Cars with diesel engines are commonly equipped with a Diesel Particulate Filter (DPF) to reduce their emissions of particulate matter (PM). Because the pressure drop within the DPF reduces engine performance, it must be predicted with accuracy. The purpose of this study was to improve the accuracy of a DPF simulation model under transient conditions by parameter optimization.
The DPF model under consideration consists of an inlet channel, a cake layer, wall layer, and an outlet channel. The pressure drop is influenced by the location, mass, and density of the deposited soot. Therefore, the model includes the following sub-models:
Sub-model 1: Calculates the soot density deposited in the wall layer
Sub-model 2: Computes the filtration efficiency and mass of the wall and cake layer
Sub-model 3: Calculates the soot density deposited in the cake layer
Because the sub-models include some empirical formulae, the first step in refining the model was to optimize their fitting parameters.
Two sets of experiments were conducted: soot loading experiments under steady-state conditions for parameter optimization, and soot loading experiments under transient conditions for model validation.
Therefore, the fitting parameters were optimized based on the experimental data using the Fletcher-Reeves conjugate gradient method, revealing that two of the fitting parameters depend on the Space Velocity (SV) and PM diameter. These parameters were therefore represented as polynomial functions of the SV and PM diameter, which were implemented in the DPF model as sub-models.
Finally, the DPF model with the polynomial sub-models was validated under transient conditions, showing that the polynomial sub-models yielded lower prediction errors than were obtained using constant fitting parameters. The revised model will be used to predict DPF pressure drops under other transient conditions such as those associated with Worldwide-harmonized Light vehicles Test Cycle (WLTC) mode.