Soot Combustion Dynamics in a Planar Diesel ... - ACS Publications

Feb 25, 2009 - The dynamic features of soot combustion on a single layer, planar diesel particulate filter (DPF) were studied using IR imaging. At a f...
0 downloads 0 Views 1MB Size
Download PDF
Ind. Eng. Chem. Res. 2009, 48, 3323–3330

3323

Soot Combustion Dynamics in a Planar Diesel Particulate Filter K. Chen,* K. S. Martirosyan, and D. Luss Department of Chemical and Biomolecular Engineering, UniVersity of Houston, Houston, Texas 77204

The dynamic features of soot combustion on a single layer, planar diesel particulate filter (DPF) were studied using IR imaging. At a feed temperature of 635 °C, the soot combustion rate was uniform all over the surface at low oxygen concentrations. At higher oxygen concentrations, local ignition occurred at either one or several locations. The maximum temperature rise of the moving fronts (>100 °C) was much higher than those attained during uniform combustion. The temperature fronts bounding an ignited zone propagated on the surface. Their peak temperature and velocity changed as they moved on the surface. The maximum temperature of a downstream moving front exceeded that of the corresponding one moving upstream. At a soot loading of 10 g/L, a hot zone formed close to the end of the DPF and the bounding temperature front propagated upstream until it conquered the whole surface. At a soot loading of 20 g/L, the position and number of the hot zones strongly depended on the oxygen concentration. In general, increasing either the oxygen concentration or the soot load increased the moving front temperature and velocity. The flow rate affected the location of the ignition point for soot loading of 10 g/L but not for 20 g/L. Introduction Automobile diesel engines have several important advantages over gasoline engines. These include better mileage, less frequent fill-ups, more robust with an average life of 300,000 miles and produce less carbon monoxide (CO). One of the main disadvantages of diesel engines is the emission of particulate matter (PM), which is a health hazard that must be removed. Currently, the most efficient device for PM removal from the engine effluents is a Diesel Particulate Filter (DPF).1 It consists of thousands of square parallel channels, with the opposite ends of adjacent channels being plugged. The exhaust gas passes through the filter porous walls into the adjacent channels, while more than 95% of the PM accumulates on the filter.2-4 The pressure drop across the filter increases with the mass of the accumulated PM, affecting the engine performance and fuel efficiency.5 Thus, the DPF is regenerated periodically to remove the accumulated PM by combustion. The exothermic combustion leads in some cases to excessive local temperature rise, which may melt the ceramic filter and destroy the DPF.6-8 To circumvent this DPF failure, it is essential to be able to predict the impact of the operating conditions and regeneration procedure on the formation of excessively high local temperatures. The PM combustion temperature rise during the DPF regeneration has been investigated both experimentally and theoretically. It is well-known that the regeneration by a moving reaction front is faster than via uniform combustion.9 Following the pioneering studies of Bissett,10,11 many follow-up simulations were published.12-22 The simulations predict that the ignition location shifts upon a change in feed temperature. Measurements of the local temperature rise by thermocouples inserted into the DPF channels were reported.23-27 The experiments revealed a sharp temperature rise during the PM regeneration. Dynamic visualization of the DPF surface enables determination of the spatiotemporal features of the combustion fronts. Herz and Sinkevitch28 used a video camera to study the combustion of soot deposits on a 3 cm diameter, single DPF layer, and on a DPF that has been cut to expose a flat surface for viewing. They reported that the presence of adsorbed hydrocarbons affected the ignition and temperature rise during * E-mail address: [email protected].

