October, 1928
INDUSTRIAL A N D ENGINEERING CHEMISTRY
1045
Importance of Mixture Ratio in Rating Fuels for Knock John M. Campbell, Wheeler G . Lovell, and T. A. Boyd RESEARCH LABORATORIES, GENERAL MOTORSCORPORATION, DETROIT, MICH.
It is shown that the knock rating of one fuel with respect to another may depend upon the carburetor setting at which the comparison is made. The influence of mixture strength is often so large that, unless careful attention is given to this factor during knock measurements, it is possible to obtain widely discordant results in spite of what may otherwise be the best of experimental technic. If knock ratings of some fuels made by different laboratories are to be comparable, the different experimenters must agree upon some general specification as to the mixture ratio at which the measurements shall be made. Because it is both the point of highest intensity and one that can be located readily, it is proposed that knock measurements be made independently for each fuel at the mixture ratio giving its maximum degree of knock.
HE tendency of a motor fuel to knock is influenced
T
by several factors. Nost of these may be classified under two general headings:
Physical: Compression ratio, timing of the spark, speed of the engine, setting of the throttle, shape of the combustion chamber, timing of the valves, efficiency of cooling temperature of jacket water, and atmospheric conditions. ( b ) Chemical: Composition of the fuel and mixture ratio. (a)
In knock-testing one customary procedure is to make comparative measurements between some standard fuel and the fuel under test. The relative tendency to knock is then expressed in terms of some one of the preceding variables which affect the tendency to knock. Since there is a choice of variables, it is not surprising that several methods for making comparative measurements of knock have been proposed. But a t the present time there does not appear to be satisfactory agreement between the results obtained by various laboratories using different methods for testing the same fuels.'J The purpose of this paper is to point out the importance of controlling mixture ratio during knock measurements, and to show that unless proper attention is given to mixture ratio it is possible to obtain widely discordant results in spite of what may otherwise be the best of experimental technic.
results for some time to come, the run was started. Bouncingpin readings were taken successively for each fuel a t each fuel level. This cycle was repeated three times. Samples of exhaust gas were taken a t each fuel level while the engine was running on gasoline containing ethyl fluid. Determination of carbon dioxide in these samples served as a key to the mixture ratio a t each fuel level. The n-heptane, the isooctane ( 2 , 2, 4-trimethylpentane), and the ethyl fluid used in this work were obtained from the Ethyl Gasoline Corporation. The ethyl fluid contained 54.5 per cent by volume of lead tetraethyl. The gasoline used as a standard in these experiments was a widely used automobile gasoline. Physical Properties of Materials SOURCE d
MATERIAL
(20'/4 n-Heptane Isooctane Benzene, c.
Jeffrey pine oil synthesis P.
I 9
Edgar, J . SOC.Automofive Eng., 22, 4 1 (1928). MacCoull, I b i d . , 22, 457 (1928).
0.6859 0.6911 0.8776
c.
96.8-99.4 97.6-99.8 78.6-79.6
Results
The results of several runs are presented in Tables I to V. From these data Figures 2 to 8 were constructed. The mixture ratios were determined in terms of conventional units as pounds of air per pound (1 pound = 0.454 kg.) of fuel corresponding to the different fuel levels by referring the percentage of carbon dioxide in the exhaust gases at each fuel level to a representative plot of exhaust gas composition against mixture ratio. These values for mixture ratio do not apply to the benzene blends nor to mixtures of heptane and octane which have physical and chemical properties that are different from those of gasoline. In plotting the readings of the bouncing-pin indicator the average of the readings for each fuel a t each fuel level was plotted instead of the individual values, in order that the charts might be read as easily as possible. T a b l e I-Comparison of B e n z e n e B l e n d s w i t h G a s o l i n e C o n t a i n i n g 2.0 cc. of Ethyl Fluid per Gallon (3.78 Liters)
I 1 I
FUEL LEVEL
BOUNCING-PIN READINGS, CC. GAS PER MINUTE
2:;
per gal.
Inches
Beuzene in Gasoline
2.0 CC.
GASOLINE
Experimental
This work was carried out upon a single-cylinder, variablecompression engine equipped with an evaporative cooling system and fitted with a bouncing-pin indicator. The compression ratio mas adjusted to give a fairly loud knock with the fuel under examination. The speed was held a t 850 r. p. m. and the spark advance was set a t 30 degrees. The fuel was metered to the engine through a mixing valve. The mixture ratio was adjusted by changing the fuel level with respect to the fuel jet in the mixing valve. Figure 1 shows the fuel reservoirs, the two float bowls independently adjustable for height by means of a rack and pinion, and the three-way cock for changing from one fuel to another. The mixing valve is shown also. Before beginning each run, the engine was allowed a preliminary warming-up period, and during this time bouncingpin measurements were taken on one of the test fuels. When the constancy of these readings was enough to assure uniform
BOILINGPOINT
C.)
