Modeling and Model-based Control of Automotive Air Paths

Abstract: The strive towards cleaner and more efficient combustion engines, driven by legislation and cost, introduces new configurations, as exhaust gas recirculation, turbocharging, and variable valve timing, to name a few. Beside all the positive effects on the emissions and fuel consumption, they all affect the air-charge system, which increases the cross-couplings within the air-path control, making it an even more complex system to control. As the SI engine uses a three-way catalytic converter, which enforces a condition of stoichiometric combustion, the amount of air flow and fuel flow are connected. This means that the air flow has a direct impact on the driveability of the engine, through the torque. As configurations are constantly improved or added, a component and model-based methodology is chosen in the thesis, as it would bring flexibility and the possibility to reuse previous developments. As it enables the engineers to keep down the development cost and at the same time bring along knowledge of the systems through the model's descriptions.The air-charge system's task is to supply the combustion chamber with the correct air mass flow, in the most energy efficient way. To be precise in the control of the air mass flow, the actuators are also constantly developed and becoming both faster and more precise. One example of this is in the first part of the thesis, where an electric servo controller for the wastegate actuation is implemented and compared against the more traditionally used actuator, controlled through a pressure difference over a membrane. As the focus for the air-charge system is the control of mass flows, how these flows can be represented by compact models is also investigated in the thesis, as compact models are beneficial for control from a computation time perspective. In the last part of the thesis a simulation study for controlling the intake manifold pressure, with a constraint on intake manifold temperature, using the throttle as actuator is investigated. Lastly, an implementation of a model predictive controller acting as a reference governor, for the throttle and intake cam phasing, in an engine test cell is demonstrated. As the controller only acts as a reference governor it makes it possible for an engineer to develop the actuator controllers independently if a closed loop model of the actuator system is supplied to the controller. The coordination, of the two actuators, is solved by letting the intake cam phasing depend on the intake manifold pressure, that is a state.