PM regeneration. Hanamura and Tanaka29,30 followed the soot trapping and combustion in a half-cylindrical wall-flow DPF using microscopic visualization. They noted that the PM combustion could proceed along the DDF channels either by a propagating or uniform combustion. Decreasing the inlet gas temperature shifted the ignition location from the re-entrance to the exit of the DPF. Gallant et al.31 developed a novel experimental system in which an IR and optical video cameras followed the oxidation of soot on a single channel filter assembly. Most previous experiments were carried out using a diesel engine exhaust during the regeneration. Unfortunately, under the dynamic operation of a diesel engine, some of the key operating conditions, such as the exhaust temperature, oxygen concentration, soot loading, and flow rate are not constant. Martirosyan et al.32 reported some preliminary IR imaging of the spatiotemporal temperature on a single channel planar DPF under controlled feed conditions. They observed a dependence of the combustion mode and number and locations of the ignition points on the controlled operating conditions. We report here the results of an extensive set of experiments aimed at determining the impact of fixed operating conditions, such as oxygen concentration, soot loading, and flow rate on the spatiotemporal temperature during the catalytic soot combustion in a single channel planar DPF. The experimental results revealed some rather surprising novel features about the evolution and dynamics of the moving temperature fronts. Experimental System and Procedure Infrared (IR) imaging was used to measure the spatiotemporal temperature during the soot combustion in a mixture of air and nitrogen on a planar catalytic single layer DPF (60 mm long, 20 mm wide), cut from a commercial DPF (NKG Insulator). Both exterior sides and bottom of the single layer DPF were sealed by ceramic glue to force all the flow through the soot layer deposited uniformly on the top of the DPF. The planar DPF was held in a stainless steel (316 L) insulated reactor (120 × 40 × 40 mm) having an IR transparent quartz window on its top. It enabled visual and infrared imaging of the

10.1021/ie8015544 CCC: $40.75  2009 American Chemical Society Published on Web 02/25/2009

3324 Ind. Eng. Chem. Res., Vol. 48, No. 7, 2009

Figure 1. (a) A schematic of the reactor. (b) A schematic of the experimental system.

temperature field on the planar DPF, located at about 3 mm from the window. Figure 1a is a schematic of the reactor holding the filter. Experiments revealed that in our system it was impossible to obtain uniform reproducible PM deposition by the exhaust from a diesel engine. The PM regeneration in a DPF is rather sensitive to the PM deposition conditions and structure.33,34 Thus, in order to have a reproducible regeneration, we used carbon black nanoparticles to simulate the PM deposit in all of the experiments reported herein. The carbon black was deposited on the top of the planar DPF using the spray of a well-mixed solution of alcohol containing carbon black nanoparticles (40 nm, Sigma Aldridge). We later used a stream of nitrogen to remove the alcohol. To check the uniformity of the deposition, each experiment was repeated at least five times. The experiments reported here were conducted using soot loadings of either 10 or 20 g/L. Preliminary experiments showed that heat loss from the reactor wall affected the temperature profiles and the location at which the ignition occurred. To minimize the impact of these heat losses, the sides and bottom of the reactor were heated by an electrical ceramic fiber heater (Watlow Instruments Co, Inc.). The wall temperature was set by a power controller (Staco energy product Co, 2PF1010). Four K-type thermocouples (diameter ≈ 0.5 mm) in the reactor monitored the temperature around the DPF. The first one, placed near the entrance, measured the inlet gas temperature. The second was located 10 mm downstream from the inlet, the third in the reactor middle, and the fourth 10 mm ahead of the outlet. The thermocouples were placed at the bottom of the single layer DPF as placing them on the top would have obscured the IR image. The thermocouples signals were recorded and processed by an Omega data acquisition board connected to a PC. A 5-mm thick mineral wool insulation was placed between the planar DPF and the reactor walls to minimize the conduction, convection, and radiative heat exchange with the reactor walls.

A schematic of the experimental system is shown in Figure 1b. The spatiotemporal temperature was measured by a high speed (up to 60 frame/s) infrared camera (Merlin, MW18, Indigo Systems) held 50 cm above the quartz window. It has a 256 × 256 indium antimonide detector array sensitive to 3-5 µm radiations. The IR signals were recorded on a PC using the ImageDesk II software. The magnitude, shape, and motion of the thermal front on the DPF were determined by image analysis software (ThermaGRAM). The IR signals were calibrated using the K-type thermocouples readings before each experiment. Typical soot combustion lasted a few minutes. The IR images of the soot combustion were recorded at the rate of 10 per second. Mixtures of pressurized nitrogen and air containing 5-15 v.% oxygen were fed to the reactor during soot regeneration. Gas flow rates were measured by rotameters (Fisher & Porter, 10A6100), which were calibrated periodically to ensure accurate flow measurements. The feed pressure was controlled by downstream pressure regulators. Compressed nitrogen was used to purge the air in the reactor before each experiment. After that, the reactor wall’s temperature was stepwise increased from room temperature to the preset temperature. The reactor was preheated to the desired experimental temperature by an electric heater (Hoskins furnace, FD303), while the reactor was fed by pure nitrogen. After the desired temperature was reached, the feed was switched to the air-nitrogen mixture. Each experiment was repeated at least five times to check for reproducibility. Experimental Results The impact of oxygen concentration (range of 5 to 15 v.%.) was studied at two different soot loadings, at a fixed gas flow rate of 12 cm3/(s · cm2) through the filter and a feed temperature of 635 °C. Thermal imaging revealed that three different modes of combustion were obtained depending on the oxygen con-

