Most studies comparing diesel/gasoline dual-fuel operation and single-fuel diesel operation in diesel engines center on time-averaged results. It seems few studies discuss differences in cyclic variability. Motivated by this, the present study evaluates the cyclic variability of combustion in both dual-fuel and single-fuel operations of a diesel engine.
Steady-state tests were done on a medium duty diesel engine with conventional direct injection timings of diesel fuel into the cylinder at one speed and three loads. In addition to single-fuel (diesel) operation, dual-fuel (gasoline and diesel) operation was studied at increasing levels of gasoline fraction. Gasoline fuel is introduced via a fuel injector at a single location prior to the intake manifold (and EGR mixing location). Crank-angle resolved data including in-cylinder pressure and heat release rate obtained for around 150 consecutive cycles are used to assess cyclic variability.
The sources of cyclic variability, namely the factors causing cyclic variability or influencing its magnitude, especially those related to cylinder charge amount and mixture preparation, are analyzed. Fuel spray penetration and cyclic variability of cylinder charging, overall A/F ratio, and fuel injection timing, tend to increase cyclic variability of combustion in dual-fuel operation. On the other hand, fuel type and fuel spray droplet size tend to increase cyclic variability in single-fuel operation.
The cyclic variability in dual-fuel operation in this study is more serious than that in single-fuel operation, in terms of magnitude, indicated by metrics chosen to quantify it. Most measures of cyclic variability increase consistently with increasing gasoline fraction. Variations of gasoline amount and possibly gasoline low temperature heat release cause higher combustion variation in dual-fuel operation primarily by affecting premixed burning.
Statistical methods such as probability density function, autocorrelation coefficient, return map, and symbol sequence statistics methods are used to check determinism. In general, the parameters studied do not show strong determinism, which suggests other parameters must be identified to establish determinism or the system is inherently stochastic. Regardless, dominant sequences and optimal sequence lengths can be identified.