dependent on relative proportions of normal and isoparaffins. Addition of a paraffin distribution value, expressed PS ratio of isoheptanes to n-heptane, gave extremely good data fit (Figure 2, A). All correlation coefficients were in the 0.99 range (Table IV), indicating that 98 to 99% of variation in the experimental data was explained by the two independent variables. Paraffin correlations were tested on the Bureau of Mines crude oil data. Accuracy of the paraffin prediction equations was satisfactory (Figure 2, B). Predicted cumulative fraction per cent paraffins averaged within 3YC a t 600OF. The small bias could easily be due to differences in experimental analysis methods. Bureau of Mines paraffin analyses were calculated from physical properties of saturated fractions separated by liquid-solid chromatography. Martin et al. (196313) used mass spectrometry to determine paraffin concentrations. VOL. 7 NO. 3 S E P T E M B E R 1968
193
ExDtl %
Exptl %
Exptl %
Figure 2. Paraffins correlations A. B.
Figure 3. Aromatics correlations (data fit) A. B.
0
Martin ef ai. data (data fit) Bureau of Mines data (prediction accuracy) C.F. Yo a t 520" F. C.F. % a t 600°F. C.F. Yo a t 700"F.
0 x 0
h
C.F. Yo a t 800" F.
A
0 X
Aromatics Correlations. Expressions for predicting increase in aromatics with increasing molecular weight required four independent variables. Cumulative fraction per cent aromatics had only limited effectiveness in estimating cumulative rate of aromatics increase. This was not surprising, since alkyl benzenes comprise the bulk of total aromatics boiling below 390OF. Higher boiling fractions contain polynuclear aromatics, aromatics condensed with naphthenes, and sulfur-containing aromatic rings, as well as alkyl benzenes. At least two additional parameters were thus needed to predict aromatic distillation trends in the higher boiling fractions. Relative distribution of C, to CI,, aromatics provided a clue to the slope of the aromatics distillation curve. Total aromatics increased steeply with increasing molecular weight when the relative proportion,of C K benzenes was high. This effect was seemingly related to large amounts of alkyl benzenes in fractions boiling above 3 9 P F . C,,, benzene concentration, expressed as per cent of C, to C,,, aromatics, was therefore used as an alkyl benzene persistence indicator. An indicator of condensation tendencies was derived from the distribution of C,,, benzene isomers. A deficiency in disubstituted CI,, benzenes appeared to accompany an abundance of condensed rings in higher molecular weight fractions. Sulfur
Table IV. Composition Calculation Constants
Multiple Regression Coefiients for C.F. W at 520" F., YI
600" F., Y2
700" F., Yi
800" F.,
Y4
x
1.001 -2.140 -1.27 0.996 Aromatics* Xi 1.249 X1 0.331 X, -0.197 XI -0.303 Z 4.06 R 0.977
0.825 -3.501 8.81 0.995
0.586 -4.517 21.06 0.991
0.412 -5.117 29.49 0.989
53.7 1.93
1.2i2 0.447 -0.256 0.262 5.74 0.964
1.154 0.356 -0.446 0.309 18.91 0.975
0.996 0.390 -0.351 0.951 17.61 0.965
11.4 21.0 46.3 0.88
Martin ef ai. data Bureau of Mines data C.F. % at 520" F. C.F. O/o at 600°F. C.F. % at 700"F. C.F. O h at 800" F.
content of the crude was added as a fourth independent variable. Regression analysis yielded reasonable multiple regression coefficients for the four independent variables (Table IV). Correlation coefficients exceeded 0.96 at each temperature level, indicating the independent variables explained 93 to 95% of experimental variation in aromatics concentrations. Fit of the Martin et ai. data (196313) was considerably better than that of the Bureau of Mines data (Figure 3). Scatter averaged about 1% for the former, compared with about 2'; for the latter. Precision of the two different experimental techniques, mass spectrometry us. silica gel adsorption, may partly account for difference in fit of. the two sets of data. Summary of Composition Correlations. Usefulness of the foregoing composition correlations depends on translation into conventional terms-i.e., per cent of fraction. Data fit in terms of volume per cent of fraction ranged from good to fair. Scatter in calculated per cent aromatics was acceptable for both sets of data (Table V). Calculated volume per cent of paraffins for the Martin et al., (1963b) data were in good agreement with experimental analyses. The Bureau of Mines paraffin data contained both large bias and sizable scatter. All of the experimental data needed to calculate hydrocarbontype composition of crude oil fractions boiling 390" to 800°F. can be derived from rapid crude oil assay. Needed experimental data consists of PONA-type analysis to 390" F., per cent of disubstituted C,,, benzenes, per cent of total C 10 benzenes, and distillationtype yields to 800" F. Physical Property Correlations. Experimental and calculated data derived from six rapid crude oil evaluation tests (Table I) were empirically correlated with physical test data from distillation assay. The 23 test crudes (Table VI) represented a fairly good geographical
Paraffins"
XI X? Z R
Paraffinsa XI = C.F. ' 4 a t 390°F. X, = ';c iso-C;/LCn-C; Z = intercept R = correlation coefficient
... ...