Ind. Eng. Chem. Res., Vol. 48, No. 7, 2009 3325

Figure 3. Dependence of the maximum temperature rise on the oxygen concentration for a soot loading of 10 g/L. The sequence of IR images corresponding to cases a, b, and c are shown in Figure 2.

Figure 2. Thermal images of soot regeneration for soot loading of 10 g/L. Feed oxygen concentration (a) 5; (b) 10; and (c) 15 v.%.

centration. Figure 2 shows IR images obtained for a soot loading of 10 g/L. For reactants mixtures containing no more than 7.5 v.% oxygen, the temperature rise during soot regeneration was essentially uniform all over the DPF surface and no moving temperature fronts formed. A typical case for a feed containing 5 v.% oxygen is shown in Figure 2a. The variation of temperature among different surface points was very small (∼5 °C). The deposited soot was uniformly consumed all over the DPF surface 22 min after the start of the experiment. The uniformity of the surface temperature during the low oxygen concentration experiments indicates that pattern formation was not due to nonuniform cordierite permeability. The reproducibility of at least five repeated experiments indicates that the carbon black was deposited uniformly. For reactant feeds containing more than 7.5 v.% oxygen the combustion was initiated at one or more ignited locations, bounded by sharp temperature fronts. The fronts propagated on the surface, igniting the soot on the surrounding colder regions. The motion of the temperature fronts became more complex as

the oxygen concentration increased. Using a feed with 10 v.% of oxygen the combustion was initiated at a single ignited location next to the end of the DPF. The hot zone was separated from the surrounding colder region by a sharp temperature front. It moved from the downstream to the upstream of the DPF, eventually conquering the whole surface (Figure 2b). The high front temperature increased the local reaction rate, leading to a complete combustion of the soot. The peak front temperature was not constant during the propagation of the front on the surface and varied in the range of 745 to 770 °C and its velocity varied from 0.18 to 0.24 cm/s. After the local soot layer was consumed, the surface temperature decreased. The cooled regenerated locations appear as green regions in Figure 2b. An upstream moving temperature front was observed by Montierth35 in experiments using fuel with regeneration assisted additives. It was also found in numerical simulations.16 A more intricate combustion mode was observed when the feed contained 15 v. % oxygen (Figure 2c). Initially, a single point ignited close to the end of the DPF. The temperature front bounding that hot zone propagated upstream, i.e., in a direction opposite to the feed flow rate. However, after some time (44 s in the case shown here) ignition occurred at a location 20 mm downstream from the inlet. One of the two temperature fronts bounding this ignited zone propagated upstream and the second downstream. The downstream moving front, with a temperature of 720 °C and velocity of 0.27 cm/s, eventually collided with one originating from the end of the PDF. The coalescence of the two temperature fronts generated one moving upstream front with a temperature of 770 °C and a velocity of 0.35 cm/s. The maximum temperature of the moving fronts increased monotonically as the oxygen concentration was increased. Figure 3 describes this dependence for a soot loading of 10 g/L. A large increase in the temperature rise occurred at the transition from uniform combustion to one by a propagating front. The maximum temperature rise was not a unique function of the oxygen concentration and it varied with the position of the moving front (720-770 °C). In general, temperature fronts moving downstream from a hot zone were higher than those moving from the same point upstream. The values in Figure 3 are the maximum observed during any experiment. Repeated experiments confirmed that these values were reproducible. As the IR images in Figure 2 show, the oxygen concentration can affect the number of ignited points and the fashion in which

3326 Ind. Eng. Chem. Res., Vol. 48, No. 7, 2009

Figure 4. Moving temperature profiles originating from end of DPF. Soot loading of 10 g/L and O2 ) 10 v.% (same experiment as that in Figure 2b).

