In-cylinder pressure-based control of premixed dual-fuel combustion

Resumen

The current climate crisis has urged the research community and manufacturers to provide solutions to make the transportation sector cleaner. Among the various technologies proposed, low temperature combustion has undergone extensive investigation. Premixed dual-fuel combustion is one of the concepts addressing the NOx-soot trade-off in compression ignited engines, while maintaining high thermal efficiency. This combustion makes use of two fuels with different reactivities in order to improve the controllability of this combustion mode over a wide range of operation. Similarly to all premixed combustion modes, this combustion is nevertheless sensitive to the operating conditions and traditionally exhibits cycle-to-cycle variability with significant pressure gradients. Consequently, advanced control strategies to ensure a safe and accurate operation of the engine are required. Feedback control is a powerful approach to address the challenges raised by the premixed dual-fuel combustion. By measuring the output signals from the engine, strategies can be developed to adapt and correct the control actions to maintain the desired operation. This thesis presents control strategies, based on the in-cylinder pressure signal measurement, applied to premixed dual-fuel combustion engines. Various objectives were addressed by designing dedicated controllers, where a special emphasis was made towards analyzing and implementing these solutions to the different levels of mixture stratification considered in these engines (i.e., fully, highly and partially premixed). At first, feedback control strategies based on the in-cylinder pressure signal processing were designed. Proportional-integral actions were selected to ensure the desired engine performance without exceeding the mechanical constraints of the engine. Extremum seeking was evaluated to track efficient combustion phasing and NOx emissions reduction. The in-cylinder pressure resonance was then analyzed and a knock-like controller was implemented to ensure safe operation of the engine. Finally, mathematical models were used to design a control-oriented model and a state observer that aimed to leverage the signals measured in the engine to improve the prediction and diagnostic capabilities in such engine configuration. The results from this work highlighted the importance of considering feedback control to address the limitations encountered in premixed combustion modes. Particularly, the use of the in-cylinder pressure measurement showed the relevance and potential of this signal to develop complex and accurate control strategies.