Hierarchical multi-time-scale predictive thermal management and fuel optimization for heavy-duty compression ignition engines

Abstract

For heavy-duty diesel engines, NOX emissions reduction is strongly constrained by fuel efficiency. This paper presents a hierarchical model predictive controller (H-MPC) for coordinated control of tailpipe NOX emissions and fuel consumption. The H-MPC uses the separation of slow and fast dynamics that exist in the engine and its aftertreatment system. The controller is synthesized with an architecture in which a high-level MPC uses a longer prediction horizon compared to the low-level predictive controller which tracks the high-level controller command and manages the thermal dynamics of the aftertreatment system. Engine load preview enables the high-level controller to estimate the desired catalyst temperature ahead of time and addresses the selective catalytic reduction (SCR) slow thermal dynamics. Calculated by the high-level controller, the intake manifold pressure, and the start of injection (SOI) crank angle is used as reference trajectories in the low-level controller that regulates fast dynamical behaviors such as engine out NOX emissions. Hardware-in-the-loop (HIL) validation of this integrated H-MPC on a rapid prototype controller shows that when the SCR catalyst temperature is above light-off temperature (warmed-up condition), the engine operation is shifted to operate with the best fuel economy since the warmed-up SCR can efficiently reduce the engine-out NOX emissions. Results indicate that up to 0.8% benefit in cycle averaged BSFC along with a 13% reduction in tailpipe NOX compared to a stock engine calibration can be achieved with the coordinated engine and aftertreatment system through H-MPC.