Figure 5. Coalescence of temperature fronts originating from two ignition points. Soot loading of 10 g/L and O2 ) 15 v.% (same experiment as that in Figure 2c).

the propagating reaction fronts move. For example, while in case b ignition occurred at one point, two ignited regions formed in case c. It is difficult to discern the dynamic features of the moving temperature fronts from the few thermal images shown in Figure 2. Thus, we show in Figures 4 and 5 the detailed axial profiles of the moving temperature fronts on the DPF. Figure 4 describes some temperature fronts originating from a single ignition point next to the end of the DPF. Some IR images corresponding to that experiment are shown in Figure 2b. About 30 s after the reactants mixture was fed to the reactor, ignition occurred close to the end of the DPF. The ignited zone was separated by a sharp temperature front from the colder part of the DPF. It moved in the upstream direction. For about the first 15 s, the amplitude of the moving temperature front was rather constant (variation of less than 5 °C). As the ignited zone moved upstream, its width and amplitude decreased. For example, the difference in the peak temperature between the profile observed at 45 and 55 s was ∼20 °C. As the temperature front moved

Figure 6. Effect of oxygen concentration on the relation between maximum front temperature and its propagation velocity. Soot loading of 10 g/L.

further upstream, the amplitude of the peak temperature and front velocity decreased. The IR images in Figure 2c are of an experiment in which temperature fronts originating from two ignition points eventually coalesced. Figure 5 describes the events leading to this coalescence. The ignition in this case occurred initially next to the planar DPF end at t ≈ 40 s, where t denotes the time after the reactant stream was fed to the rector. The temperature front bounding this hot region propagated upstream. At t ) 44 s, a new ignition zone formed at about 2 cm from the planar DPF inlet. The temperature of this second ignited zone increased with time. The temperature fronts bounding this hot zone moved in opposite directions, one upstream and the second downstream. At t ≈ 50 s, the upstream temperature front, which originated from the end of the DPF, collided with the downstream moving one that originated from the second ignited zone. This caused the two hot zones to coalesce into one. The temperature of this hot zone exceeded by a few degrees that of either one of the two original hot zones. The peak temperatures of the moving fronts and their velocities changed as they moved on the DPF. In general, when the ignition occurred next to the DPF end the peak temperature and velocity decreased as the front moved upstream. When a hot zone formed in the middle of the DPF, the front moving downstream had a higher peak temperature and velocity than that moving upstream. The increase or decrease of the peak front temperature led to corresponding change in the front velocity. Figure 6 describes the relation between the peak front temperature and front velocity at several oxygen concentrations. The data suggest a linear dependence. The slope for feed mixtures containing 15 v.% oxygen is steeper than those containing a lower oxygen concentration. To check the impact of the soot loading, we conducted a series of experiments at a soot loading of 20 g/L, which was twice that of the loading in the previous set of experiments. We used the same feed temperature (635 °C) and reactant velocity in the DPF channels (12 cm3/(s · cm2)) as in the previous set. We found that the number of ignition points and spatiotemporal temperature patterns were affected by the change in the loading. Increasing the soot load caused the dynamic features of the spatiotemporal temperature patterns to be richer and more complex than those at a lower soot

Ind. Eng. Chem. Res., Vol. 48, No. 7, 2009 3327

Figure 7. IR images of soot combustion for soot loading of 20 g/L and oxygen concentration of (a) 7.5; (b) 10; (c) 12.5; and (d) 15 v. %.

loading. Figure 7 shows thermal images of the DPF surface during combustion of the soot at four oxygen concentrations. At low soot loading (lower than 5 v.%), the combustion was uniform all over the surface, and this case is not shown here. At higher oxygen concentrations, the combustion was initiated at some location from which moving temperature fronts originated. Figure 7a shows IR images for an experiment with an oxygen concentration of 7.5 v.%. About 13 s after introduction of the reactant mixture, ignition occurred at two locations, one close to the upstream section of the DPF, the second at about its center. The downstream front propagating from the first ignition zone and the upstream front originating from the second ignited zone collided. The collision led to a higher local temperature (yellow region in 7.a, t ) 24 s) than that of the original fronts. The higher temperature increased the local reaction rate. Thus, by t ) 35 s, all of the soot in that region was consumed, and it cooled down (green region). The soot combustion and cooling in the center of the large hot region in the downstream of the DPF led to its splitting into two disjoint hot (red) regions at t ) 45 s. Thus, while the soot combustion originated in this case from two ignited zones, the image at t ) 45 s may mislead one to assume that it originated at three ignition regions. The soot combustion mode at higher oxygen concentrations was rather different from that at oxygen concentration of 7.5 v.%. Ignition occurred only at one point for oxygen concentrations in the range of 10-15 v.%, but its location was a sensitive function of the oxygen concentration. At an oxygen concentration of 10 v.% (Figure 7b) the ignition occurred at the upstream section of the DPF. As the bounding temperature front moved downstream, it conquered the whole surface, leading to complete soot combustion. The front temperature increased as it moved downstream. For reactants mixture of 12.5 v.% oxygen the ignition occurred at the downstream section of the DPF (Figure 7c). The temperature front bounding this ignition zone moved in the upstream

