228
INDUSTRIAL AND ENGINEERING CHEMISTRY
T'ol. 19, No. 2
Comparison of Gasolines b y Analytical and Engine Tests' By Donald R. Stevens and Samuel P. Marley MELLONINSTITUTE OR INDUSTRIAL RESEARCH, UNIVERSITY OF PITTSBURGH, PITTSBURGH, PA.
ITHIN recent months much attention has been given specialIy designed carburetor having three cups maintaining to the possible value of cracked gasolines as anti- the same fuel level, all feeding through the same metering knock motor fuels. Cracked gasolines are believed orifice. The temperature of air supplied to the carburetor to differ from straight-run gasolines in having a higher content was regulated by means of an exhaust stove. A loadof olefins, and possibly of naphthenes and aromatics. The controlling device and power-recordinginstruments were used. relatively high nonknocking value of these types of hydroDetonation Indicator carbons has been generally recognized. It is generally agreed that the various methods so far proThe indicator is a mod5cation of the bouncing-pin type. posed for the chemical analysis of the complex hydrocarbon Instead of leaving the pin free, as in the Midgley gas-generatmixtures present in commercial gasoline are not rigidly exact. ing indicator, the pin is tapped into the indicator piston and Recently Egloff and Morrel12have combined what are possibly held rigidly at the top by lock nuts, the tension of the main the most c o n v e n i e n t of s p r i n g being adjustable. these methods into a pro- 11 ,I The shock of the detonacedure for the determination Eighteen gasolines composed entirely of petroleum tion is transmitted to an Of four Of were analyzed for their content of paraffins, naphanvil secured to the short ents-namelyl thenes, aromatics, and unsaturated hydrocarbons acarm of an indicating lever, naphthenesl aromatics, and the long arm of which travels cording to the method recently described by Egloff unsaturated hydrocarbons. and Morrell. The same gasolines were then tested for over a scale. Movement of this lever is damped by a Using data Obtained by Ridetonating tendency by engine test, using a directvery light spring. The inthey the reading detonation indicator. Comparison of the dicator is so adjusted that Of naphthenes benzene equivalents calculated from analysis with and unsaturated when the engine is not dethose determined by engine tests shows that the agreebans in terms Of t o n a t i n g the lever shows ment is moderately fair for about half the fuels studied, but slight movement. The thus arriving at but that rather wide discrepancies are found in the a r o m a t i c equivalent for degree of detonation in the other cases. I t is recommended that fuels be studied each fuel analyzed. The engine is controlled by adfor detonating tendency, only by direct engine test. use of further data of Rijusting the spark timing, card04 permitted reading the carburetor adjustments,
W
I S D USTRIAL AND ENGINEERISG CHEMISTRY
February, 1927
Table I-Description NO.
1 2 3
5 6 7 8 9 10 11 12 13 14 15 16 17 18 a
GASOLINE
INITIAL
DRY
A . P. 1. 52.8 58.8 55.9 58.7 60.9 66.7 02.7 63.2 62.9 60.3 58.8 58.1 59.2 56.6 55.2 56.6 57.1 46.9
c.
c.
.~
118 132 131 112 106 108 132 128 110 122 128 116 110 107 117 115 138 166
Table 11-Analysis
No.
GASOLINE
of Gasolines UXSATU-
RAT&D
ARO- N A P S - PARAFMATICS THENES FINS
%
%
%
%
3
Straight-run midcontinent A Straight-run midcontinent B Special 9
2.8 2.6 1.7
3.9 5.0 9.4
19.7 19.5 18.9
73.6 72.9 70.0
4 5
Motor A Special B
5.2 7.5
10.0 11.4
16.9 18.5
67.9 62.6
6 7
Special C California
3.5 2.2
10.4 7.3
28.1 31.7
58.0 58.8
8 9
Straight-run Pennsylvania Motor C
2.5 3.9
Trace 3.2
14.9 10.7
82.6 82.2
17.8 17.4 16.8 13.0 15.0 19.6 20.2
20.6 20.2 23.1 18.0 18.5 22.5 23.0
12.1 12.8 12.1 12.6 12.6 11.1 8.6
49.5 49.6 48.0 66.4 53.9 46.8 48.2
10 11
12 13 14 15 16
Untreated from P. S. D.-A Treated from P. S. D.-B Cracked Smackover Motor B Cracked midcontiaent Untreated gasoline from P. S. D.-C Untreated gasoline from P. S. D.-D
17
Cracked Pennsylvania
11.7
16.7
6.7
64.9
18
Vapor-phase cracked
29.5
44.7
12.4
13.4
Method of Test
I n making a series of tests, the engine is run until uniform temperature has been attained, and speed, load, and other conditions are then adjusted so that the standard fuel will give the desired reading on the indicator. With a standard fuel t'he same degree of detonation can be obtained from day Table 111-Analysis NO.
