Modifying the Classical One Dimensional Catalyst Model to Include Axial Conduction and Diffusion

Abstract

Lean NOx trap (LNT) catalytic aftertreatment devices are one potential option for the reduction of oxides of nitrogen (NOx) in the exhaust of compression ignition engines. They work through a controlled modulation between a storage phase that captures NOx over an alkali earth metal and a regeneration phase that reduces the stored nitrates on the surface using a rich pulse of injected fuel or via stoichiometric engine operation. This rich phase has an associated fuel penalty while being relatively difficult to control through temperature and chemical species. In order to improve system efficiency, a number of researchers have proposed dual leg LNT systems using two LNTs, one of which is always storing while the other is undergoing regeneration. The majority of the exhaust flows through the storage LNT while only a small fraction (low space velocity) advects across the regeneration LNT. This increases the regeneration residence time, improving effectiveness and decreasing the amount of fuel used. From an LNT simulation standpoint, most researchers utilize the classical onedimensional (1D) aftertreatment model constructed from the Euler equations of motion that neglect axial conduction and diffusion. This paper explores the applicability of this model under low flow situations prevalent in a dual leg LNT system through a carbon monoxide lightoff experiment. The authors chose this type of experiment in order to focus purely on fluid mechanics and not the choice of LNT reaction mechanism. The results suggest that a Navier–Stokes (N–S) version of the 1D aftertreatment model is preferred for the regeneration leg of a dual LNT system. Moreover, the authors provide the solution of such a model within this paper.