Characterization and Analysis of Ceria-Coated Gasoline ...ardous particulate matter emissions from vehicles using gasoline direct ignition (GDI) engines. This paper describes the soot
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Characterization and Analysis of Ceria-Coated GasolineParticulate Filter
Harikesh Arunachalama, Gabriele Pozzatob, Mark A. Hoffmanc, Simona Onoria,∗
aDepartment of Energy Resources Engineering, Stanford University, Stanford, California 94305, USAbDipartimento di Elettronica, Informazione e Bioingegneria, Politecnico di Milano, 20133 Milano, Italy
cDepartment of Mechanical Engineering, Auburn University, Auburn, Alabama 36849, USA
Abstract
Gasoline particulate filters (GPFs) are practically adoptable devices to mitigate haz-
ardous particulate matter emissions from vehicles using gasoline direct ignition (GDI)
engines. This paper describes the soot accumulation and regeneration experiments con-
ducted on a ceria-coated GPF installed downstream of a three-way catalytic converter
in a vehicle operating a GDI engine. Using the geometric design parameters of the
coated GPF, the total volume of cordierite and the total trapping volume of exhaust gas
in the GPF were calculated. The measured pre-GPF air-fuel ratio was used to determine
the volume fraction of the exhaust gas constituents. Oxygen density and the specific
heat of the exhaust gas were obtained as a function of temperature using the computed
volume fractions. Finally, the amount of soot mass oxidized during a regeneration
event was evaluated using the measured parts per million levels of pre- and post-GPF
CO2 gas. These parameters are essential to characterize the dynamic performance of
a GPF. Data acquired from experiments, and the aforementioned parameters serve as a
foundation for the development of mathematical models for virtual sensor deployment
and assessment of GPF performance across different initial soot loading and operating
temperature conditions.
Keywords: gasoline direct injection, gasoline particulate filter, catalytic washcoat,
Figure 1: (a) Side view of the sectioned coated GPF, and (b) axially parallel channels which are alternativelyplugged at each end with cordierite. A two-dimensional view of an inlet channel, porous wall, and an outletchannel of the GPF is shown here.
which can negatively impact engine performance and fuel economy. To minimize this
negative impact, the soot trapped in the GPF must be periodically removed. This is
accomplished via regeneration, i.e. oxidation of soot at elevated temperatures in the
presence of oxygen [28].
The structure and internal design of the coated GPF used in the experimental cam-
paign of this work is presented in Fig. 1. The washcoat material is primarily composed
of ceria (CeO2). Precious metals are loaded within the cerium to provide catalytic
reaction benefits, while the cerium provides the scaffolding for the precious metals and
oxygen storage ability that enhance soot oxidation reactions inside the GPF.
3. Experimental Characterization of a Ceria-Coated GPF
Experimental work, data acquisition, and data analysis associated with the soot
accumulation and soot regeneration stages were conducted at the Chassis Dynamome-
ter laboratory located at the Clemson University International Center for Automotive
Research.1
1The authors of this manuscript were affiliated with the Department of Automotive Engineering, ClemsonUniversity, Greenville, SC 29607, USA, when the experimental work and model development studies wereundertaken.
7
Preprints (www.preprints.org) | NOT PEER-REVIEWED | Posted: 28 May 2018 doi:10.20944/preprints201805.0397.v1
Figure 3: (a) Isometric view showing the front and sectioned interior of a coated GPF. The cross-sectionshown here is assumed to be semi-circular, and the frontal view represents the maximum number of channelsfrom top to bottom. Alternating channels and plugs span the entire frontal view, as shown in (b). All channelsand plugs have a thickness of (hchannel + hwall). (c) represents the cross-section of a single channel anda single plug.
semi-cylinders, Nch is expressed in terms of Ncross as:
Nch = 2 ·{Ncross ·
Ncross2·
0.5× π4D
2
D × D2
}⇒ Ncross =
√4×Nch
π≈ 80
(1)
The substrate diameter, D, is expressed in terms of Ncross and (hchannel + hwall)
as:
D = Ncross · (hchannel + hwall)
⇒ hwall =D
Ncross− hchannel = 0.215 [mm].
(2)
Figure 4 (a) represents a single inlet/outlet channel pair sectioned in the axial di-
rection. Figure 4 (b) illustrates a three dimensional view of a single inlet channel. Note
that the outlet channel is a mirror image of the inlet channel. The walls have a porosity
fraction of φwall, such that the volume fraction of the cordierite in the wall is equal to
10
Preprints (www.preprints.org) | NOT PEER-REVIEWED | Posted: 28 May 2018 doi:10.20944/preprints201805.0397.v1
Figure 4: (a) Two-dimensional view of an inlet channel, porous wall, and an outlet channel of the coatedGPF, and (b) three-dimensional view of the inlet channel incorporating the cordierite plug at the downstreamend.
4.2. Calculation of the volume fraction of the exhaust gas constituents
The volume fraction of species i in the exhaust gas is the ratio of the volume Vi that
it occupies and the total volume of all the exhaust gas species:
Yi =Vi∑j Vj
(5)
If the exhaust gas constituents follow ideal gas behavior, they satisfy the ideal gas
equation [29]:
P · Vi = ni ·R · Tgas, (6)
where ni represents the number of moles of exhaust gas constituent i, P and Tgas
represent the pressure and temperature of the exhaust gas, and R is the universal gas
constant. The ideal gas equation satisfied by the exhaust gas is:
P ·∑j
Vj =∑j
nj ·R · Tgas (7)
Using (6) and (7), Yi is expressed in terms of the total number of moles of the exhaust
gas constituents, ntotal:
Yi =ni
ntotal(8)
Under nominal operating conditions, the volume fraction of oxygen, YO2, is set
12
Preprints (www.preprints.org) | NOT PEER-REVIEWED | Posted: 28 May 2018 doi:10.20944/preprints201805.0397.v1
Figure 5: (a) Volume fraction of O2 and CO2 in the exhaust gas, and (b) Volume fraction of N2 and H2Oin the exhaust gas, during the course of the regeneration event.