OLEFINS
GASOLINE
Synth. 17 14 18 1
2210
c.
284' C.
%
426 426 425 428 408 430 430 440 475 416 410 436 436 436 427 434 458 410
392" C.
%
28.5 24.5 21.5 28 32 30 22 32 27 30.5 30.5 31 26 26 23 24.5 17 17.5
RECOVERY
%
%
55 51 52 53.5 65 58.5 50 67 51 65.5 61 55 53 45 41 48 40.5
91 91.5 93 91 94 90 92 91 86 94 95 91 88 84.5 84 89 85 94
5s
96 97 97 96 95 94 97 95 93.5 96.5 97 96 94 95.5 96 95.5 97 97
.
Pressure-still distillate.
composition, but different source, are given by gasolines 1 and 2 as compared with 3, and 4 as compared with 5, the two latter being partly cracked, blended products of uncertain origin. Pennsylvania cracked gasoline (17) and a vaporphase cracked gasoline (18) are each characterized by a composition quite different from that of the other gasolines studied. In order to test the accuracy of the method of analysis, synthetic mixtures of pure hydrocarbons, one of each of the classes considered, were made up and analyzed. The hydrocarbons were normal heptene from Jeffrey pine oleoresin, hexylene prepared from secondary hexyl alcohol by passing the vapor oyer reduced tungstic oxide a t 400" C., methylcyclohexane, and C. P. toluene. Mixtures were made to duplicate gasolines 1, 14, 17, and 18. Only 14 and 17 were analyzed, but all four were used in engine tests. The composition and results of analysis are presented in Table 111.
1 2
of Gasolines
GRAVITS
Straight-run midcontintent A Straight-run midcontinent B Special A Motor A Special B SDecial C California Straight-run Pennsylvania Motor ~ ~C -. . . ~ Untreated from P. S. D.a-.i Treated from P. S.D.-B Cracked Smackover Motor B Cracked midcontinent Untreated from P. S. D.-C Untreated from P. S. D.-D Cracked Pennsylvania VaDor-Dhase cracked
229
Cracked Pennsylvania Cracked midcontinent Vapor-phase cracked Straight-run midcontinent A
of Known Mixtures AROMATICS
Found
%
%
11.7 l5,0 29.5
8.0 18.2
2.8
to day with only minor adjustments of the engine and the indicator. These adjustments are required because of uncontrollable variations in condition of spark-plug points, breaker points, valve and piston seal, etc. With a standard degree of detonation obtained on the standard fuel, the supply is changed, by turning a three-way cock, to the fuel under test; and after the engine has adjusted itself to operation on the new fuel, readings are made on the indicator. Changes back and forth from the test fuel to the standard fuel are made several times. Usually the readings for each fuel are constant, but any slight variations are minimized by averaging all readings. Enough benzene or kerosene (as the case may be) is then blended with the standard fuel to bring the degree of detonation, as shown on the indicator, down or up to that of the fuel being tested. In securing the results reported in this paper, separate sets of tests were made by two observers a t times several weeks apart. Very satisfactory duplication was obtained. I n the data presented by Egloff and Morrell the equivalence of olefins, naphthenes, and aromatics was calculated on the basis of data obtained by Ricardo, indicating that 5 per cent of olefins had the same effect in reducing detonation as 4 per cent of naphthenes and 1 per cent of aromatics. Bpparently these equivalents of Ricardo were not in all cases obtained by the use of pure hydrocarbons. The authors have therefore studied the effect of the hexylene, methylcyclohexane, and toluene mentioned above. Two series of tests were made, one by adding each hydrocarbon in turn to gasoline 1, straight-run midcontinent products, and the other by adding the same proportions to gasoline 4,an average motor gasoline. The same results were obtained in both fuels, indicating that 2 per cent of hexylene and 2 per cent of methylcyclohexane gave the same effect as 1 per cent of toluene. These results were duplicated in several series of tests with different carbureting equipment, and were further checked by a rather elaborate system of intercomparisons. It will be evident that hexylene probably gave greater reduction than would have been obtained with heptylene, which has about the same boiling point as toluene and methylcyclohexane. It is not believed that the difference would have been great enough to change the general nature of the results, and, in general, these tests are regarded as indicating a 2 : 2 : 1 equivalence at a 5.5 to 1 compression ratio. This conclusion does not take into account the
Synth. 70 16.7 18.5 44.7 3.9
Found 70 15.3 13.9
~XAPHTHENES Synth R 6.7 12.6 12.4 19.7
Found
PARAFFINS Synth.
