A Statistical Approach to Improve the Accuracy of the DPF Simulation Model under Transient Conditions 2019-01-0027
For reduction of Particulate Matter (PM) emitted from a diesel engine, a Diesel Particulate Filter (DPF) is mounted on a diesel car in general. Since a high pressure drop through the DPF causes low engine performance, it is important to predict the pressure drop with high accuracy. The purpose of this study is to improve the accuracy of the DPF simulation model under transient condition by means of an improved parameter optimization.
The DPF model had been created that consisted of an inlet channel, a cake layer, wall layers and an outlet channel. The pressure drop is influenced by the soot deposition place, mass and density. Therefore the model included following sub-models:
• Sub-model 1: Calculation of the soot density deposited in the wall layer
• Sub-model 2: Calculation of filtration efficiency and the mass of the wall layers and the cake layer
• Sub-model 3: Calculation of the soot density deposited in the cake layer
Firstly some parameters called fitting parameters had to be optimized, since these three sub-models included some empirical formulas.
Secondly, two types of experiments were carried out. One was soot loading experiment under steady-state conditions for parameter optimization. The other was soot loading experiment under transient conditions for model validation.
Thirdly, the optimal values of the fitting parameters were found by Fletcher–Reeves conjugate gradient method. The optimization results reveal that the fitting parameters depend on Space Velocity (SV) and PM diameter. Based on the results, these parameters were implemented into the DPF model as polynomial functions of SV and PM diameter.
Finally, the DPF model with the polynomial functions was validated under transient conditions. The simulation results show that the prediction errors were reduced compared to the model with constant values. This approach will be applied for a pressure prediction under other transient conditions like WLTC mode.