Article
`
`pubs.acs.org/IECR
`
`A Model Development for Evaluating Soot-NOx Interactions in a
`Blended 2‑Way Diesel Particulate Filter/Selective Catalytic Reduction
`‡
`‡
`†
`†,⊥,* Kushal Narayanaswamy,
`Soo-Youl Park,
`Steven J. Schmieg,
`and Christopher J. Rutland
`†
`Engine Research Center, University of Wisconsin-Madison, 1500 Engineering Drive, Madison, Wisconsin 53706, United States
`‡
`General Motors Global Research and Development, 30500 Mound Road, Warren, Michigan 48092, United States
`*S Supporting Information
`
`ABSTRACT: The 2-way diesel particulate filter/selective catalytic reduction (DPF/SCR) emission reduction system has been
`considered as a potential candidate for future emission standards owing to its advantages in cost savings and packaging flexibility.
`For the 2-way device, Cu−zeolite is coated inside the DPF substrate as nitrogen oxide (NOx) reducing (DeNOx) catalytic
`material. Therefore, when exhaust gas passes through the 2-way device, NOx reduction and soot filtration occur simultaneously.
`However, the operating characteristics of the combinatorial device might be different from individual DPF and SCR devices. In
`this work, a previously developed model was improved to include soot filtration and oxidation. The model has been tested and
`validated with experimental data from a reactor flow bench in a systematic manner and applied to capture the effect of soot
`deposits on NOx reduction performance in a 2-way DPF/SCR device. Accordingly, the soot oxidation characteristics of a 2-way
`device are investigated with various feed gas compositions. Then the effect of soot deposit on the SCR reaction is investigated in
`terms of deterioration of DeNOx performance and the interaction between soot oxidation reactions and DeNOx SCR reactions.
`
`1. INTRODUCTION
`As carbon dioxide (CO2) emissions regulation become more
`stringent, automakers are increasing their attention on
`technologies that decrease fuel consumption. For powertrain
`development,
`the hybridization of
`the power source and
`increasing efficiency in the internal combustion engine itself are
`viable technologies. Consequently, interest in diesel engines has
`increased because they have better fuel efficiency than gasoline
`engines.
`Heterogeneous combustion is one feature of diesel
`combustion. This makes locally favorable conditions for the
`formation of NOx and PM (particulate matter) inside the
`combustion chamber. Under stringent emission regulations,
`automakers are forced to decrease tailpipe emissions, and a
`viable solution is the use of emission aftertreatment systems.
`Representative technologies for emissions aftertreatment are
`(1) oxidation of carbon monoxide (CO), hydrocarbon (HC),
`and nitric oxide (NO) in a diesel oxidation catalyst (DOC), (2)
`NOx reduction in a lean NOx trap (LNT) or urea/ammonia
`selective catalytic reduction (SCR) catalyst, and (3) PM
`reduction by filtration in a diesel particulate filter (DPF).
`SCR is generally accepted as one of most promising
`alternative NOx reduction technologies. It has wider operating
`temperature range and higher conversion efficiency than other
`NOx reduction technologies. However, a disadvantage of SCR
`is it requires a storage tank of ammonia or urea and an injection
`system. Thus, it results in larger packaging requirements.
`A typical diesel aftertreatment system is composed of DOC,
`NOx reduction catalyst, and DPF placed in series in the exhaust
`system. In recent years, catalyst suppliers are introducing
`multifunctional devices for minimizing packaging problems. In
`particular, both soot filtration and NOx reduction by SCR are
`implemented in a 2-way DPF/SCR device by coating SCR
`
`catalyst on a DPF substrate. Figure 1 shows the schematic
`diagram for a blended 2-way DPF/SCR device.
`Integration of DPF and SCR functionalities into a single
`device might cause other problems which do not appear in
`separate devices. The operation of the 2-way device might be
`different
`from that of conventional DPF and SCR in
`accordance with different performance characteristics. He et
`al.1 tested 2-way DPF/SCR by installing it in a pickup truck.