Found
'7,
R,
"n
9.0 13.5
65;O 53.9 13.4 73.6
67.7 54.4
INDUSTRIAL A N D ENGINEERING CHEMISTRY
230
possibility that most petroleum naphthenes may be pentamethylene] rather than hexamethylene, derivatives. Both the new ratio and the 5;4: 1 ratio of Ricardo and of Egloff and Morrell have been used in calculating the results given below. Engine Tests
Table IV-Aromatic
and Benzene Equivalents AROMATICBENZENE AROMATIC
EQUIVA- EQUIVA- EQUIVALENTS
No.
GASOLINB
CALCD. FROM
ANAL-
LENTS OF
LENTS
R
OR
Kb GASOLINE 1
PLUS ADDED BENZENE (1) (2)
ADDEDTO GASOLINE 1
% by WOE.
-
-
-
B7 B30
15.7 36.6
21.1 40.6
15.3 21.0 17.5 24.4
B18 B20
25.7 27.5
30.4 32.1
18.1 15.7
26.2 24.2
B40 B35
45.7 41.1
49.1 44.8
Straight-run Pennsylvania Motor C
4.2 6.2
8.7 10.5
K3 K17
-
-
Untreated from P. S. D.-A Treated from P. S. D.-B Cracked Smackover Motor B Cracked midcontinent Untreated from P. S. D.-C Untreated from P. S. D.-D
27.2 26.9 29.5 23.7 24.6 29.2 29.2
35.5 35.3 37.5 30.8 32.3 37.8 37.4
B B B B B B B
41.1 36.6 30.2 27.5 34.7 36.6 32.0
44.8 40.6 34.7 32.1 38.9 40.6 36.3
17
Cracked Pennsylvania
20.7
25.9
K19
-
-
18
Vapor-phase cracked
53.7
65.6
B48
52.9
55.9
1 2 3
Straight-run midcontinent A Straight-run midcontinent B Special A
9.4 10.4 14.5
4 5
Motor A Special B
6 7
Special C California
8 9 10 11 12 13 14 15 16
15.1 16.0 19.7
table gives only the indicator readings of the mixtures and the natural gasolines, both against the standard fuel, motor gasoline A. Table V-Engine
No.
In Table IV are given the percentages of benzene or kerosene required to m a k e the standard gasoline equal to each test gasoline,the aromatic equivalent of each gasoline as calculated from the analyses in Table 11, according to both the 5:4:1 and the 2: 2; 1ratios, and the equivalents (according to both ratios) of the fuels resulting from the addition of the required amount of benzene to the standard fuel, to balance each test fuel.6 The calculation of the figures in the last two columns may be shown by an example. Gasoline 13 required 20 per cent of benzene added to the standard fuel (gasoline 1) for b a l a n c e . Since only 80 per cent of the standard fuel was present in the mixture, the aromatic equivalent of only the 80 per cent Detonation Indicator should be considered. 15.1X 0.80 = 12.1, which plus 20 = 32.1, the aromatic equivalent of the mixture.
35 30 23 20 28 30 25
(1). using 5 : 4 : 1 ratio; (2) using 2 : 2 : 1 ratio. B indicates benzene added.
b K indicates kerosene added.
The mixtures of pure hydrocarbons listed in Table I11 were also tested in the engine, with the results shown in Table V. No benzene balance figures were determined, and the 8 The general correctness of most of these figures at a somewhat higher compression ratio has been estimated by H. G . Smith at Port Arthur, Texas.