Figure 8. Dependence of the maximum temperature rise on the oxygen concentration for soot loading of 20 g/L. The sequence of IR images corresponding to cases b, c, and d are shown in Figure 7. At a, uniform combustion occurs.

direction until it conquered the whole surface. For 15 v.% oxygen (Figure 7d) the ignition occurred at the center of the planar DPF. The two bounding temperature fronts moved in opposite directions. The temperature and velocity of the front moving downstream exceeded those of the one moving upstream. A comparison of Figures 2 and 7 shows that the change in the soot loading affected the combustion modes and the dynamic of the temperature fronts. The dependence of the maximum front temperature on the oxygen concentration at a soot loading of 20 g/L is shown in Figure 8. The IR images corresponding to cases b, c, and d are shown in Figure 7. As in the case of the 10 g/L (Figure 3), the maximum temperature

3328 Ind. Eng. Chem. Res., Vol. 48, No. 7, 2009

8.5 cm3/(s · cm2)). Further increase in the flow rate led to a monotonic increase of the temperature rise. The flow rates at which the local maximum and minimum were obtained were rather insensitive to the values of either the oxygen concentration or the soot load. For a soot loading of 10 g/L the ignition point location was insensitive to changes in the oxygen concentration but sensitive to changes in the flow rate. The ignition point was near the entrance for flow rates up to 5 cm3/(s · cm2). Further increase in the flow rate shifted the ignition point to the center of the DPF. The ignition point was at the end of the DPF for flow rates exceeding 8.5 cm3/ (s · cm2). For a soot loading of 20 g/L the ignition point location was affected by the oxygen concentration (Figure 7) but not by the flow rate for all the oxygen concentrations and flow rates covered by these experiments. Discussion and Concluding Remarks Figure 9. Effect of oxygen concentration on the maximum temperature of upstream and downstream moving fronts. Loading of 20 g/L.

Figure 10. Impact of filter wall flow rate on maximum front temperature at several O2 concentrations and soot loading of 10 and 20 g/L. Feed temperature of 620 °C.

was a monotonic increasing function of the oxygen concentration. The moving combustion front mode led to a much higher peak temperature than the uniform combustion mode. The changes in the oxygen concentration affected the location of the ignition point and the direction in which the reaction fronts moved. Ignited or regenerated zone in the center of the planar DPF are bounded by temperature fronts that move in opposite directions. In general, the fronts moving downstream had a higher temperature and velocity than the one moving in the upstream direction. Figure 9 describes this dependence of the front velocity on the direction of motion for soot loading of 20 g/L. The difference in the velocities of the two fronts is largest at the higher oxygen concentrations. A change in the vehicle speed changes the exhaust flow rate and, hence, the flow rate per unit surface area of the DPF. We conducted some experiments to determine the impact of the flow rate at several oxygen concentrations and soot loadings of either 10 or 20 g/L. The temperature rise exhibited a moderate dependence on the flow rate (Figure 10). At low flow rates, the temperature rise increased as the flow rate was increased. It eventually reached a local maximum (at flow rate ∼7 cm3/(s · cm2) in these experiments). Further increase in the flow decreases the maximum temperature rise until a local minimum was reached (at about