Per cent
-I
~
I
cc. 0.64 0.66 0.45 0.26 0.57 0.69 0.61 0.47 0.25 0.63 0.65 0.60 0.50 0.31 0.64
16%
18%
20%
22%
cc.
cc.
cc.
cc.
0.72 0.66 0.62 0.43 0.63 0.71 0.75 0.64 0.44 0.63 0.70 0.73 0.70 0.51 0.63
0.62 0.68 0.60 0.30 0.50 0.65 0.65 0.59 0.38
0.62 0.61 0.50 0.26 0.43 0.60 0.61 0.55 0.35 0.55 0.61 0.61 0.60 0.30 0.50
0.50 0.49 0.44 0.21 0.32 0 50 0.50 0.50 0 30 0.44 0.48 0 52 0.50 0.30 0.41
0,51
0.64 0.69 0.64 0.35 0.53
~
A study of the figures indicates that the rating of one fuel with respect to another depends upon the carburetor setting a t which the determination was made. For example, in Figure 2 a t the carburetor adjustment corresponding to a mixture ratio of 14.8 pounds of air per pound of gasoline, 17 per cent benzene in the standard gasoline appears to be equivalent to 2.0 cc. of ethyl fluid per gallon in the same standard gasoline, whereas a t the carburetor adjustment
correspondirrg to :in air-gasoline mixture rat,io of 12:1, it would take 22 per cent of beliaene to be equivalent to the same amount of ethyl fluid. Likewise, the data plotted in Figure 3 show that 5.0 cc. of ethyl fluid per gallon in the stnlidard gnsoliiic may be deterinii~cdas equivnleiit to concciitmt,ions of from 30 to 39 pcr cent benzene in tile siniidard gnaoline, :~ccording to the cnriioretor adjiistrnclit, xrhen tlrc latter is the same for lmth fuels. Thcrc is, tlierefore, a considerable range in composition of benzene hleiids w l l i c h appear t o 1iaT-e a tenderrcy to I~IXJI!~; eqoivnlent to a given concentration of lcnd tetr:ietliyl in gasoline. This rmge is iiidicakd by tlie shaded portion of Figure 4. Tlic sliadrd :ma 011 the chart may be used also to interpret the range of lead tetraetliyl eqoimlent, to a g i m i iicnzetre hleiid. Thus, it shows thnt from 1.3 to 2.0 cc. of lend tetr:iethyl per gallo~i niny be found equivalcrrt lo 25 per cent beiieeiie in g-asoliiic. tile vnliic found depeiiding upou the carburetor setting at tlie time t,lie determiiiation was ninde.
Tlic last, titree r u i ~ susing mixtures of n-heptane and isoiictane as tlie standard fuel show a similar effect, although the range of composition having equivalent ratings against lead tetraethyl at different nlistuve r:stios is not so wide as for benzene. Figire 8 is a plot, of heptane-octane composition against lend totraetliyl in the sbndard gasoline, a t mixture r:itios between 11.5:l aiid 15:I for gasoline. Table Ill-Comparison of Fuel Confainins 48 Per Cent n-llepfnne and 52 Per Cent IsoZftane w i t h Gnseline Alone and with Gasoline C:ontaininC 0.5 EC. Of Ethyl Fluid per Gallon ..
I
l k ~ ~ : N c I x < ~ - 1RCAi>ih.CiS, ~rl4 C C . C A S Pill . w l ~ c , T n
0.80
1.75 2.00 2.25 2 so 2 73
.
an
Table 11-Cornpa
a.25 1.75 2.oo 2.25
2.50 2.75
Inihrs 3.oo 3.50 3,25 2 . no 1 .73 1 ,613 1.75 2.00 2.25 2.50 3.m
1.50 1.76 2.0 2.26 2.60 a 011 1 io
PEI re328
9.3
cc.
e'.
Ci.
ci.
cc.
0.26
0.60
0.51
11.41 0.68 0.60 0.75 0.50
0.31
0.41) 0.48
10.4 11 6 12.8 13.2 14.u
0.6:I
0.60
0.62 n.6n 0.62 0.50 0.42 11.27 11.61 0.71 0.64 I!. ii2 0 .48
13.8 12.4 11.4 10.8 9.4
0.25
0.71 11.75
0 ,7!)