Volume 5 of Fraction
...
Aromaticsb
X I = C.F. L/c at 39OCF. X , = 5; Cl0benzenes X i = 7 disubstituted C I ,benzenes Xi = c i sulfur Z = intercept
R = correlation coefficient
194
Table V. Composition Correlations Data Fit
I B E C PRODUCT RESeARCH AND DEVELOPMENT
Fraction
Volume Tr Aromatics. Ac. Dei'. Martin Bureau of et al., Mines crudes crudes
Volume iParaffins, Ac. Dec. Martin Bureau of et al., Mines crudes crudes
390-520" F. 3.7 4.9 2.7 9.2" 520-600" F. 2.3 5.0 4.1 14.2" 600-700°F. 5.8 ... 3.4 700-800" F. 7.1 ... 4.7 "Not corrected for bias of +3.7 and 14.2CC for 39Oo-52Oo and 520"-600" fractions, respectively.
source range and reasonable variations in quality of all major fractions. Seventeen physical properties of the nine major fractions made up the dependent variables for regression analysis. Nine primary composition indicators (Table VII) were used as independent variables. These included paraffin contents of fractions boiling to 800" F., as calculated from prediction equations derived in the foregoing section. Paraffin level calculated for the 70OC to 800°F. fraction was used as the estimated paraffin content of higher boiling fractions. Normal paraffin concentrations, estimated by the peak height ratio method, were calculated directly from the crude oil chromatogram for both kerosine and gas oil fractions. Analogous values, but including ~ - C X through n-Clj, were used as estimates of n-paraffin levels in all fractions boiling above 700" F. Total aromatics concentrations are likely of limited importance in determining physical properties of high boiling fractions. Properties are more likely to depend on distribution of aromatics types-Le., degree of condensation, type of substitution, etc. Consequently, quality correlations on fractions boiling higher than kerosine did not include total aromatics as a n independent variable. Rather, the two aromatic distribution indicators derived for total aromatics predictions were used directly. A third ring parameter variable was a substituent chain length indicator, per cent of n-butyl benzene expressed as per cent of total C,o benzenes. Polarographically measured condensed aromatics and pentane insolubles or asphalt content were the fourth and fifth aromatics variables. Boiling range distributions variables were also needed for most of the high boiling fractions. These included variable cut points on the lube oil fractions, distillation data (ASTM D1160) where inconsistent fractionation was suspected, and a definition of boiling range distribution in the widest boiling fractions.
Table VI. Test Crudes for Quality Correlations Crude
Specific Grauity at 60" F.
Wt. 70 Sulfur
0.8702 0.8338 0.8822 0.8453 0.8265 0.8649 0.8304 0.8418 0.8403 0.8193 0.8044 0.8383 0.8504 0.8256 0.8260 0.8428 0.8388 0.9065 0.8189 0.8639 0.8762 0.8054 0.8299
1.60 0.14 1.40 0.25 0.89 1.61 0.26 0.11 0.34 0.50 0.38 0.42 0.39 0.47 0.35 0.22 0.29 2.12 0.91 1.78 1.60 0.35 0.53
Gach Saran (Kharg, Iran) South Louisiana mix Oklahoma mix Illinois basin Magnolia (Ark.) Wesson (Stephens, Ark.) Bentonia (Miss.) Chatham (Ohio) Canadian blend Michigan North Louisiana m b Canadian-Michigan mix Rocky Mountain sweet West Texas intermediate Illinois mid-continent Illinois basin East Texas Mississippi Washita-Fredericksburg (Miss.) Paluxy (Miss.) North Dakota intermediate Citronelle, Ala. North Dakota light
Table VII. Range in Primary, Composition Type Variables Mean % Sulfur % Nitrogen Calculated 5'0 paraffins Kerosine (340"-520°) Gas oil (520"-700") Light neutral (700"-800") n-Paraffins P.H.R. from GLC Kerosine (34Oo-52O0) Gas oil (520"-700") Higher boiling (7OO0-80O0) Aromatics distribution indicators Alkyl benzene persistence Aromatic condensation Substituent chain length Pentane insolubles Polarographic condensed aromatics