IR measurements of the spatiotemporal temperature on a planar DPF provide important information about the special dynamic features that can be encountered during soot combustion in DPF and their sensitivity on the operating conditions. The main advantage of these experiments is that they provide information about the temporal temperature all over the surface. This information cannot be attained by experiments in which a few thermocouples are inserted in the DPF. The disadvantage of our experiments is that they provide only qualitative but not quantitative information about the maximum temperature rise and the temperature front dynamic features in a commercial DPF. The reason for this is that the heat loss from the planar DPF surface is much higher than that inside a DPF. Thus, the temperature rise in a DPF is expected to exceed that observed in our experiments. Moreover, experiments using a planar DPF cannot account for the impact of the radial heat conduction and wall heat loss on the temperature field in a DPF, which may lead to local hot regions. The experiments revealed some very intricate and unexpected features of the moving temperature fronts during soot combustion and their sensitive dependence on the operating conditions. The peak temperature rise of moving fronts is much higher than those during uniform combustion. Thus, predictions of the maximum temperature rise require one to know what mode of combustion exists during the regeneration. Radial gradient of temperature and nonuniform PM deposition may even lead to a situation in which different combustion modes may exist in different DPF channels. It is important to conduct extensive simulations and experiments to check the relations between the soot combustion observed on a planar DPF and that in a commercial one. Using a soot loading of 10 g/L ignition was usually initiated at the end of the DPF. However, with soot loading of 20 g/L, the ignition was initiated at various locations on the DPF, as shown in Figure 7. We do not have an explanation for this rather sensitive and reproducible dependence of the ignition location on the operation conditions. However, this sensitive dependence, which has not been observed or predicted before, is worth pointing out. Our, as well as many other, laboratory experiments are of cases in which the soot is uniformly deposited on the DPF. However, the PM deposition may not be uniform in a commercial DPF, nor may the PM combustion be complete and uniform at the end of the regeneration. This nonuniformity can affect the local reaction rate and corresponding rate of heat release, as well as the convective heat removal by the gas flowing through the filter. With time, the accumulation of

Ind. Eng. Chem. Res., Vol. 48, No. 7, 2009 3329

unconverted PM in some channels may lead to a buildup of the soot in these channels. This local increase in the PM load has two adverse effects. It can increase the local reaction rate and at the same time decrease the rate of convective cooling by the gas flowing through that PM deposit and at the same time increase the local reaction rate. Such a scenario may lead to excessive local temperature rise. It is important to gain an understanding and an ability to predict the impact of such nonuniformities on the maximal temperature rise and combustion mode. Extensive theoretical and experimental studies have been examined on the temperature rise of reaction fronts in packed bed reactors used to carry out conventional chemical reactions.36-38 Some of the qualitative features of the moving temperature fronts on a planar DPF are similar to those in catalytic packed bed reactors. In both cases, the temperature of the downstream moving fronts exceeds those of the upstream moving ones. However, in packed bed reactors, the front tends to attain an asymptotic constant temperature, shape, and velocity relatively fast. Similar qualitative features of moving reaction fronts are exhibited also during selfpropagating high temperature synthesis (SHS) processes.39 Our planar DPF experiments revealed some qualitative features that have not been reported to occur in packed bed reactors. For example, we observed formation of more than one ignited zone. This led to formation of temperature fronts moving in opposite directions that eventually collided, leading to a high temporary temperature rise. It is important to gain an understanding of what causes ignition at several locations and if and when it may occur in a commercial DPF. Figure 10 describes the dependence of the temperature rise and velocity of the moving reaction fronts on changes of one operating condition. In practice, several operating conditions may change simultaneously upon a change in the diesel engine load. For example, a shift from fast driving to idle leads to changes in the exhaust flow rate, oxygen concentration, and temperature. A rapid shift may lead in some cases to a transient temperature rise higher than those encountered under either the original or final stationary operating conditions. It is well-known that a rapid change in the feed to a packed bed reactor may lead to a transient temperature rise that largely exceeds that obtained under either the initial or final steady state. This is referred to as the “wrong way behavior”.40,41 We are currently studying these effects and will report our results in a future publication. Acknowledgment We wish to thank the NSF for financial support of this research and the NKG company for providing the DPF monolith. Literature Cited (1) Johnson, V. T. Diesel Emission Control in Reviews the Last 12 Months. SAE Tech. Pap. 2003, 2003-01-0039. (2) Howitt, J. S.; Montierth M. R. Cellular Ceramic Diesel Particulate Filter. SAE Tech. Pap. 1981, 810114. (3) Abthoff, J.; Schuster, H. D.; Langer H. J.; Loose G. The Regenerable Trap Oxidizer-an emission Control Technique for Diesel Engines. SAE Tech. Pap. 1985, 850015. (4) Adler, J. Ceramic Diesel Particulate Filters. Int. J. Appl. Ceram. Technol. 2005, 2 (6), 429. (5) Schmidt, N.; Root, T.; Wirojsakunchai, T.; Schroeder, E.; Kolodziej, C.; Foster, D. E.; Suga, T.; Kawai, T. Detailed Diesel Exhaust Particulate Characterization and DPF Regeneration behavior measurements for Two Different Regeneration Systems. SAE Tech. Pap. 2007, 2007-01-1063.

