Switched-Mode Urea Dosing Control Design for Selective Catalytic Reduction in a Diesel Engine
2020-03-06T14:17:07Z (GMT) by
The increasingly-stringent regulations on tailpipe NOx are difficult to meet with engine management strategies in isolation. The Urea-SCR aftertreatment system technology offers a key solution to this problem. However, the complexity of chemical reactions, strong temperature dependence of effectiveness, and the transient nature of disturbances make the system very challenging to control and offer an opportunity to develop creative control techniques.
Urea-dosing controllers developed for previous work in our research group require customization to the specific drivecycle on which they are being run. The goal of the present work is to develop controllers that are able to work on multiple drivecycles without using information specific to the drivecycle. To meet the control objective of maximizing the NOx reduction while maintaining the instantaneous concentration of gaseous ammonia at the tailpipe, referred to as NH3 slip, under a defined ceiling, one slip-reference controller and two switched-mode controllers are proposed. The slip-reference controller consists of a model-based component which uses the observed storage fraction estimate, and a feedback component that uses only NH3 slip information. The two switched-mode controllers switch between slip-tracking and storage-tracking mode depending on a switching logic. The first switched-mode controller uses a constant storage-reference and a temperature-dependent switching logic. The second one is designed to have predictive capability and uses a lookup-table to vary the storage-reference with time. By default, at any time, the controller is assigned the same mode as it was in the previous timestep and is changed to slip-tracking or storage-tracking when the observed storage value is either less than or greater than the storage reference by a certain margin, respectively.
Studies from simulation show the effectiveness of the three control strategies for three different drivecycles. The effects of uncertainty in model parameters, errors in urea dosing, and inaccuracies in the observer's initial conditions are also studied.