Although the proper timing of the spark is as essential as the spark itself and the electrical and mechanical devices for producing the spark have been many, little attention has been given to the study of spark-advance. An error in timing of ± 20 deg. in a low-compression engine, or of ± 15 deg. in most other engines, has been shown experimentally to cause a loss of 10 per cent from the best power and economy, provided other conditions remained the same. Hand or semi-automatic control can average hardly closer than ± 15 deg. to the correct advance because the speed and the load combinations are constantly changing on the road.
Two important phases mark the spark-advance problem. The practical question as to whether the requirements of optimum spark-advance at various combinations of load and speed are such as might be controlled automatically is apparently answered affirmatively; it can be satisfactorily represented by additive functions, one of speed only and one of load or intake-suction only. Hand adjustment would still be needed to take care of the difference between a clean engine and a dirty one or of a cold engine. When once an engine has been warmed up, however, automatic controls could maintain the proper spark-advance, thereby increasing in practice the power, flexibility and economy of the engine.
The second phase is that of scientific analysis. Inasmuch as combustion takes time, and as the engine is rotating during the combustion process, spark-advance is essential, for it maintains a definite relation between the progress of the explosion and the motion of the piston of the engine. This relation should be such that half the rise of pressure during combustion would occur at the dead-center position of the piston. Analysis, however, both theoretical and experimental, shows that one-half of the pressure rise occurs substantially at three-fourths of the explosion-time, that is, the interval between the point of ignition and the pressure peak. This, then, is the numerical basis for the relations of explosion-time, engine speed and optimum spark-advance.
The existing data, relating to the explosion-time as affected by the mixture-ratio, the size of the combustion-chamber, turbulence, dilution with dead or exhaust gases and the temperatures preceding the explosion, are reviewed. Density is shown not to affect the explosion-time. The factor commonly supposed to be density, which demands an increased spark-advance as the engine is throttled, is in reality dilution with exhaust gas, which increases as the throttle closes, and the cause of the faster explosion in a high-compression engine than in one of low compression is the temperature preceding ignition.
A simple mathematical law, connecting the explosion-rate and turbulence and derived from experiments on bombs, is shown to be applicable to engines, and the manner of its application to the turbulence factor of any engine is indicated. This opens the way to quantitative experiments on turbulence in various designs of engine, hence to the development of designs for producing the greatest amount of turbulence, if such development should seem desirable, as turbulence is the factor that makes really high rotative speeds compatible with a good power output. An equally important factor in making explosions in engines occur more quickly than those in bombs is the heat produced adiabatically during the compression stroke.
The slowing-up of combustion on account of the dilution of the charge with exhaust gas was measured experimentally, and the results are tabulated and compared with the numerical extent of the dilution. Assuming that the dilution ratio is the ratio of the total charge to the quantity of new gas, the slowing-up of combustion because of dilution is demonstrated experimentally to be about proportional to the cube of the mass-dilution ratio.
The scientifically valuable part of the paper is the naming of the factors that affect the explosion-time of an engine, the giving of mathematical expressions for their laws of action and the finding of the numerical
values of constants for such factors as turbulence and dilution. The measurement of the optimum spark-advance is made available as a research method for investigating the reaction rates of combustion, and hence of all that related group of topics now of interest to automotive engineers.
By measuring the optimum spark-advance, the combustion rates of gasoline with and without “anti-knock,” or tetraethyl-lead, were measured. Although the quantity used was 20 times the normal amount, no change in the reaction rate of combustion was found when the combustion remained normal, that is, without detonation. When detonation occurred without anti-knock, the reaction times with anti-knock followed those to be expected with normal combustion. Detonation apparently changed the combustion habit as if it produced an abnormal top to the combustion.