(6) Konstandopoulos, A. G.; Kostoglou, M.; Skaperdas, E.; Papaioannou, E.; Zarvalis, D.; Kladopoulou, E. Fundamental Studies of Diesel Particulate Filters: Transient Loading, Regeneration and Aging. SAE Tech. Pap. 2000, 2000-01-1016. (7) Koltsakis, G.; Haralampous, O.; Margaritis, N.; Samaras, Z.; Vogt, C. D.; Ohara, E.; Watanabe, Y.; Mizutani, T. 3-Dimensional Modeling of the Regeneration in SiC Particulate Filters. SAE Tech. Pap. 2005, 200501-0953. (8) Guo, Z. H.; Zhang, Z. Y. Hybrid Modeling and Simulation of Multidimensional Processes for Diesel Particulate Filter during Loading and Regeneration. Numer. Heat Transfer, Part A 2007, 51, 519. (9) Konstandopoulos, A. G.; Kostogloua, M. Reciprocating Flow Regeneration of Soot Filters. Combust. Flame 2000, 121 (3), 488–500. (10) Bissett, E. J. Mathematical Model of the Thermal Regeneration of a Wall-Flow Monolith Diesel Particulate Filter. Chem. Eng. Sci. 1984, 7/8 (39), 1233. (11) Bissett, E. J; Shadman, F. Thermal Regeneration of Diesel Particulate Monolithic Filters. AIChE J. 1985, 31 (5), 753. (12) Kobashi, K.; Hayashi, K.; Aoki, H.; Kurazono, K.; Fujimoto, M. Regeneration Capability of Diesel Particulate Filter System Using Electric Heater. SAE Tech Pap. 1993, 930365. (13) Optis, C. N.; Johnson, J. H. A 2-D Computational Model Describing the Flow and Filtration Characteristics of a Ceramic Diesel Particulate Trap. SAE Tech Pap. 1998, 980545. (14) Koltsakis, G. C.; Stamatelos, A. M. Modes of Catalytic Regeneration in Diesel Particulate Filters. Ind. Eng. Chem. Res. 1997, 36, 4155. (15) Miyairi, Y.; Miwa, S.; Abe, F.; Xu, Z.; Nakasuji, Y. Numerical Study on Forced Regeneration of Wall-flow Diesel Particulate Filters. SAE Tech. Pap. 2001, 2001-01-0912. (16) Konstandopoulos, A. G.; Kostoglou, M.; Housiada, P.; Vlachos, N.; Zarvalis, D. Multichannel Simulation of Soot Oxidation in Diesel Particulate Filters. SAE Tech. Pap. 2003, 2003-01-0839. (17) Stratakis, G. A.; Pontikakis, G. N.; Stamatelos, A. M. Experimental Validation of a Fuel Additive Assisted Regeneration Model in Silicon Carbide Diesel Filters. Proc. Inst. Mech. Eng., Part D 2004, 218, 729. (18) Sbrizzai, F.; Faraldi, P.; Soldati, A. Appraisal of Three-dimensional Numerical Simulation for Sub-micron Particle Deposition in a Micro-porous Ceramic Filter. Chem. Eng. Sci. 2005, 60, 23–6551. (19) Pontikakis, G.; Stamatelos, A. Three-Dimensional Catalytic Regeneration Modeling of SiC Diesel Particulate Filters. J. Eng. Gas Turbines Power 2006, 128, 421. (20) Dardiotis, C. K.; Haralampous, O. A.; Koltsakis, G. C. Catalytic Oxidation Performance of Wall-Flow versus Flow-Through Monoliths for Diesel Emissions Control. Ind. Eng. Chem. Res. 2006, 45, 3520–3530. (21) Frey, M.; Wenninger, G.; Krutzsch, B.; Koltsakis, G. C.; Haralampous, O. A.; Samaras, Z. C. 2D Simulation of the Regeneration Performance of a Catalyzed DPF for Heavy-duty Applications. Top. Catal. 2007, 42, 237–245. (22) Bogdanic, M.; Behrendt, F.; Mertins, F. The Influence of a 2-component Model on the Computed Regeneration Behavior of an Uncoated Diesel Particulate Filter. Chem. Eng. Sci. 2008, 63, 2601. (23) Jelles, S. J.; Makkee, M.; Moulijn, J. A.; Acres, G. J. K.; PeterHoblyn, J. D. Diesel Particulate Control Application of an Activated Particulate Trap in Combination with Fuel Additives at an Ultra Low Dose Rate. SAE Tech. Pap. 1999, 1999-01-0113. (24) Bhatia, G.; Gunasekaren, N. Heat-Up of Diesel Particulate Filter: 2D Continuum Modeling and Experimental Results. SAE Tech. Pap. 2003, 2003-01-0837. (25) Haralampous, O. A.; Koltsakis, G. C. Oxygen Diffusion Modeling in Diesel Particulate Filter Regeneration. AIChE J. 2004, 50 (9), 2008. (26) Belloir, M.; Sakushima, N.; Lahjaily, H. A CDF Study to Optimize the Injection Strategy for Diesel Particulate Filter Regeneration. SAE Tech. Pap. 2007, 2007-01-0164. (27) Haralampous, O. A.; Koltsakis, G. C.; Samaras, Z. C.; Kraemer, L.; Heimlich, F.; Behnk, K. Control Strategies for Peak Temperature Limitation in DPF Regeneration Supported by Validated Modeling. SAE Tech. Pap. 2007, 2007-01-1127. (28) Herz, R. K.; Sinkevitch, R. M. Reactors for Investigation of Soot Combustion on Filter Surfaces. Carbon 1986, 4 (24), 457–462. (29) Hanamura, K.; Suzuki, T.; Tanaka, T.; Miyairi, Y. Visualization of Combustion Phenomena in Regeneration of Diesel Particulate Filter. SAE Tech. Pap. 2003, 2003-01-0836. (30) Hanamura, K.; Tanaka, T. Visualization Study of Heterogeneous Reaction of Particulate Matter in Regeneration of DPF, Proceedings. JSAE Annual Congress 2005, 22, (5), 23. (31) Gallant, T. Experimental Diesel Particulate Filter Capabilities at PNNL, DEER Conference, Chicago, IL, August, 20-24, 2006. (32) Martirosyan, K. S.; Chen, K.; Luss, D. Behavior Features of Soot Combustion in Diesel Particulate Filter, Chem. Eng. Sci. (2009).