0.82
(1.68
0.68 0.80
0 48
0.70
0.52 0.73
0.60 1).80
0.78 0.65 0.47 0.43 0.64 0.81 0.78
0.72 0 58 13.41'
0.36 0.52 0.68 0.68
0.60 0.411 0.41 0.60 0.65
0.68 0.60
n..so I1 39
0.49 11.62 0.60 tr.45 0.28 0.40 0.59 0.51 0.42 0.31 0.29 0.49 0.60 0.58
11.35 0.58
0.41 0.50 0.60 0.40 0.31 11.25 0.40 0.66 0.62 0.54 0.46 0.26
3.00 3.25 1.75 2 (!0 2.25 2.60 2.75
n.mr
0.70'
0.60 0.46 0.35 0.:36 0.65 0.75 0.71 n.62 0.48
0.38
0.20
11.31 0.6U 0.66 0.65 1!.50 0.83
~
3.oa
1j.47
0.66 0.76 0.70 0.61
0.60 0.88
0.26
3.25
~
0.50 0.36 0 :io
ci
-__
C.ll 0.27 0.37
0.88 0.31 0.26 0.17
0.28
0.44 0.40 0.80 0.32 0.15 0.26 0.41 0.41 11.35 11.2x
.2-,. .,-aX-
3.00 2 25 2 ,(10
1 . 0
1a.n
2.00 2.25' 2~5n
13.8
2.75
12.0
3.lE
:I, 2 5 3.50 3.2.5 S.lI@ 2.75 2.50 2.25
a.no 3.26
0.53
I1.4i
0.4!)
11.37 I,.lY 11.47 11.3!)
0.35 0.54 0.48 0.48 0.54
0.45 0.46
0.4s
0.41)
0.40 I1 .211
0.:1z
0.20
0.10 0.12
0.81 0.41
n.si
0.51 0.48 0.17
o.28 0 . I!! O.OX 0.13 0 . 30
0.39 C1.40
0.41 0.31 0.14
0.50 0.44 0.24 11.41) 0.40 0.411 11.40 11.60 0.48 0.31 0.22 0.08 0.27 0.44 0.48 0.52 0.60
0.38 0.17
October, 1928
I N D CSTRIAI, ,450 ESGISEERISG CHEJIISTRY
FUEL LEVEL Figure 2-Comparison of Benzol Blends with Gasoline Containing 2.0 cc. of Ethyl Fluid per Gallon
Figure 3-Comparison of Benzol Blends with Gasoline Containing 5.0 cc. of Ethyl Fluid per Gallon
per Gallon
Figure 6-Comparison of Fuel Containing 38 Per Cent n-Heptane a n d 62 Per Cent Isoiictane with Gasolines Containing 1.5 a n d 2.0 cc. of Ethyl Fluid per Gallon, Respectively
Discussion
One fundamental asumption in almost every method for measuring knock has been that changes in the variable!: which influence knock lial-e the same effect upon all fuels. This assumption is decidedly erroneous n-hen applied to changes in mixture ratio. I t is possible to have two fuels, d and B , of different composition which knock just alike a t one carburetor adjustment. although a t a leaner adjustment d may knock less than B and a t a richer adjustment A may knock more than B. As an illustration. consider 18 per cent benzene and 2.0 cc. of etlijd fluid per gallon in Figure 2 as fuels -4 and B , respectively. At the carburetor setting corresponding to about 13.8 pounds of air per pound of gasoline these two fuels knock alike. but a t leaner mixture ratios 18 per cent benzene knocks less and a t richer mixture ratios more than 2.0 cc. of ethyl fluid per gallon. I n other words, the rate of change in tendency to knock with respect to changes in carburetor adjustment varies according to the chemical composition of the fuel.
Therefore. in order that different laboratories may be able to o b t a i n comparable a n d reproducible m e a s u r e m e n t s of knock, it is necessary to decide upon some c o in m o n m i x t u r e s t r e n g t h a t which knock-testing shall be done. This mixture ratio must be adhered to closely b e c a u s e the tendency to knock is sensitive to comparatively s m a l l c h a n g e s in- mixture ratio.
1047
Figure4-How Mixture Ratio Affects Relationship between Antiknock Properties of Benzol a n d Lead Tetraethyl
Figure 8-How Mixture RatioAffectsRelationship between Antiknock Properties of n-Heptane-Isotictane Mixtures, a n d Lead Tetraethyl
1048 The question arises-what
INDUSTRIAL A N D ENGINEERING CHE-JlISTRY mixture ratio shall be chosen?