0.76 0.09 55.1 42.7 31.8 0.53 0.45 0.39 19.6 42.9 4.2 2.0 244
Range 0.110.0137 31 24
-
0.470.400.35-
Correlation Results
2.72 0.29
Regression analyses indicated the composition parameters used adequately described physical properties of the high boiling fractions. Twenty-seven of the 45 correlation coefficients were greater than 0.90 (Table V I I I ) , and only two were less than 0.80. Best results, in general, were obtained on those tests reflecting over-all chemical composition-Le., specific gravity, refractive index, etc. Fit of the specific gravity data was particularly good. Differences between calculated and experimental gravities averaged less than 0.0040 for five of the seven high boiling fractions and did not exceed 0.010 in any case. Quality characteristics evaluated for kerosine included
84 58 46 0.63 0.50 0.45
13.7 - 24.3 19.4 - 55.1 0.0 - 6.5 0.1 - 13.8 134 -432
Table VIII. Summary of Quality Correlations-Correlation Light Neutral Quality Pammeter Specific gravity" U.O.P. "K" factor Cetane No. Refractive index Aniline point Ring no. Smoke point Carbon residue Freezing point Pour point 5'~Wax Thermoviscosity 100" viscosity 210" viscosity Viscosity index Retained ductibility Petrolene VGC
Cracking Kerosine Stack Gas Oil (340"-520") (520"-EP) (52O0-7OO0) 0.982 0.950 0.966
... ...
0.950 0.872
...
...
0.974
0.963 0.943
...
... ... 0.859
0.953
... ...
... 0.801
...
0.831
...
...
0.924
...
... ...
...
Ci-200" specific gravity, R
=
...
... ... ...
Coefficients ( R )
Heavy Neutral
Bright Stock
(700"-800") Raw Dewaxed
(800"-900") Raw Dewaxed
(900"-EP) Rax Dewaxed
0.961 0.947 0.946 0.963
0.868
0.956
0.910
...
...
...
...
... ... ...
... ...
0.741
...
0.870
...
...
0.911
...
... ...
...
... ...
...
...
...
0.959
... ... ... ... ...
...
...
... ... ...
...
...
0.953
...
...
...
... ...
0.821 0.791 0.921
...
... =
0.949
...
...
...
...
...
...
...
0.893; 200"-390" specific gravity, R
0.858
0.935
...
...
... ...
...
0.822
...
...
...
0.861
...
...
... ...
0.816
... ...
... ... ... ...
...
0.800 0.901 0.898 0.970
Asphalt
0.852 0.881 0.922
...
.
.
I
... ... 0.914 0.915
0.923. ~~
VOL. 7 NO. 3 S E P T E M B E R 1968
~
195
four experimental tests and three calculated values (U.O.P. "K" factor, cetane index, and ring number). All correlations were satisfactory (Table VIII). Data fit on the primary experimental tests was quite good (Figure 4 ) . Calculated freezing points averaged within 1" of experimental, and scatter in smoke points averaged only 1.4 mm. Gas oil quality correlations were good for both the four experimentally measured physical properties and the two calculated values (cetane number and U.O.P. "K''). Excellent fit of the refractive index data (Figure 5 ) was particularly gratifying. Primary quality tests on the three lube cuts were wax content and viscosity index of the dewaxed oil. Viscosity
0.82
I
"K"
U.O.P.
o , 8 3 ~S p e c i f i c
index correlations were excellent, with 85 to 94% variation in experimental data explained by the rapid assay composition parameters. Calculated viscosity index data averaged within 5 units of experimental for three lube fractions (Figure 6). Wax content correlations were also good on the two lighter fractions, but data fit on the bright stocks was relatively poor. Specific
I
Gravity
Slack
Wax
(6OOF.)
0,941
J
Factor
(60°F
92 Viscosity
16
8
0.94
,
Index
24
Carbon Residue
/
(Wt. %) 0 . 7 9 K
,
,
,
0.79 0.80 0.81 0 . 8 2
, 0.83
v,
,
,
,
11.8
12.0
12.2
Freezing
Point
11.6
0
35b Smoke
Point
I
05
20
20
40
60
80
Exptl
1.5 I OI 15 Exptl
0 5
2.0 2 0
Figure 6. Lubricating oils quality correlations
0 t i g h t neutral (700"-800" F.)
x
Bright stock (900"-aspholt)
ExptI
Exptl
Heavy neutral (800"-900" F.)