`Cu-zeolite was used as catalytic material and test conditions
`were the US city (FTP) and highway (US06) drive cycles. NOx
`conversion efficiency across the 2-way DPF/SCR device was
`75% for FTP and 83% for US06 mode with 60 000 miles of
`engine aging. Also they compared filter regeneration perform-
`ance between the 2-way DPF/SCR and a conventional
`catalyzed DPF, and the results
`show that
`regeneration
`performance in the 2-way device is almost the same as that
`of a catalyzed DPF. However, CO release during the
`regeneration is much higher with the 2-way device.
`Lee et al.2 showed how NOx reduction performance is
`affected by filter regeneration over a 120 000 mile aged 2-way
`DPF/Cu−zeolite SCR catalyst device. They suggested that
`more ammonia should be injected to compensate for the
`amount of ammonia oxidation at filter regeneration temper-
`atures. However, no quantified results for this effect were
`presented.
`Fuel penalty during filter regeneration is another important
`aspect. To improve low temperature soot oxidation for
`minimizing fuel penalty, oxidation catalysts are coated on
`DPF substrates. Usually, noble metals are used as a catalytic
`
`Received: August 3, 2012
`Revised: October 7, 2012
`Accepted: November 12, 2012
`Published: November 12, 2012
`
`© 2012 American Chemical Society
`
`15582
`
`dx.doi.org/10.1021/ie3020796 | Ind. Eng. Chem. Res. 2012, 51, 15582−15592
`
`BASF-2011.001
`
`

`
`Industrial & Engineering Chemistry Research
`
`Article
`
`Figure 1. Schematic diagram of blended 2-way DPF/SCR.
`
`The coordinate system andcomputational domain for
`equations 1−6 are shown in Figure 2.
`4
`4
`a
`
`(
`
`C
`
`b,
`
`i
`
`I
`
`−
`
`C
`
`wf,
`
`|
`i y
`
`==
`)
`0
`
`0
`
`(1)
`
`cI
`k a
`
`(
`
`u C
`w b,
`
`i
`
`I
`
`)
`
`+
`
`I
`u C
`
`(
`
`b,
`
`i
`
`I
`)
`
`+
`
`∂∂
`
`x
`
`i
`
`wf,
`2
`
`+
`
`
`k S C(
`w w
`
`wf,
`
`i
`
`−
`
`C
`
`)
`
`ws,
`
`i
`
`=
`
`0
`
`RR
`
`i
`
`(2)
`
`(3)
`
`2
`∂
`=
`
`C y
`
`(
`
`u C
`w wf,
`
`i
`
`)
`
`−
`
`D
`w
`
`∂
`
`∂∂
`
`y
`
`
`k S C(
`w w
`
`wf,
`
`i
`
`−
`
`C
`
`)
`
`ws,
`
`i
`
`+
`
`I
`
`−
`
`C
`
`)
`
`0
`
`=
`
`|
`i y
`wf,
`∂
`C
`wf,
`∂
`y
`
`i
`
`=
`
`y
`
`0
`
`(4)
`
`b,
`
`i
`
`cI
`
`k C
`(
`
`I
`C u
`i
`b,
`
`w
`
`=
`
`C
`
`wf,
`
`|
`i y
`
`=
`
`· −
`u
`0 w
`
`D
`w
`
`O
`
`(
`
`C
`
`b,
`
`i
`
`−
`
`C
`
`wf,
`
`|
`i y
`
`)
`
`=
`
`w
`
`cO
`k a
`
`4
`
`O
`)
`
`b,
`
`i
`
`−
`
`4
`a
`
`(
`
`u C
`w wf,
`
`|
`=
`i y w
`
`)
`
`+
`
`(5)
`
`∂∂
`
`O
`u C
`(
`x
`=
`
`0
`
`∂
`C
`wf,
`∂
`y
`
`i
`
`(6)
`
`O
`)
`
`= −
`
`D
`w
`
`b,
`
`i
`
`−
`
`C
`
`|
`i y
`
`=
`
`0
`
`wf,
`
`cO
`
`
`
`k C(
`
`=
`y w
`O represent
`I Cwf,i Cws,i nd Cb,i
`the
`The symbols Cb,i
`concentration of species i measured at the inlet channel, gas
`stream inside the wall, solid surface inside the wall and outlet
`
`Figure 2. Discretization of computational domain and coordinate
`system.
