Oxidation of PM

Oxidation of particulate matter in a diesel particulate filter is known to be due to two global chemical reactions: oxidation of carbon in the particulate matter with oxygen, commonly known as thermal oxidation of PM, and oxidation of carbon in the particulate matter with nitrogen dioxide (N O₂), known as N O₂-assisted oxidation of PM. Thermal oxidation of PM was considered the only form of oxidation in diesel particulate filters [40–45] primarily when the DPF’s did not have catalytic coatings. Models focused on oxidation of PM via 3 reactions: thermal, catalytic and N O₂-assisted [13]. This framework was for the particulate matter in the PM cake layer and it extended the framework for thermal oxidation and catalytic oxidation laid out in earlier works [46,47].

Since one of the primary goals of this research work was to determine the kinetics of PM oxidation, a review of literature covering this topic was carried out. Over the years, various research works have focused on deriving the kinetics of PM oxidation. The earlier works on this topic were focused on thermal oxidation of PM with oxygen (O2). One of the important kinetic parameters that have been reported in open literature is the activation energy of PM oxidation. Recent research works have reported activation energy values of PM oxidation as follows in Table2.1.

Table 2.1: Activation energies of PM oxidation reported in open literature Literature Activation Energy

Source Oxidant

[kJ/mol]

[48] 137±8.7 Diesel Soot O₂

132±5.1 Model Soot O₂

A recent work on PM kinetics of diesel particulate matter [49] studied the effect of fuel type (conventional fuel versus soy-based biodiesel) on the oxidation of PM. The authors also observed that the B100-derived PM is more reactive at the same temperature compared to ULSD-derived PM. This difference was attributed to the presence of more mobile carbon (MC) in Biodiesel-derived PM compared to ULSD-derived PM. Figure2.7shows a relevant result obtained in this work, which shows the reaction rates (shown in log-scale) obtained from oxidation of PM generated using ULSD and B100 fuels, illustrating the higher total reaction rates for oxidation of 40% of the same initial PM at different temperatures, the only difference being the source of PM (fuel-type).

For this research work, the starting point for PM kinetics came directly from an analysis of experimental data from references [54] (for passive oxidation) and [55] (and active regeneration) based on Arrhenius plots and the total reaction rates being assumed equal to N O2-assisted in the passive oxidation experiments and thermal (O2) in the active regeneration experiments.

Figure 2.7: Total reaction rates obtained from oxidation of PM derived from ULSD (red squares) and B100 (green circles) fuels[49] - Reprinted with permission from the authors of SAE Paper No.

2010-01-2127 - AppendixL.

2.4 Oxidation of Gaseous Species (Catalytic Reactions) and Back-Diffusion of N O

The oxidation of gaseous species (HC, CO and N O) in the catalytic wash-coat of a CPF was a significant improvement of the model that was carried out in the MTU CPF high-fidelity model from the existing model [56]. One of the first works about this topic was from Aristotle University Thessaloniki in Greece [52,53] which focused on the development and validation of a numerical model based on the convection-diffusion-reaction (C-D-R) framework. The reaction framework for HC and CO oxidation were adapted from [57] and that for N O oxidation was as cited in [58]. The inclusion of this framework also implied that the diffusion of all gaseous species was automatically considered in the overall model, out of which the possible diffusion of N O2 against the direction of flow of exhaust gas could possibly be significant at certain operating conditions as shown in [53]. Such a case was illustrated in reference [53] as shown re-produced in Figure2.8 and2.9 for two different axial locations, one near filter entrance and one near filter exit, showing that the N O₂ concentration in the PM cake layer is affected by the diffusion of N O₂ produced in the PM cake layer, thus resulting in back-diffusion of N O₂. To be noted here is that this work assumed that the catalytic reactions (such as oxidation of N O to produce N O2) take place in the entire thickness of the substrate wall. This could only be valid for cases where the catalyst depth of penetration is 100% of the substrate wall thickness, which is an ideal case. Another publication by the same first author [59] also described the modeling of back-diffusion of N O₂ with the conclusions that the results are sensitive to the effective

diffusivities of N O2 in the cake and wall and that further work was required to determine the effective diffusivities experimentally.

Figure 2.8: N O₂concentration profiles at three different temperatures near filter entrance (z/L = 0.25) [53] - Reprinted with permission from SAE Paper No. 2004-01-0696 c 2004 SAE

International.

Figure 2.9: N O₂concentration profiles at three different temperatures near filter exit (z/L = 0.95) [53] - Reprinted with permission from SAE Paper No. 2004-01-0696 c 2004 SAE International.

The reference [51] advanced the modeling of oxidation of gaseous species such as HC, CO and N O by detailing the mechanism and reactions considered as well as inhibition factor parameters. Experimental studies that quantified the effect of back-diffusion of N O₂ on oxidation of PM were carried out at MTU [60,61]. The authors observed that the calculated ratio of N O₂ consumed by the stoichiometric oxidation reaction with PM (assuming a stoichiometric ratio of 7.7:1) to N O2available at the CPF inlet for the duration of some passive oxidation experiments was greater than one. This meant that the N O₂ available at the CPF inlet was insufficient to fully account for the N O₂ consumed by the oxidation reaction with PM. This calculation also assumed that the entire reaction of PM was with N O₂and not thermal (O₂). In some experiments, this ratio was as high as 4.5[61].