Development of a Control-Oriented Ignition Delay Model for GCI Combustion

Increasingly stringent pollutant emission limits and CO 2 reduction policies are forcing the automotive industry toward cleaner and decarbonized mobility. The goal is to achieve carbon neutrality within 2050 and limit global warming to 2 °C (possibly 1.5 °C) with respect to pre-industrial levels as stated in both the European Green Deal and the Paris Agreement and further reiterated at the COP26. With the aim of simultaneously reducing both pollutants and CO 2 emissions, a large amount of research is currently carried out on low-temperature highly efficient combustions (LTC). Among these advanced combustions, one of the most promising is Gasoline Compression Ignition (GCI), based on the spontaneous ignition of a gasoline-like fuel. Nevertheless, despite GCI proving to be effective in reducing both pollutants and CO 2 emissions, GCI combustion controllability represents the main challenge that hinders the diffusion of this methodology for transportation. Several works in the literature demonstrated that to properly control GCI combustion, a multiple injections strategy is needed. The rise of pressure and temperature generated by the spontaneous ignition of small amounts of early-injected fuel reduces the ignition delay of the following main injection, responsible for the torque production of the engine. Since the combustion of the pre-injections is chemically driven, the ignition delay might be strongly affected by a slight variation in the engine control parameters and, consequently, lead to misfire or knocking. The goal of this work was to develop a control-oriented ignition delay model suitable to improve the GCI combustion stability through the proper management of the pilot injections. After a thorough analysis of the quantities affecting the ignition delay, this quantity was modeled as a function of both a thermodynamic and a chemical–physical index. The comparison between the measured and modeled ignition delay shows an accuracy compatible with the requirements for control purposes (the average root mean squared error between the measured and estimated start of combustion is close to 1.3 deg), over a wide range of operating conditions. As a result, the presented approach proved to be appropriate for the development of a model-based feed-forward contribution for a closed-loop combustion control strategy.