3330 Ind. Eng. Chem. Res., Vol. 48, No. 7, 2009 (33) Konstandopoulos A. G.; Skaperdas E.; Masoudi M.; Microstructural Properties of Soot Deposits in Diesel Particulate Traps, SAE Tech. Paper No. 2002-01-1015, 2002. (34) Konstandopoulos, A. G.; Kostoglou, M.; Vlachos, N; Kladopoulou, E. Advances in the science and technology of diesel particulate filter simulation. AdV. Chem. Eng. 2007, 33, 213–275. (35) Montierth M. R. Fuel Additive Effect Upon Diesel Particulate Filters, SAE, paper 840072, 1984. (36) Wicke, E.; Vortmeyer, D. Zundozonen heterogener reactionen in gas-durchstormten kornerschichten. Ber. Bunsen-Ges. Phys. Chem. 1959, 63, 145. (37) Rhee, H. K.; Foley, D.; Amundson, N. R. Creeping reaction zone in a catalytic fixed bedreactor: a cell model approach. Chem. Eng. Sci. 1974, 28, 607.

(38) Kiselev, O. V.; Matros, Yu. Sh. Propagation of the combustion front of a gas mixture in granular bed of catalyst. Combust., Explos., Shock WaVes 1980, 16, 152. (39) Merzhanov, A. G.; Rumanov, E. N. Physics of reaction waves. ReV. Mod. Phys. 1999, 71, 1173. (40) Crider, J. E.; Foss, A. S. Computational studies of transients in packed bed reactors. AIChE J. 1966, 12, 514. (41) Chen, Y.-C.; Luss, D. Wrong-way behavior of packed-bed reactors: Influence of interphase transport. AIChE J. 1989, 35, 1148.

ReceiVed for reView October 15, 2008 ReVised manuscript receiVed December 24, 2008 Accepted January 14, 2009 IE8015544