Identification and characterization of complex dynamic structure in spark ignition engines

The goal of this investigation was to identify and characterize complex dynamic structure in cycle-to-cycle variations from a spark ignition engine and to investigate the effect of key engine parameters on cyclic dispersion patterns. New data analysis techniques from nonlinear dynamics and chaos theory were used to reveal previously unrecognized patterns in cycle-resolved measurements of the combustion process. Results from these analyses indicated a transition from stochastic behavior to noisy nonlinear determinism as equivalence ratio was decreased to very lean conditions. The transition to deterministic behavior appeared to occur via a period-doubling bifurcation sequence. Stochastic, periodic, and possibly chaotic behavior were observed over the range of the bifurcation. The level of communication between successive combustion events and the complexity of the dynamics were a strong function of the level of residual gases in the cylinder. The transition from stochastic to deterministic behavior was observed in both a single-cylinder research engine and a multi-cylinder production engine, indicating the basic phenomena involved are likely the same for many engine designs. Experimental patterns were also compared with patterns predicted by a recently proposed engine model. This comparison supported the hypothesis that the combustion instability develops as a noisy period-doubling bifurcation sequence. The model was also used to demonstrate that multi-cylinder synchronization may occur in engines due to noise components common to all cylinders.; Key engine parameters were observed to have a significant effect on the characteristics and development of lean combustion instability. Specifically, the effects of engine coolant, fuel-air charge motion, fuel injection timing, and ignition timing were investigated under lean conditions. Several of these parameters were capable of enhancing lean combustion stability and should be considered in designing control schemes for extending the effective lean limit in spark ignition engines. The observation of dynamic structure under lean conditions may have important implications for engine diagnostics and control since cyclic dispersion is not a purely random process.