The mounting evidence of anthropogenic climate change necessitates a significant effort to improve the internal combustion (IC) engine and reduce its adverse environmental impacts due to its ubiquitous use powering ground transportation in the world today. Homogenous Charge Compression Ignition (HCCI) engines present a promising opportunity to reduce the environmental consequences of using IC engines by reusing exhaust from one engine cycle to initiate combustion on the following engine cycle. The presence of high retained exhaust ratios in HCCI engines results in dilute, low-temperature combustion that achieves greater efficiencies and lower CO2 and NOx emissions than conventional spark-ignited or diesel engines. However, three critical obstacles prevent them from being widely adopted: first, unlike conventional IC engines, HCCI engines lack a direct combustion trigger to determine when combustion occurs, and that lack of a direct trigger makes specifying combustion timing challenging. Second, the high quantities of retained exhaust create a strong physical link between engine cycles, resulting in undesirable dynamics that could potentially lead to engine misfire at certain operating conditions. Finally, the high quantities of retained exhaust also prevent the engine from inducting as much fuel and air as possible, limiting the load range of the engine.
This dissertation addresses all three of those obstacles by investigating the abilities of different actuators to control combustion timing and improve the dynamics at certain HCCI operating conditions that could be used to expand the load range of HCCI engines. A simple, physical model that represents one engine cycle as a discrete-time, nonlinear system captures the oscillatory dynamics present at certain HCCI operating conditions on an experimental engine. The physical model provides physical intuition about the sources driving the oscillations and the control actions needed to reduce them. A linearized version of the model depicts the source of those oscillations on a root locus, and shows that a negative real axis pole in a discrete-time, linear dynamical system drives the oscillations.
Three different actuators, exhaust valve closing timing, pilot fuel injection timing, and main fuel injection mass, each reduce the oscillations. For each actuator, a linearization of the physical model illustrates how each actuator can be used with simple linear control laws to improve the dynamics at HCCI operating conditions. Then, the actuators are compared to each other on three bases: the difficulty of the control problem associated with using the particular actuator to reduce the oscillations, the difficulty of implementing the actuator in a production vehicle, and the effectiveness of each actuator at reducing the oscillations.