*3.4. Experimental Tests in the Engine Test Cell*

Both the soot loading and regeneration stages were performed under properly selected operating conditions in order to consider the influence of the exhaust flow rate on soot loading in terms of soot flow rate and soot cake properties. In particular, steady-state conditions were used to perform the soot loading, with a standard injection strategy consisting of pilot, pre- and main injections. In this phase, a constant value for the exhaust flow rate was used, aiming at monitoring the effective soot accumulation. In fact, two main terms are responsible for the pressure drop across the DPF:


During the filtration, the soot mass flow rate was calculated using the following formula:

$$
\dot{m}\_{\text{soot}} = \frac{\dot{m}\_{\text{exh}}}{\rho} \ast \mathbb{C}\_{\text{soot}} \tag{2}
$$

where . *mexh* and *ρ* are the mass flow rate (in kg/h) and density (in mg/m3) of the exhaust gas at the inlet of the DPF, while *Csoot* is the PM concentration in mg/m<sup>3</sup> measured by the smoke meter. During the experimental tests, the soot concentration across the filters was continuously monitored in order to estimate the DPF efficiency, in terms of PM, using the following equation:

$$\eta\_{fillration} = \frac{\mathcal{C}\_{\text{soot\\_in}} - \mathcal{C}\_{\text{soot\\_out}}}{\mathcal{C}\_{\text{soot\\_in}}} \ast 100\tag{3}$$

where *Csoot*\_*in* and *Csoot*\_*out* are the PM concentrations in mg/m3 upstream and downstream of the filters, respectively.

The measurement of torque, fuel consumption, exhaust flow rate, T and DP across the filter was carried out to compare the engine performance during the filtration stage. The PSD measurements were carried out upstream and downstream of the CDPF. The size-dependent filter removal efficiency during the accumulation was evaluated for each diameter using the following equation:

$$\eta\_{filration} = \frac{(Nparticles)\_{in} - (Nparticles)\_{out}}{(Nparticles)\_{in}} \ast 100\tag{4}$$

where the subscripts "*in*" and "*out*" indicate the number of particles before and after the filtration, respectively. Moreover, the PN emission evolution during the active regeneration was detected. This represents a critical issue in a future perspective; in fact, the regeneration procedure could be entailed in the type of approval cycle (i.e., RDE).

The experimental tests during the soot loading phase were performed under four different engine loads in the range of 22–36%, with an engine speed equal to 2140 rpm and standard ECU calibration. The main engine variables are reported in Table S4. Engine load conditions were selected to compare filter performance also when the passive regeneration occurred with the CDPF (e.g., loads equal to 33% and 36%).

Regarding the active regeneration, PSD was correlated with the time history of DP across the CDPF and inlet/outlet temperatures. According to the description reported in a previous work [43], the procedure was actuated when a soot loading of 5 g/L had been reached: this quantity, not far from that reported in the literature [55,56], was recommended by the filter manufacturer as a value that does not impact on engine performance. The comparison between the CDPF and the bare DPF is shown in Figure 8 in terms of normalized pressure drop and temperature during the regeneration phase in the same engine operating conditions. The beneficial effect of the catalyst deposited on the DPF is evident: the DP curve decreases, meaning that soot oxidation occurrence is needed at temperatures lower than 400 ◦C for the CDPF (Figure 8a) and about 600 ◦C for the bare DPF (Figure 8b). Regarding the contents of the following sections, the whole regeneration process of the CDPF was divided into three consecutive phases (Figure 8a):


**Figure 8.** Trend of pressure drop and temperature across the (**a**) CDPF and (**b**) DPF during the regeneration.

Particle Emission Measurements

In this section, the PSDs at the CDPF outlet are presented and discussed; the results are compared with those obtained for the DPF [43] to emphasize the differences in terms of particle emissions. The plots shown in Figures 9–12 correspond to the raw experimental PSD data measured in the engine tailpipe. Due to the complex setup adopted for the measurements and the intrinsic oscillations due to the typical wave propagation of the engine exhaust gas flow to be analyzed, the PSD data were expected to be noisy. However, in this case no digital filter to generate a smoother data distribution was applied to render this aspect for readers and better evaluate the quality of the data and the difficulty in obtaining them. Regarding the particle size determination, Figures 9–12 show the mobility diameter (Dm) in a log scale; therefore, the range of values mentioned in the text can be considered correct despite the data oscillation trend. Moreover, the PSDs are represented on the Y-axis in Figures 9, 10 and 12 as dN/dln (Dm), where N is the number density and Dm is the mobility diameter.

**Figure 9.** PSDs at the CDPF outlet for test case #1 during the accumulation (blue line) and just after the active regeneration ended (red line).

**Figure 10.** Comparison of outlet PSDs during accumulation for test cases #1 and #2 (**left side**) and for test cases #3 and #4 (**right side**).

**Figure 11.** PN removal efficiency of the CDPF for test cases #1 and #4.

Figure 9 shows the PN concentration at the CDPF outlet for test case #1. The PSD bimodality [43] was still observed, and, after a minimum around 20–30 nm, the particle number increased for larger diameters (blue line). Differently from what was recently measured by authors with a regular DPF in the same conditions [43], a remarkable difference between the empty and partially filled filter was observed; in particular, an appreciable increase in the range of 10–100 nm was clearly visible (red line). The latter is due to the high reactivity of the CPDF, which ensures complete filter emptying during the active regeneration. However, the slight PN increase disappears after a few minutes of accumulation.

