Experimental Analysis and Control of Recompression Homogeneous Charge Compression Ignition Combustion at the High Cyclic Variability Limit.

Abstract

The automotive industry currently faces many challenges pertaining to strict emissions and fuel consumption constraints for a sustainable society. These regulations have motivated the investigation of low temperature combustion modes such as homogeneous charge compression ignition (HCCI) as a potential solution to meet these demands. HCCI combustion is characterized by high efficiency and low engine-out emissions. However, this advanced combustion mode is limited in the speed-load operating space due to high pressure rise rates for increased loads. Often higher loads are run at later combustion phasings to reduce pressure rise rates, however high cyclic variability (CV) can also be a limiting factor for late combustion phasings. This work presents advancements in the understanding of high variability dynamics in recompression HCCI as well as methods for control of CV and load transitions which typically encounter regions of high variability.

Standard in-cylinder pressure based analysis methods are extended for use on high variability data. This includes a method of determining the trapped residual mass in real time. Determination of the residual mass is critical in recompression HCCI because of the combustion's sensitivity to the thermal energy contained within the residual charge. Trapping too much or little residuals can lead to ringing or misfires and CV, respectively.

Various levels of CV are studied using large experimental data sets to ensure statistical relevance. The cycle resolved analysis of this data has allowed for the development of a predictive model of the variability associated with lean late phasing combustion. This model is used to develop control which can suppress cyclic variability at steady state.

Knowledge about steady state control of CV and its oscillatory dynamics is further applied to the development of an adaptive controller. The adaptive controller uses a parameter estimation scheme in the feedforward component of a baseline midranging structure. The adaptive feedforward component enables the ability to correct for modeling errors and reduces parameterization effort. Experimental results demonstrate that the control is effective at navigating through large load transients while avoiding excess amounts of variability. Additionally, the actuators spend more time in a region of high authority when compared to non-adaptive control.