A Switched, Controls-Oriented SCR Catalyst Model Using On-Line Eigenvalue Estimation 2009-01-1284
Selective catalytic reduction (SCR) of NOx is coming into widespread use for diesel exhaust emissions control in passenger cars, light trucks, and commercial vehicles. Because of the transient nature of these applications, modeling is a critical element of the controls development process. For software-in-the-loop simulation, the model must run in real-time while still retaining first order accuracy. Furthermore, if used as an embedded system or nonlinear observer, the allowable time step must not be shorter than the control module clock rate. Unfortunately, the time scales for ammonia storage decrease exponentially with temperature. The end result is a trade-off between spatial resolution, real-time performance, and temperature range. If the SCR catalyst is placed downstream of a particulate filter, this issue is even more acute due to the high temperatures that occur during regeneration.
A switched catalyst model is proposed that breaks this trade-off. The model computes equilibrium ammonia coverage distribution and catalyst eigenvalues on-line. This is accomplished with a generalized approach that is amenable to a wide range of reaction mechanisms. In this particular case, the model includes ammonia adsorption-desorption with a coverage dependent activation energy. It also includes NO2/NOx ratio effects, NO to NO2 conversion, ammonia oxidation, and N2O formation. When the eigenvalue magnitude exceeds a threshold, the original state equations are replaced with pseudo-state equations. This preserves model order while enabling larger time steps. Model validation is achieved by comparing to published flow reactor measurements for a copper-zeolite coating on a 400/7 cordierite substrate. Model capability and robustness are demonstrated through simulation of an ESC test and a filter regeneration event.