Figure 4. Kerosine quality correlations Specific
88
Specific
U0 P
Gravity
"K" F a c t o r
0 90
U 0 P "K" Factor
Gravity
i y,,,, ( 6 0 "F
)
112 0 2
0 ?
0
8
0 06
B t
,
8
~~~
0
1
d
/
0
0
086 0 84 0
118
0.86
088
090
II 6
084
086
088
II 8
120
122
Carbon Residue /20
Refractive
Index
9
looo
190-
Viscosity
-0
8
180-
7
170
I
0.2
6 6 Exptl
7
8
9
ExptI
Figure 5. Gas oil quality correlations 196
1
I & E C PRODUCT RESEARCH A N D DEVELOPMENT
0
0.4 Exptl
0.6
0
0
0
II 8
Anil;:;
120
I22
)Point
*/ 170
180 190 Exptl
Calculated f r o m analyses o f blending f r a c t i o n s
Figure 7. Cracking stock quality correlations
200
for all fractions. High n-paraffin contents had a positive effect on freezing point, pour point, and wax content; and this eeect was also consistent through bright stock. The alkyl benzene persistence indicator and polarographically measured condensed aromatics were the least effective ring parameters. A high degree of condensation, as suggested by the aromatic condensation indicator, contributed to poor burning qualities (carbon residue and smoke point) and increased viscosity. High values for the substituent chain length indicator contributed to lower gravities, improved freezing characteristics, and higher viscosities. Effects of high sulfur contents were consistently positive for both gravity and carbon residue.
Data fit for the six quality parameters of cracking stocks ranged from excellent to relatively poor. Correlation coefficient for the U.O.P. "K" factor was significantly lower than that obtained for specific gravity (0.872 us. 0.950), suggesting that boiling range distribution was not adequately defined by the independent variables. Carbon residue data included one questionable experimental number. Regression analysis of cracking stock carbon residues included an experimental value of 0.72% for one crude (Figure 7). Residue calculated from the regression equation was only 0.43't, but this compared well with the 0.42% calculated from carbon residues measured on individual fractions comprising the cracking stock blend. Correlations on the three asphalt quality tests were acceptable (Figure 8). Calculated values for both specific gravity and petrolene VGC averaged within 0.009 unit of experimental values. Significance of Composition Parameters. Statistical significance of any single independent variable in any single test was not achieved. Significance in terms of reasonable directional effects and consistency with molecular weight was achieved. High paraffin contents contributed to low gravity (negative effect), and the effect was consistent
Specific
Gravity
Information derived from the 3-hour crude oil evaluation scheme is summarized in Table I X . Fifty-nine quantity and quality parameters are derived from the six experimental tests. Computations are cumbersome but are easily and rapidly handled by computer. The six experimental tests described do not necessarily represent the only approach to utilizing the concepts developed in this work. The gas chromatographic analysis
Petrolene
(60OF.)
I
Discussion of Results
VGC
Retained Ductility
(IOOOF.)
I
I
1.02f
;1.00:
'-
I
0,984
/ 1'
0,96t/
,
,
,
0.96 0.98 1.00 1.02
0.82
0.84 0.86 0.88 0.90
Exptl
0
Exptl
40
80
120
Exptl
Figure 8. Asphalt quality correlations
T a b l e IX. I n f o r m a t i o n Derived f r o m R a p i d C r u d e Oil E v a l u a t i o n
Light iVaphtha C, 200" F. of crude Specific gravity
; 1 '
Light Naphtha r?:?"-POO~ E'.
' i of crude ..,
O.N.
Light Neutral Lube-770O0-80Uo F Rau Deuaxed ('c of crude .. Specific gravity Specific gravity Carbon residue ., , 5; wax ... 100' viscosity 210' viscosity V.I.
Refo rni er IVap h tha
Kerosine
200"-.Y90aF .
.'140"-320=F .
L ; of crude Specific gravity L ; parafins ' naphthenes '.c aromatics
L c of crude Specific gravity U.O.P. "K" Cetane no. Ring no. Freezing point Smoke point Themnoviscosity
Heaull Seutral Lube-800"-90Oo F . Rau Dewaxed 'C of crude ... Specific gravity Specific gravity Carbon residue ... ... wax 100' viscosity 2100 viscosity V.I. i-
Gas 011 520"- 70P F .
Cracking Stock 521% Asphalt
c,1 of crude Specific gravity U.O.P. "K" Cetane no. Pour point
'; of crude Specific gravity U.O.P. "K" Pour point Aniline point Carbon residue 210" viscosity
R.I. 100" viscosity
-
Bright Stock-900O-Asphalt Rau Dewaxed 'c of crude ... Specific gravity Specific gravity Carbon residue ... L-
c WaX
...
