Model-based ammonia slip observation for SCR control and diagnosis

The control of selective catalytic reduction (SCR) systems, via NH $_{\text{3}}$ injection, requires from a precise estimation of the SCR load in order to ensure NOx reduction by minimizing ammonia slip. This article aims to resolve the cross-sensitivity of current NOx sensors at the outlet of the SCR, by providing the control unit with an estimation of NOx and ammonia slip. The problem of discerning between NOx and ammonia slip is solved by identifying an intermediate variable representing the SCR load. The SCR load is estimated by combining the mass conservation principle between the inlet and the outlet of the SCR and a NOx reduction model, via an extended Kalman filter. Current models and observers have several limitations to represent the real behavior of the SCR along all the operating conditions; on one hand, when relying on the mass conservation, small errors at models are integrated, leading to important bias on the SCR load and on the other hand, the dynamics at the SCR must be precisely modelled for an adequate adaptation of the model. The main focus of the developed algorithm is to use a simplified model which might be used for ammonia slip estimation, being aware of current limitations of SCR models in real operation. Experimental results in a EURO 6 compression ignited (CI) engine show the potential of such observation in transient conditions and an adequate correlation with external ammonia measurements provided by additional sensors available on the test bench.