Sensitivity Study on Thermal and Soot Oxidation Dynamics of Gasoline Particulate Filters

Gasoline particulate filters (GPFs) are devices used to filter soot emitted by gasoline direct injection (GDI) engines.

A numerical model for a ceria-coated GPF presented in a previous paper by H. Arunachalam et al. in 2017 was developed to predict internal temperature and soot amount combusted during regeneration events. Being that both the internal temperature and the accumulated soot cannot be directly measured during real-time operation and owing to their critical importance for GPF health monitoring as well as regeneration scheduling, the above model turns out to be a valuable tool for OBD applications. In this paper, we first conduct a stochastic analysis to understand the relation between the model parameters and the initial value of the ceria (IV) oxide volume fraction, as a deterministic value for such a state is not known. A particle swarm optimization (PSO) algorithm was employed to define what type of relationship the model parameters were with respect to the initial state of the ceria (IV) oxide volume fraction. A sensitivity study is then conducted over the model parameters to study parameter identifiability from system measurements. Effects of the initial temperature and initial amount of soot were studied as well. Results indicated that the model is most sensitive to the activation energy of GPF regeneration, agreeing with previous studies. Additionally, the model was shown to be able to predict the GPF temperature with less than 5% error when there was at most 20% uncertainty in the parameters. The results of the relationship between ceria (IV) oxide and the parameters, as well as the sensitivity analysis can be used simultaneously in the future for observer-based design.