`
`15583
`
`dx.doi.org/10.1021/ie3020796 | Ind. Eng. Chem. Res. 2012, 51, 15582−15592
`
`material for soot oxidation and numerous studies have been
`conducted to examine this effect. Song et al.3 measured soot
`oxidation with a Pt/CeO2 catalyst, and they showed that the
`oxidation temperature decreases by 130 °C compared to that of
`an uncatalyzed device.
`In this work, Cu−zeolite is used as a SCR catalytic material
`for NOx reduction. Moreover, limited studies have shown that
`copper is also a catalytic material for soot oxidation. Levin et
`al.4 studied how copper fuel additives affect soot oxidation.
`They showed that it expands vehicle operability by reducing
`particulate mass loading and related external energy con-
`sumption required during regeneration. Murphy et al.5 added
`metal chlorides dissolved in methanol
`to diesel soot and
`showed that the copper decreased the ignition temperature by
`about 100 °C.
`to
`this study is to develop a model
`The objective of
`understand the soot-NOx interaction in a 2-way DPF/SCR
`device. A previously developed model describing the transport
`and reaction phenomena inside a wall flow type substrate (Park
`et al.6) forms the basis of this study. This model incorporated
`SCR kinetics and was validated under
`temperature pro-
`grammed desorption tests and steady state NOx conversion
`tests. For
`soot filtration and oxidation, mathematical
`formulations based on previous research work by Bisset7 and
`Konstandopoulos8 is incorporated. In this paper the existing 2-
`way DPF/SCR model is expanded to include soot filtration and
`oxidation. For this purpose, a soot cake layer model is added
`and both NO2- and O2-based soot oxidation reactions are
`included. The model has been validated in a systematic manner
`with available experimental data to capture the effect of soot
`deposits on NOx reduction performance of the 2-way DPF/
`SCR device.
`
`2. MATHEMATICAL FORMULATION FOR 2-WAY
`DPF/SCR
`2.1. Species Transport Equations. The basic foundations
`for building a model for an aftertreatment device are transport
`equations for species, energy, and momentum. The governing
`equations for flow and temperature of a wall flow-type device
`are based on Bisset.7 They are composed of conservation of
`mass, momentum, and energy for the inlet and outlet channel.
`Darcy’s law for pressure drop across the porous media is used
`to relate the inlet and outlet channel pressures. Only single inlet
`and outlet channel are solved under the assumption that all
`channels are identical.
`The species transport equations for each species under a wall
`flow substrate were introduced by Park et al.6 These are shown
`in eqs 1−6, and they are applied to inlet channel (eqs 1 and 4),
`outlet channel (eq 5 and 6) and porous filter wall (eqs 2 and 3).
`
`BASF-2011.002
`
`

`
`Industrial & Engineering Chemistry Research
`
`Article
`
`there are 6n unknowns and 6n equations. A solver for the
`equation is designed by discretizing the differential equations of
`eq 1, 2, and 5. Equations 1 and 5 do not include the diffusion
`term because it is so slow as compared to the convective
`transport along the channel axial direction, but eq 2 includes
`the diffusion. Thus, eqs 1−6 should be solved simultaneously at
`a given location of x, and then the solution marches to the next
`channel axial position. As seen in the rate expression of “NO
`oxidation” and “Fast SCR reaction” in Supporting Information,
`Table S2, the nonlinear terms are included, so the finally
`discretized algebraic equations are nonlinear system equations.
`The solver was programmed by Fortran using an internal
`nonlinear system equation subroutine which is available in the
`International Math and Static Library (IMSL).
`2.2. Model
`for Soot Filtration and Oxidation. To
`describe the soot mass conservation inside the 2-way device,
`models for soot filtration and soot oxidation are needed. Soot
`filtration occurs in two different modes.8,13−16 First, at the
`initial stage of the filtration process, soot is deposited inside the
`filter wall, and this is called deep bed filtration. After filling up
`the substrate, soot deposition starts to occur on the filter wall
`and forms a soot cake layer. Konstandopoulos et al.8
`contributed much on the development of a model for soot
`filtration and oxidation. The model for soot filtration used in
`this study is based on their work and additional implementation
`details can be found in refs 8 and 17.
