Nonlinear Model Predictive Control for Cold Start Selective Catalytic Reduction

Selective catalytic reduction (SCR) is emerging as a key technology for reducing emissions of nitrogen oxides (NOx) from diesel vehicles, but the temperature dependence of the governing chemical kinetics are highly nonlinear and industry standard techniques of limiting ammonia injection until the catalyst reaches operating temperature leave room for improvement of NOx reduction. Cold start emissions constitute a significant fraction of urban NOx emissions, due to low road speeds and short journeys precluding the catalyst from reaching operating temperature quickly. We demonstrate that nonlinear model predictive control (NMPC) provides the desired control performance in adhering to the required constraints and meeting the complex control objectives regardless of catalyst temperature. The results include improved overall NOx reduction during a typical test cycle including cold start, without design effort specifically for low temperature operation. We also show that the controller is amenable to real-time implementation for use in a vehicle.