On automotive engine intake manifold dynamic modeling, estimation, and control

A first principles mean value dynamic model of an engine's intake manifold has been developed that includes external exhaust gas recirculation (EGR) and variable cam timing (VCT). Using a perfect mixing assumption, mass conservation linearity, and an engine pumping model, it was possible to separate fresh air dynamics from external exhaust gas recirculation dynamics. This approach gives a simple solution that can be easily implemented and utilized in real-time discrete control systems. In total, the model, included both fresh air dynamics and external exhaust gas recirculation dynamics, has three parameters. Experimental data validated the modeling approach and real-time implementation.; An external exhaust gas recirculation control system, including an upstream control valve and a downstream measuring orifice, was modeled using thermodynamic orifice flow equations and mass conservation. A simplified nonlinear approximation to the full adiabatic orifice model was analytically shown to provide low approximation error when constrained by physical engine operation limits. Steady state engine dynamometer data verified the approximation to the EGR system.; The validated intake manifold dynamic model and EGR system model were utilized to develop an EGR flow estimation using a limited sensor set. Namely, a sliding observer was developed for dynamically estimating EGR flow primarily using a throttle body mass air flow sensor and a differential pressure across the downstream EGR measuring orifice. A sliding observer was selected because of the nonlinear model coupling and calibration ease.; From the intake manifold dynamic model and the EGR system model, a simulation was developed for use in controller and estimator design. The sliding observer was first validated using the simulation. Experimental vehicle data was then used to validate the observer. A very accurate EGR flow estimate was obtained with high bandwidth.; Two controllers were then developed using the models, simulation, and observer. A conventional PID type controller was compared and contrasted in simulation to a time-optimal controller. A linearization technique was used in developing the time-optimal controller. To give a worst case comparison, the PID controller used estimated EGR flow in feedback control, while the time-optimal controller assumed perfect measurement. In other words, the PID controller used the sliding observer in the loop while the time optimal controller was provided with a perfect flow measurement. Finally, the tuned PID controller was experimentally validated in a vehicle.