Asphalt (Bottoms) 5; of crude Specific gravity Pet. VGC Retained ductility
100" viscosity 210" viscosity V.I.
VOL. 7 NO. 3 S E P T E M B E R 1968
197
procedure, in particular, was developed several years ago and does not take full advantage of present day technology. For example, capillary column GLC analysis for Cg through ClC,aromatics (Martin and Williams, 1963a) would yield more accurate aromatics distribution data and likely improve composition prediction accuracy. Similarly, urea adduction followed by GLC analysis (Martin et aE., 196313) could provide a more accurate measure of n-paraffin distribution and improve accuracy of physical property evaluation. The composition prediction techniques described could also be easily adapted to a partial distillation assay of crude oil.
through courtesy of the Bartlesville Petroleum Research Center, Bartlesville, Okla. Literature Cited
Gaylor, V. F., Jones, C. N., Landerl, J. H., Hughes, E. C., Anal. Chem. 36, 1606 (1964). Holliman, W. C., Smith, H. M., McKinney, C. M., Sponsler, C. R., United States Department of the Interior, Bureau of Mines Tech. Paper 722, 1950. Martin, Ronald L., Winters, John C., Anal. Chem. 35, 1930 (1963). Martin, Ronald L., Winters, John C., Williams, Jack A,, “Composition of Crude Oil, Geological Significance of Hydrocarbon-Type Distributions,” Sixth World Petroleum Congress, Section V , Frankfurt, West Germany, June 1963. RECEIVED for review January 18,1968 ACCEPTED June 13,1968
Acknowledgment
Distillation assay data was supplied by R . E. Farrell, The Standard Oil Company (Ohio). Retain samples of crudes analyzed by the Bureau of Mines were obtained
KINETICS OF DEPOSIT FORMATION FROM HYDROCARBONS Effect o f Trace Sulfur Compounds W I L L I A M
F .
T A Y L O R
A N D
T H O M A S
J .
W A L L A C E
Government Research Laboratory, Esso Research and Engineering Co., Linden, N . J . 07036 Trace levels of various pure sulfur compounds markedly influence the rate of deposit formation from hydrocarbons a t 200” to 450°F. in the presence of oxygen. Structural effects are important, as individual sulfur compounds differ greatly in their effect on the rate of deposit formation. The addition of various pure sulfur compounds including thiols, sulfides, disulfides, and condensed thiophenes to an essentially sulfur-free hydrocarbon material at the 1000-p.p.m. S level increased the rate of deposit formation to a factor of 20. The addition of diphenyl sulfide and dibenzothiophene a t the same level did not increase the rate of deposit formation. The sulfur compounds which increase the rate of deposit formation decompose a t the conditions studied into radical fragments which initiate the complex free-radical autoxidation reactions that lead to the formation of deposits.
high-speed supersonic aircraft, aerodynamic heating causes metal skin temperatures to rise considerably above those encountered in subsonic aircraft. I t has been estimated for a plane such as the Supersonic Transport (Mach 2.7) that the temperature of an uninsulated fuel tank could rise to 430°F. (Chemical Week, 196’7). Other studies have shown that hydrocarbon jet fuels exposed to such high temperature stress can degrade and form deposits (Churchill, 1966). One particular problem area is the formation of deposits in fuel wing tanks which contain puddles of residual liquid hydrocarbon and hydrocarbon vapors. Taylor and Wallace (1967) reported the general features of the complex kinetic reaction system in which such deposits are formed. The deposits are a result of free-radical, autoxidation reactions. Trace levels of sulfur compounds were found to influence the deposit formation process. However, the rate of deposit formation IN A
198
I & E C PRODUCT RESEARCH A N D DEVELOPMENT
of various fuels could not be related quantitatively to the total sulfur content of the fuel. The results indicated that fuels of very low sulfur content are relatively stable and fuels of high sulfur content are fairly unstable. The effect of trace levels of various pure sulfur compounds on the rate of deposit formation was investigated in the present study. Experimental Apparatus. Details of the kinetic unit used to measure the rate of deposit formation were reported by Taylor and Wallace (196‘7).
The main section of the unit consists of a glass tubular reactor which has five separate reactor heaters, each independently controlled by its own Gardsman temperature controller. The reactor heaters are controlled so that the fuel encounters a sequence of rising temperature zones as it flows down the reactor. Carefully weighed metal strips, approximately 1.0 x 10 cm., are positioned
