NOx Trap Model For Lean Burn Engine Control

One of the most environmental problems nowadays is the reduction of the global CO2 emissions. In the field of automotive transportation, car manufacturers in Europe plan to reach the level of 140 g(CO2)/km in 2008) [1]. At this level, one of the most viewable solution is the use of lean-burn engines. This kind of engines are known for their combustion efficiency, but they present a major inconvenience: NOx emissions and their post-treatment. Some technical solutions have been provided in the past years such as DeNOx catalysts. However these solutions won't be compatible with future European legislations: limits of diesel engine emissions will reach 0.25 g(NOx)/km in 2005. One possible technology to overcome - such severe limits - is the NOx trap catalyst proposed by TOYOTA [2], which is already used by some car manufacturers on serial vehicles (such as PSA [3] and VW [4]). NOx trap catalysts are known to work within two phases: Trapping NOx when the engine is running “lean” Regeneration (NOx reduction) when the engine is running “rich” Implementation of such technology requires advanced strategies of engine control, which consist in periodically rising fuel to air ratio from lean to rich. The design of such strategies requires real-time models describing the trapping and regeneration process. The aim of this work is to propose such a NOx trap catalyst model. This model describes the trapping phase. It includes gas temperature and composition variations. The numerical simulations and comparisons with real data obtained from an experimental synthetic gas bench at IFP show the interest of such physical models. The experimental setup consists of a quartz reactor containing the NOx trap sample, connected upstream to gas mass flow controllers and an analyzer bench downstream. Such NOx trap models can be easily connected to global engine simulations in order to design optimized regeneration strategies. The model has been derived after studying the slow and fast dynamics of the reactions occurring in the NOx trap. So, it was possible to deal with all these reactions using only one “simple” first order differential equation.