Torque based combustion property estimation and control for diesel engines

Doctoral thesis, 2012

Modern diesel engines are becoming increasingly complex as a result of demands on the reduction of both fuel consumption and emissions. This escalating complexity not only applies to the engine itself, but also to its control system. In this context, the interest in closed-loop engine control is growing as this control strategy offers increased accuracy and robustness, as well as a reduced need of control system calibration, compared to traditional open-loop engine control systems. However, the concept of closed-loop control requires information about the controlled process, in this case the combustion events in the cylinders, and additional sensors are thus needed. The most suitable sensor configuration for the acquisition of combustion information is still subject to research. This thesis deals with estimation and control of diesel engine combustion properties based on crankshaft torque measurements. Methods are presented that describe a combustion event in the torque domain and estimate combustion properties either directly or by first reconstructing the corresponding cylinder pressure. The proposed combustion property estimation methods are evaluated using both simulations and experimental data. Combustion net torque, a novel torque domain combustion description, is a central concept in this thesis. Techniques for combustion net torque based estimation of both entire burned mass fraction traces and the 50% burned mass fraction combustion phasing measure are presented. These techniques are also implemented in a real-time engine control system and used in order to successfully demonstrate torque based closed-loop combustion phasing control online. This experimental demonstration illustrates how disturbances that effect the combustion phasing can be detected and counteracted using crankshaft torque measurements.