(a) (b)
Regeneration Event
Regeneration Event
Figure 6: (a) Exhaust gas temperature at the GPF inlet, and (b) variation in O2 gas density with time as afunction of the exhaust gas GPF inlet temperature.
4.3. Calculation of ρO2and Cp,gas
ρO2and Cp,gas are exhaust gas properties that are dependent on the species con-
centration. ρO2 is determined as a function of Tinlet using the ideal gas equation:
PO2· VO2
= nO2·R · Tinlet (12)
The mass of oxygen trapped inside the coated GPF, mO2, is mathematically expressed
as ρO2· YO2
· Vexh. Converting VO2and nO2
in terms of mass and density terms, the
14
Preprints (www.preprints.org) | NOT PEER-REVIEWED | Posted: 28 May 2018 doi:10.20944/preprints201805.0397.v1
Figure 7: (a) Pre- and post-GPF air-fuel ratio, and (b) variation in the specific heat capacity of the exhaustgas, over the course of the regeneration event.
above equation is reformulated as:
PO2·(mO2
ρO2
)=
(mO2
MO2
)·R · Tinlet (13)
The density of oxygen, ρO2, is then expressed as:
ρO2=PO2·MO2
R · Tinlet(14)
ρO2is evaluated using the exhaust gas temperature at the GPF inlet. Exhaust gas
pressure at this location is assumed to be equal to atmospheric pressure. Figure 6
(a) presents the exhaust gas temperature profile at the GPF inlet during a regenera-
tion event. Corresponding to this temperature profile, the dynamic variation of ρO2 is
presented in Fig. 6 (b).
Cp,gas is determined as a function of Tinlet using the volume fraction of each con-
Figure 8: (a) Raw measurements of pre and post-GPF CO2 ppm levels with the observation windows high-lighted, (b) magnified view of observation window 1, and (c) magnified view of observation window 2.
where Cp,N2(Tinlet), Cp,O2
(Tinlet), Cp,CO2(Tinlet), and Cp,H2O(Tinlet) are specific
heat capacities of the individual exhaust gas constituents. Their values as a function of
temperature are provided in the NIST-JANAF tables [32].
Figure 7 illustrates a representative variation in exhaust gas heat capacity during a
regeneration event. The decrease in the overall specific heat capacity is due to a de-
crease in the volume fraction of triatomic molecules, CO2 andH2O, whose more com-
plex molecular structures have a greater number of vibrational and rotational modes to
absorb energy than the simpler, diatomic molecular structures of N2 and O2.
4.4. Calculation of the soot mass oxidized, mc,exp, during regeneration
The experimental measurements of CO2 ppm levels are processed to account for
time delays associated with exhaust gas transport through the GPF and the FTIR ana-
lyzer. The following sequence of steps are implemented before determining the amount
of soot oxidized during a regeneration event:
1. Temporally shift the post-GPF CO2 ppm data to align with the pre-GPF CO2
ppm data. This shift eliminates the transport delay associated with exhaust gas
flow through the GPF.
16
Preprints (www.preprints.org) | NOT PEER-REVIEWED | Posted: 28 May 2018 doi:10.20944/preprints201805.0397.v1
Figure 9: (a) Pre and post-GPF CO2 ppm levels before shifting (original data), and (b) CO2 ppm levelmeasurements after shifting the post-GPF CO2 ppm by 0.8 s.
2. Temporally shift pre- and post-GPF CO2 ppm to correlate the measured data
with the onset of a regeneration event. This shift eliminates the transport delay
associated with exhaust gas flow through transport lines to the FTIR analyzer.
Since the transport lines for both pre- and post-GPF ppm measurements are of
the same length and both analyzers use the same flow rate, both pre and post-GPF
CO2 measured data are shifted by the same amount. After employing this shift, the
FTIR species concentration data more accurately aligns with the mass flow and lambda
signals.
Using the raw measured CO2 ppm data, two time windows were observed to un-
derstand the magnitude of the time shift required. This is illustrated in Fig. 8. During
nominal engine operation, the pre- and post-GPF CO2 ppm measurements must be
nearly equal. The time instant at which the pre and post-GPF CO2 ppm levels reached
a peak/trough were analyzed. The post-GPF measurements achieved their correspond-
ing peak/trough value with a 0.8 s time delay with respect to their pre-GPF counter-
parts. Hence, this value was chosen to perform the time shift in step 1. The CO2 ppm
levels after this implementation are shown in Fig. 9.
To perform step 2, the start and end time of the regeneration event must be first
identified. The start time, ts, is chosen as the first time instance when the post-GPF
CO2 ppm level, CO2,out, exceeds the pre-GPFCO2 ppm level, CO2,in. The end time,
17
Preprints (www.preprints.org) | NOT PEER-REVIEWED | Posted: 28 May 2018 doi:10.20944/preprints201805.0397.v1