Vol. 19, No. 2
17 14 18 1
Tests with Known Mixtures
GASOLINE Cracked Pennsylvania Cracked midcontinent Vapor-phase cracked Straight-run midcontinent A
INDICATOR READING Synthetic
Natural
20-50 2 M O 20-4 20-45
20-35 20-12 20-8 20-33
Discussion of Results
It will be seen that in the range of cracked midcontinent gasolines (gasolines 10 to 16) the method of Egloff and Morrell gives moderately fair agreement. As soon as materials of widely different types are exanlined, however, discrepancies appear. Attention may be called to gasoline 3, which in the engine balanced a mixture of 30 per cent benzene and 70 per cent standard fuel, equivalent to 36 per cent aromatics by the Egloff and Morrell method of calculation, although its aromatic equivalent by their method of analysis is only 14 per cent. A similar condition is shown by gasoline 6, which balanced a 40:60 mixture of benzene and standard fuel, equivalent to 45 per cent aromatics, although the Egloff and Morrell aromatic equivalent (by analysis) is only 18 per cent. Again, a similar condition is shown by gasoline 7, an uncracked California distillate. Comparison of gasoline 6 with gasoline 12 shows further discrepancy. The former has an aromatic equivalent, by analysis, of 18 per cent, but balanced a 40 per cent benzene60 per cent standard fuel mixture, while the latter has an aromatic equivalent, by analysis, of 29 per cent, and balanced a mixture of 23 per cent benzene with 77 per cent of standard fuel. The analyses and engine tests here show complete disagreement. When gasolines of similar composition as shown by analysis are considered, further differences appear. The most outstanding case is that of gasolines 1 and 3, which correspond closely in analysis and in calculated aromatic equivalent, the latter differing by only 5 per cent; the engine tests of which, however, show that 1 is hard-knocking gasoline and 3 ranks well as an antiknock fuel. Comparison of gasoline 17, Pennsylvania cracked, and gasoline 9, a commercial motor gasoline] brings out the fact that they have about the same knocking tendency] although strikingly different in chemical composition and calculated aromatic equivalents. The reverse is shown by gasolines 8 and 9, which have fairly close calculated aromatic equivalents, but widely different knocking tendencies. It should be noted, however, that the Egloff and Morrell method of analysis and calculation gives very satisfactory agreement with the motor tests in the case of gasoline 18, a vapor-phase cracked product which contains only 13 per cent of paraffins. An examination of the results obtained by the use of the 5 : 4: 1 and the 2: 2: 1 ratios shows that both give rather wide differences from the results found. In general, the 2:2:1 ratio seems to give slightly better agreement. From these results it might be assumed that a great deal of additional information is needed before a definite conclusion can be reached. The results with the pure hydrocarbon mixtures listed in Table V indicate that the composition of commercial gasolines is far more complex than might be assumed from the clrtssification given by the method of analysis discussed. The one case of satisfactory agreement between synthetic and natural fuels occurred in a mixture (vapor-phase cracked) containing only 13 per cent of paraffins. With the cracked midcontinent gasoline, the synthetic fuel knocked harder than the standard, while the natural fuel knocked less than the standard. Since
February, 1927
INDUSTRIAL AND ENGINEERING CHEMISTRY
a normal paraffin hydrocarbon was used in the synthetic mixture, the explanation may lie in the possibility that the paraffin hydrocarbons in the natural, cracked fuel are of branched-chain structure. Similar considerations apply in a less degree to the cracked Pennsylvanian and the straightrun midcontinent fuels. This point may serve also to explain the case of gasolines 8, 9, and 17, containing by analysis an appreciable aromatic equivalent, but actually requiring, by engine test, from 3 to 19 per cent of kerosene mixed with the standard for balance. Conclusions
a result of chemical analyses and engine tests on eighteen gasolines, it would appear that the method of evaluation of motor fuels here considered has given results agreeing approximately with engine tests for about half of the fuels studied. The other gasolines examined showed rather wide discrepancies between the findings of analysis and engine test. It is not believed that the antiknock value of gasolines can be satisfactorily determined by this method. 1-AE
231
2-Vsing pure hydrocarbons, an equivalence in knock reduction of approximately 2 : 2 : 1 has been found for a naphthene, an olefin, and an aromatic hydrocarbon, as represented by methylcyclohexane, hexylene, and toluene. 3-Qualitative evidence has been found for the idea that there are striking differences in the detonating tendency of the paraffin hydrocarbons present in different gasolines. Normal heptane was found to knock harder than petroleum paraffins and this may indicate the desirability of branched-chain paraffins as motor fuels. 4-The authors do not know of any dependable method for determining the detonating tendency of motor fuels except that of direct engine tests. Acknowledgment
The authors wish to acknowledge their indebtedness to
W. A. Gruse, of Mellon Institute, for valuable suggestions, and to J. 0. Timms, W. H. Ragsdale, and E. C. Martin, who aided in the experimental part of these investigations.
