Model Predictive Urea Dosing Control Strategy for Heavy-Duty Diesel Vehicles

Abstract: Stricter requirements on the reduction of Nitrogen Oxides (NOx) in the emissions of heavy-duty diesel vehicles drives development for more efficient aftertreatment systems. An ammonia covered catalyst is one of the most successful technologies in reducing NOx by converting it into the harmless byproducts water and nitrogen. The ammonia injection control is however difficult due to nonlinearities and the impact of external exhaust parameters. The ammonia coverage ratio depends heavily on the surface temperature of the catalyst and a rapid increase in surface temperature would lead to a rapid decrease in ammonia storage capabilities. If the storage capabilities decrease below the current level of stored ammonia, the excess ammonia will flow into the exhaust and convert to NOx, an undesired phenomenon due to the cost of and the pollution caused by the ammonia released, often referred to as ammonia slip. This issue is further amplified by the fact that the problem is asymmetric, that is injected ammonia cannot be actively removed but has to be reduced by the reaction with the NOx present in the exhaust. As such, it is very important to keep the level of ammonia storage ratio low enough to avoid slipping but at the same time sufficiently high to obtain a satisfactory NOx conversion efficiency. These two issues are the main reasons why feedback control has proven to be difficult to implement to solve the dosing problem. As one has to store a lot of ammonia in order to obtain a satisfactory conversion of NOx, one often cannot react to rapid temperature increases in the catalyst. As such, one often experiences a lot of ammonia slip during these scenarios. In this report it is shown that utilizing predicted parameters of the exhaust in a model predictive controller reduces the ammonia consumption by 38% while also improving the tracking of the NOx conversion reference by 5.5%.