Real-Time Embedded Models for Simulation and Control of Clean and Fuel-Efficient Heavy-Duty Diesel Engines

The ever increasing demand for fuel economy and stringent emission norms drives researchers to continuously innovate and improve engine modes to implement adaptive algorithms, where the engine states are continuously monitored and the control variables are manipulated to operate the engine at the most efficient regime. This paper presents a virtual engine developed by modeling a modern diesel engine and aftertreatment which can be used in real-time on a control unit to predict critical diesel engine variables such as fuel consumption and feed gas conditions including emissions, flow and temperature. A physics-based approach is followed in order to capture vital transient airpath and emission dynamics encountered during real driving condition. A minimal realization of the airpath model is coupled with a cycle averaged NOx emissions predictor to estimate transient feed gas NOx during steady state and transient conditions. The complete airpath and NOx emission model was implemented on a rapid prototyping controller and experimentally validated over steady state and transient emission cycles. The overall performance of the reduced order model was comparable to that of the full state model in predicting the transient behavior of engine airpath dynamics and NOx emission.