Influence of an Antiknock Compound in a Gas-Ion Oxidation’ By S. C. Lindz and D. C. Bardwell FIXEDNITROGEN RESEARCH LABORATORY, BURSAW
OF SOILS, WASHZNGTON,
D.c.
The actual comparison of the rates, with and without diethyl selenium, of the slow oxidation of methane under the ionizing influence of alpha-radiation does not indicate any retardation by ihe antiknock compound but rather some acceleration. The interpretation of this and its possible bearing on antiknock theory are discussed. UMEROUS efforts have been made to find a satisfactory theory for the action of the antiknock compounds of M i d g l e ~ . ~In 1924 Wendt and Grimm4 made experiments which they believed would support an electronic theory of detonation and at the same time explain its suppression by antiknock compounds. Their experiments consisted in passing air that had been ionized by passage through an arc, over a pool of benzene with or without dissolved tetraethyl lead or aniline. They found that a smaller proportion of ions reached an electroscope placed in the air stream beyond the benzene pool when the latter contained an antiknock than when it did not. From this they concluded that the function of the antiknock is to cause the more rapid recombination of gas ions, and if we admit the latter to be a positive factor in detonation, its suppression would result. More recently, however, Clark, Brugmann, and Thees have used x-radiation as a more constant source of ionization and fail to find that the presence of an antiknock influences the rate of recombination of the ions. The opportunity for a more direct test seems to be offered by studying the influence of an antiknock on a slow oxidation known to be proceeding under ionizing influence. The slow oxidation of methane a t ordinary temperature under the ionizing influence of radon was chosen as a suitable reaction. On account of its greater volatility, selenium diethyl6
N
1 Presented under the title “Etlects of Antiknock in Slow Oxidation of Methane under Influence of Ionization” before the Division of Gas and Fuel Chemistry at the 72nd Meeting of the American Chemical Society, Philadelphia, Pa., September 5 t o 11, 1926. 2 Present address, University of Minnesota, Minneapolis, Minn. I THISJOURNAL, 16, 421 (19’23) I b i d . , 16, 890 (1924). 6 I b i d . , 17, 1226 (1926). 6 Mr. Midgley was kind enough t o furnish a supply of selenium diethyl.
was selected as the antiknock compound, instead of lead tetraethyl. Experimental Method
The method employed was essentially that which the writers have used in measuring the velocity of various reactions.’ It consisted in saturating a mixture of C& 202 with selenium diethyl a t 22.4’ C., where the vapor pressure of the latter is 38.4 mm., and then diluting samples of the saturated gas to give mixtures 0.001, 0.01, and 0.046 (almost saturated a t 25” C.) molars with respect to selenium diethyl. These mixtures were introduced into the glass reaction spheres of about 2 cm. diameter to which suitable amounts of radon had been added. The course of the reaction was followed by means of the decrease of pressure (see tables).
+
Experimental Results
Preliminary tests showed that the oxidation of methane in the absence of L‘antiknock’l could be readily measured manometrically.9 The oxidation proceeds, apparently in one step, completely to form water and carbon dioxide: CHI
+ 202 = Cot + 2Hz0
Chemical determination of carbon dioxide affords a final check on the reaction assumed. Calculations
The kinetics have been calculated by use of the general equation : 7 J. Am. Chem. Soc., 41, 531, 551 (1919); 46, 2003 (1924); 47, 2675 (1925); 48, 1556, 1575 (1926). 8 “Molar” in this paper means the mol fraction. 0 Lind and Bardwell, J . A m . Chcm. Soc., 4S, 2335 (1926).