Figure 10 shows a comparison of the catalytic and the bare filters. In Figure 10a, the PSDs for test cases #1 and #2 (low engine speed) are reported: the two filters exhibited very similar results in terms of both shape and concentration across the whole range investigated. On the other hand, in Figure 10b (test cases #3 and #4—high engine speed) the CDPF presents a slight increase in PN in both test cases. Focusing on test case #3, the peak of 5 nm particles is three times higher than that observed for the DPF (green and black lines, respectively). The trend was observed up to 40 nm, after which the difference between the PSDs measured with the two filters was negligible. Similar remarks could be made with reference to the comparison of the PSDs obtained in test case #4. The increase in the number of 5 nm particles confirmed that, in the CDPF, the passive regeneration occurs starting from 33% of engine load [21]. In fact, while for the standard filter the PN removal efficiency was comparable, for the CDPF at 36% of engine load a slight reduction was detected. This results in partial filter emptying, thus allowing a reduction in the frequency of active regeneration. It is worth remarking that the particle emissions detected were much lower

than those achieved during active regeneration; therefore, a reduction in both pollutants and fuel consumption can be achieved by operating the CDPF rather than the DPF.

The above results were also confirmed by the filtration efficiency estimation, as defined by Equation (3), shown in Figure 11.

In accordance with the DPF results, the highest PN removal was observed for test case #1, due to the lowest spatial velocity. In particular, the CDPF seemed to be very effective also for the sub-23 nm particles and exhibited a constant efficiency, below 40 nm, close to 99%. The narrow difference, compared with the DPF [43], is due to the lower mean pore diameter which affects the particle removal efficiency. On the other hand, in test case #4 the passive regeneration event leads to a slight decrease in efficiency for particle diameters below 10 nm. This phenomenon was not noticed for the standard DPF, which exhibited very close values for all the investigated conditions. For particles larger than 60 nm, the filtration efficiency began to decrease, and the CDPF exhibited similar behavior to the DPF.

Figure 12a shows the PSDs measured at the CDPF exit during the start-up of the regeneration. The dashed black line represents the PN distribution obtained by actuating the fuel after injection. The concentrations of particles with sizes below 10 nm were one order of magnitude lower than in the DPF and remained almost constant for the whole particle size range investigated; the PSDs are like those measured during the accumulation phase, as proof of the high reactivity of the catalyst. The benefits of the CDPF are even more striking comparing the PSDs when the fuel post-injection is actuated (blue and red lines). In fact, no significant increase in particle emissions was observed; moreover, the results highlight the absence of a correlation between the concentration peak of particles with sizes below 10 nm and the post-injection fuel quantity. In particular, the particle emissions were three orders of magnitude lower than the standard filter and remained two orders of magnitude lower for particle sizes larger than 50 nm. The slight increase in the concentration of particles in the range of 30–60 nm (red line) was probably due to the start of the soot oxidation, which resulted in the soot layer fragmentation responsible for a lower filtration efficiency towards the particles in this range.

Figure 12b shows the PSDs measured at the CDPF exit during the second phase of the regeneration process (200–450 s). The plots refer to different mean inlet temperatures to analyze the evolution of PSDs in the second stage of regeneration. Compared to the results obtained with the standard filter, the CDPF exhibited a bimodality of the PSDs also observed during the accumulation phase, with a peak of 5 nm particles equal to <sup>3</sup> × <sup>10</sup><sup>9</sup> #/cm3, three orders of magnitude lower than for the DPF. Very similar values were detected during this stage, which emphasizes the improved soot oxidation thanks to the catalytic activity. The reduction in PN concentration at the outlet of the CDPF was confirmed by the sharp increase in the outlet temperature: while for the standard filter it exceeded the inlet temperature after 380 s [43], for the catalytic filter, this occurred much earlier, after 220 s. Moreover, the temperature increase across the filter reached 100 ◦C, two times higher than that detected across the DPF. Given these considerations, it can be asserted that the local oxidation temperatures were much higher and that complete combustion of the particles was achieved.

Figure 12c shows the PSDs at the end of the regeneration, identified by the quasiconstant pressure drop across the CDPF and the decrease in the outlet temperature. The magenta and cyan lines refer to two consecutive measurements collected in a time window of one minute in the same inlet conditions. A PN peak of 2 × 1011 was detected for particles with sizes below 10 nm in the later measurement (cyan line), which is more than one order of magnitude higher than the earlier measurement (red line). Furthermore, an appreciable increase in the range of 20–40 nm was noticed, which yielded a PSD shape not observed in the earlier measurements. This behavior can be explained considering that, despite the same inlet temperature being used, the DPF core exhibited a higher temperature in the former case, thus promoting the oxidation of the particles with lower diameters. Furthermore, this result confirms the complete emptying of the CDPF, previously evidenced in Figure 9.

The dashed black line in Figure 12 shows the mean PSD during the whole regeneration process for the CDPF. The comparison with the corresponding values detected for the DPF, represented by the dashed red line, highlights the impressive reduction in particle emissions achieved by the CDPF in the whole range investigated. In particular, the peak of the 5 nm PN dropped from 4 × 1012 to 3 × 1010, two orders of magnitude lower, while for the particle size above 50 nm the reduction reached three orders of magnitude, from 1010 to 107.

This very important result highlighted that the use of a well-designed CDPF can allow the oxidation of accumulated soot at lower temperatures and decrease in particle emissions during the regeneration phase, not only with respect to a bare DPF, but also with respect to recent studies in the literature, in which increases in emissions of particles of sizes above and below 23 nm were observed [45,57].
