Quantitative Analysis of Ash Density and Ash Distribution inside DPF Honeycomb Channels Based on X-ray Computed Tomography 2019-01-0979

Simulation of soot and ash deposits in diesel particulate filters (DPF) often assumes uniform distributed cake-layer and/or plug accumulation at the very end section of the inlet channels, which may not reflect some conditions in the field. For example, cake-layer thickness changes along the filter length, and plugs show up not only at the end section but also at the middle section or even near the inlet section. This paper presents detailed microscopic analytical techniques which have been developed and applied to quantitatively derive the density and distribution of ash deposits inside DPF honeycomb monoliths. The ash loading experiments were done in a combined engine/burner test facility. Specifically, X-Ray tomography (X-Ray CT) was used extensively, which has the advantage of non-contaminating the ash deposits. A unique 2D and 3D data processing procedure was developed so that quantitative and statistical analyses could be done to extract ash layer/end-plug deposition information. With the technique, the quantitative ash distribution information will be useful as inputs to theoretical model for better and more accurate analysis and prediction. The key feature of the technique lies in analyzing the samples in multiple perspectives, to be described in the paper, in deriving the 3D information.

Catalyst coated cordierite DPF was used to collect lab generated soot and ash from a single cylinder diesel engine and oil-burner. DPF loading process continues until the pressure drop reaches 10 kPa at a space velocity of 40000/h, and regeneration process follows with duration of 1 hour. After cycles of loading and regeneration, the DPF filter was removed from the bench and scanned by the X-Ray CT system. On one hand, thousands of 2D images were sliced out from the DPF and analyzed with self-developed program with MATLAB, which can provide the ash packing density and distribution information of both cake-layer and end-plug ash accumulation. On the other hand, 3D DPF ash volume was post-processed to extract all of the ash plugs (end-channel or mid-channel) and their spatial distribution, radially and axially along the channels.

With the 2D image processing and comparison with standard powder samples, collected ash can be classified into a few different groups and measured separately; hence, distribution of ash with different densities and fractions of each density group can be characterized. For scanned 3D volumetric data, spatial distribution, plug length and volume can be summarized with the 3D measurements and analysis. Finally, information about the number of plugged channels and the plug ratio of each channel can be obtained as well.

With the specific image processing and statistical analysis, described in the paper, more details of the ash deposits can be examined. These results can be used as input for modeling work, generate accurate predictions for calibrations in DPF feedback control, and more importantly for solving the main problems that could potentially lead to DPF failure.