`The mathematical expression of soot mass conservation is
`described depending on the form of the deposit. Equation 9
`represents a mass conservation equation for the soot deposited
`inside the filter wall. It is applied to each wall element.
`∂
`C
`w,soot
`∂
`t
`In this equation, the soot concentration changes in time due to
`oxidation and filtration processes. The symbol Rf,w is the soot
`filtration rate by deep-bed filtration and is calculated using the
`incoming flux of soot particles and filtration efficiency.8 The
`first term on the right-hand side represents the catalytic soot
`oxidation rate, and it is expressed by eq 10.
`=
`at
`k
`RR
`S
`[O ]
`w,soot cat
`2
`2
`
`(9)
`
`(10)
`
`+
`
`R
`
`f,w
`
`NO
`2
`
`−
`
`RR
`
`Oc
`
`at
`2
`
`= −
`
`RR
`
`V
`j
`
`Oc
`
`Here kcat is a reaction constant for the soot oxidation reaction
`which occurs on the surface of soot particles and it has units of
`[m/s]. kcat is modeled by a modified Arrhenius equation which
`includes a linear dependency of the pre-exponential factor on
`temperature. The activation energy for each soot oxidation
`pathway is summarized in Supporting Information, Table S6.
`Sw,soot is surface area per unit volume [m2/m3]. For the soot
`deposited inside the filter wall, Sw,soot can be estimated by the
`following equation.
`=
`×
`A C
`S
`MW 1000
`(11)
`w,soot
`s w,soot
`C
`Here As is a surface area per unit mass of soot particles [m2/kg]
`and MWC is the molecular weight of carbon.
`The soot deposited in the cake layer is assumed to behave
`like a shrinking or inflating layer.18 Under this assumption, the
`density or concentration of soot cake layer denoted by Ccake
`stays constant. The soot mass conservation equation is
`represented in eq 12.
`∂
`w
`cake
`∂
`t
`
`f,cake
`
`+
`
`R
`
`NO
`2
`
`−
`
`RR
`
`Ot
`
`h
`2
`
`= −
`
`RR
`
`A
`
`C
`cake
`
`j
`
`(12)
`
`15584
`
`dx.doi.org/10.1021/ie3020796 | Ind. Eng. Chem. Res. 2012, 51, 15582−15592
`
`channel, respectively. In eq 1 and 5, kc is a mass transfer
`coefficient between the fluid in the channel and walls (or cake
`layer) in the inlet and outlet channel, respectively. The filter
`wall and soot cake layer are porous media, so the flow in the
`inlet channel has a suction flow as well. Hwang et al.9 studied
`how the heat transfer coefficient changes with the intensity of
`suction or blowing flow in a rectangular duct with porous wall
`and suggested an experimental correlation as presented in eq 7.
`This experimental correlation is directly used to get the mass
`transfer coefficient under the assumption of Lewis (Le) number
`of ∼1. However, eq 7 is derived when only one of four walls is
`porous but each DPF channel has
`four porous walls.
`Nevertheless,
`in the case of engineering application, several
`studies6,10 used eq 7 and they showed acceptable correlation
`with experimental results both for species and energy transport.
`recently, Kostoglou et al. derived the Nu−Rew
`Most
`correlations based on an analytical approach for a wall flow
`monolithic catalytic reactor.11 Comparisons of
`transport
`coefficient between single and four porous walls will be studied
`further and updated to the simulation code.
`=
`−
`+
`−
`2
`Re0.367
`Nu
`
`
`2.712
`<
`<
`Re
`20
`0
`w
`Equation 3 represents species balance between mass transfer
`and surface reaction. However, in eq 3, the surface area per unit
`volume (S) has a very large value so Cwf,i and Cws,i are almost
`equal. It represents that the overall reaction in a porous media
`is mostly limited by reaction kinetics. The mass transfer
`coefficient from gas stream to solid surface inside the soot cake
`layer or filter wall which is denoted by kw can be calculated by
`experimental correlation suggested by Dwivedi12 as seen in eq
`8.