It seems hardly practicable to select a certain mixture ratio at random. This would require the constant use of gasanalysis apparatus or some device for measuring both air and fuel in order to make sure that the mixture ratio was properly adjusted. The procedure would become quite complicated with fuels which have different products of combustion and different physical properties. The mixture ratio for maximum power would not be suitable because power measurements are not sufficiently sensitive to changes in mixture ratio to make a satisfactory adjustment possible. There remains one outstanding and comparatively simple solution. For every fuel there is a definite mixture ratio a t which maximum knock occurs; and, if the determinations are made a t the mixture ratio for maximum knock with
Vol. 20, No. 10
each individual fuel, there is available a standard for obtaining reproducible results by different laboratories. Conclusions
1-The tendency of a fuel to knock is very sensitive to changes in mixture ratio. 2-Since the change in tendency to knock with changes in carburetor adjustment varies according to the composition of the fuel, it is necessary to choose some definite ratio mixture for knock-testing work if reproducible results are to be obtained. 3-The use of mixture ratios giving maximum knock for each fuel is suggested as a convenient means for obtaining more consistent results between different laboratories.
Action of Accelerators and Inhibitors upon the Oxidation of Liquid Hydrocarbons T. E. Layng and M. A. Youker UNIVERSITY OF ILLINOIS, URBANA, ILL.
A n apparatus has been devised and a method described for determining the effect of various inhibitors and accelerators of knock upon the slow oxidation of hydrocarbon fuels. Data are given to show the effect of various substances upon the slow oxidation of n-heptane and its normal oxygen derivatives, gasoline, and kerosene. A surprising similarity is shown between the action of lead tetraethyl and various compounds of sodium and potassium, and also a difference in the action of lead tetraethyl and these compounds of sodium and potassium upon the oxidation of hydrocarbons, in the gas or liquid phases.
HE status of the mechanism of oxidation of hydro-
T
carbons, or the cause of detonation in the internalcombustion engine, is still in an unsatisfactory state. The use of antiknock substances, notably lead tetraethyl, has added a new factor and has been the cause of increasing vastly the amount of useful information which must be obtained for the solution of this problem. I n a recent summation of our knowledge of this subject Clark1 shows that a t the present time no theory is without its defects, but the peroxide theory supported by the work of Riloureu, Dufraisse, and Chaux,2Callendar,3 and more recently of M a r d l e ~appears ,~ the most promising. This theory postulates that detonation a t high compression ratios is due to the formation of explosive peroxides by the oxidation of liquid droplets of the fuel. The above investigators and also Lewis6 studied the action of oxygen upon various hydrocarbon fuels in the presence of catalysts a t different temperatures. This method of study has also been advocated by Clark.'j The present investigation was undertaken in the hope of designing a more efficient apparatus and thus producing a means of securing more data in regard to the action of catalyst of oxidation and the causes of detonation. The results of preliminary work are presented herein. J. SOC.Aulomolloe Eng., 21 (1928). Chimre el induslne, 18, 3 (1927). 8 EngZnee7ing (London), 121 477 (1927). 1 J. Chem. SOC.(London), 1928,872. 6 Ibrd., 1927, 1556. * IND. END.CHEM., 17, 1210 (1925). 1
2
Apparatus
The experiments on which Moureu and his associates based their conclusions were conducted with the fuel to be tested and the oxygen in a small glass bulb with a manometer attached. When the bulb was immersed in a hot bath, the rise of mercury in the manometer was a measure of the oxygen absorption and hence also of the rate of peroxide formation. Lead tetraethyl and other antiknock substances, such as the aromatic amines, retarded the absorption of oxygen. Peroxides were detected in the final products when oxygen had been absorbed. The method of recording pressures in the apparatus of Lewis and Mardles was similar to that employed by Moureu. I n each case it was possible for some of the fuel to be distilled and condensed in the manometer connections. Accordingly, an apparatus (Figure 1) was designed which was extremely simple and eliminated any possibility of distillation. The Pyrex bulb is of an average volume of 160 cc. The side arm permits easy cleaning, the introduction of solid catalysts, and the withdrawal of gas samples. The mercury in the base seals off the bulb and gives a column of mercury when the bulb is heated which enables changes in the pressure 'of the enclosed system to be detected. Since the entire bulb is immersed, changes in pressure under constant-temperature conditions can only be caused by the thermal decomposition of the fuel or its chemical union with oxygen. In this apparatus there has been introduced the added factor of mercury in contact with the fuel when in the liquid phase, and with the fuel and oxygen in the case of the gaseous phase. However, no mercury compounds have been shown to have any effect upon the rate of oxidation, and since in all the tests with different accelerators and inhibitors a comparison is made with the fuel alone, any eflect of the mercury mag be assumed a constant for a given fuel a t a given temperature. Before using, the bulb was cleaned in hot sodium hydroxide solution, washed out with dilute hydrochloric acid, and rinsed several times in distilled mater. It was then washed out twice with acetone and dried. It was found to be very important to clean the bulbs very carefully.
