Model predictive control: A unified approach for urea-based SCR systems

Despite the fact that urea-based selective catalytic reduction (SCR) of NOx is a key technology for achieving on- and off- highway diesel emission standards, significant control challenges remain. Transient operation, combined with dramatic changes in catalyst dynamics over the operating range, cause highly nonlinear system behavior. Moreover, these effects depend on catalyst formulation and new catalysts continue to be developed. With many controllers, any difference in catalyst formulation, converter size, and engine emissions calibration require control system re-tuning. To minimize control development effort, this paper presents a novel "generic" controller for SCR systems. Control action is grounded in a physics-based, nonlinear, embedded model. Through the model, controller parameters are adjusted a priori for catalyst formulation and converter size. The few remaining tuning levers are quite intuitive, and require no special knowledge of controls theory. The controller design is a type of model predictive control, and includes novel features for on-line determination of simplified, linear models as well as urea dosing limits for NH3 slip control. Robust performance is achieved using feedback from NOx and NH3 sensors. Controller capability is demonstrated for SCR inlet conditions corresponding to an EPA 2007 heavy-duty diesel operating on the European Steady-State Cycle (ESC) and US EPA heavy-duty FTP cycle.