Previous Emissions Work

A concerted effort is on-going to develop novel control methods for aftertreatment systems in automobiles. One of our goals is to combine the state of the arts tools in chemical engineering and in controls to provide an effective framework for modeling, optimization, adaptation, and model-based control for future generation of after treatment systems for lean-burn gasoline direct injection and diesel engines. Our objectives are to lay the foundations for developing (i) reduced-order, control-oriented models that can incorporate the fundamental principles of conservation equations and species reactions, (ii) data-driven and based on system identification methods while capturing the dominant physio-chemical dynamics/interactions of the Urea SCR system, (iii) procedures for first principle-based model calibration from limited data sets and parametrize data-driven models from either experimental data or first principles based model simulation data, and (iv) enable the design and development of initial adaptive control strategies for active control of urea injection predicated on the nonlinear structure of the data-driven models.

A class of physically based control-oriented models of urea-selective catalytic reduction (SCR) after-treatment systems have been derived.; Starting from first-principles, appropriate simplifications are made in the underlying energy and species equations to yield simple governing equations of the Urea-SCR. The resulting nonlinear partial differential equations are linearized and discretized to yield a family of linear finite-dimensional state-space models of the SCR at different operating points.; It is shown that this family of models can be reduced to three operating regions that are classified based on the relative NOx an NH3 concentrations. Within each region, parametric dependencies of the system on physical mechanisms are derived. A further model reduction is shown to be possible in each of the three regions resulting in a second-order linear model with sufficient accuracy. These models together with structured parametric dependencies on operating conditions set the stage for a systematic advanced control design that can lead to a high NOx conversion efficiency with minimal peak-slip in NH3. All model properties are validated using simulation studies of a high fidelity nonlinear model of the Urea-SCR, and compared with experimental data from a flow-reactor.

Recent results using a closed-loop adaptive Proportional-plus-Integral (PI) controller show a NOx conversion efficiency of over 95%, with a maximum NH3-slip of less than 5 ppm. An adaptive PI-controller was simulated on the full chemistry model, and shown to be capable of delivering over 90% of NOx conversion efficiency at a peak Ammonia Slip of less than 2 ppm. Preliminary experimental results from a test vehicle have demonstrated that with little to no tuning, efficiency levels of SCR above 80% can be achieved as adaptation proceeds, when driven in city traffic, with a mean ammonia slip of around 20 ppm.

  • Active-Adaptive Emission Control