`
`w
`
`Re0.0212
`
`w
`
`
`Re0.000443
`w
`
`3
`
`(7)
`
`0.365
`.386
`Re
`
`w0
`
`=
`
`0.765
`0.82
`Re
`
`+
`
`ε
`p
`
`Sh
`1/3
`Sc Re
`
`w
`w
`where Sh and Sc represent Sherwood number and Schmidt
`number, respectively.
`Depending on the y-location, the domain should be classified
`as soot cake layer or catalytic filter wall. The mass transfer
`coefficient (kw) diffusivity (Dw) and surface area (S) are porous
`wall properties and determined depending on their position.
`The reaction rate for species i which is denoted by RRi is also
`determined depending on its y-location. While the soot cake
`layer involves only the soot oxidation reactions, the catalytic
`filter wall
`includes SCR related reactions as well as soot
`oxidation reactions. An additional assumption included in this
`study is that
`the soot deposit
`inside the filter wall also
`undergoes catalytic oxidation in the presence of Cu−zeolite
`catalyst. The reaction mechanisms are summarized in
`Supporting Information, Table S1 and their rate expressions
`are presented in Table S2. For ammonia desorption,
`the
`activation energy which is denoted by EA,2 is dependent on the
`ammonia surface coverage. According to Park et al.,6 the
`exponential
`formulation for dependency of
`the activation
`energy on the ammonia surface coverage works much better
`than linear formulation so this exponential formulation is still
`used in this research. The validation of eq 1−6 and rate
`expressions for SCR are well described in their study.6
`The reaction rate term, RRi can include a product of
`concentrations of two or more species and it also couples eqs
`1−6 through all trace species. The unknowns in eqs 1−6 are
`|
`|
`o for n trace species, so
`I, Cwf,i, Cws,i, Cwf,i
`y=0 Cwf,i
`y=w and Cb,i
`Cb,i
`
`(8)
`
`BASF-2011.003
`
`

`
`Industrial & Engineering Chemistry Research
`Here, wcake is a thickness of soot cake layer. The superscript “th”
`represents thermal oxidation. For the soot cake layer, thermal
`oxidation is a main reaction pathway for soot oxidation because
`most of the catalytic material is coated inside the filter wall.
`
`3. EXPERIMENTAL SETUP AND TEST PROCEDURE
`Experimental data from 2-way DPF/SCR catalyst core samples
`(sample size ≈ 0.95 in. diameter by 2.38 in. length) provided by
`a catalyst supplier was used for model validation. The washcoat
`containing the catalyst powder was supported on 300 CPSI/12
`mil cordierite wall flow monolith substrate and aged prior to
`testing. Supporting Information, Table S3 lists the hydro-
`thermal aging conditions and the dimensions of the catalyst
`samples used in the experiments. The aging was meant to
`simulate approximately 120 000 miles of vehicle operation. The
`catalyst sample cores were tested in a fixed-bed flow reactor
`system and FTIR (Fourier transform infrared) spectroscopy
`was used for measuring the concentration of each gas species.
`Additional details on the experimental setup and tests can be
`found.19
`The aged core samples were then loaded with soot particles
`by amounts of 1.0 to 3.0 g/L using a Euro-4 V6 diesel engine
`operating on diesel certified fuel (cetane number 42) and with a
`DOC mounted upstream of the filter carrier. The catalyst
`samples were preheated to 500 °C in N2 gas in order to
`eliminate volatile organics and any adsorbed ammonia or NOx
`from the catalyst. Then the catalyst samples were cooled to 200
`°C before testing. The experimental work is focused on
`elucidating both the soot oxidation characteristics for the 2-way
`device and the difference in SCR performance between clean
`and soot loaded samples. Data from the following set of
`controlled reactor experiments were used to validate the model
`in a systematic manner.
`NO oxidation and NO2 dissociation experiments: As a
`precursor to soot oxidation inside the 2-way device, NO2
`formation is examined because it is an active soot oxidizer.
`NO2 dissociation is examined to be able to differentiate its
`impact with and without the presence of soot.
`Soot oxidation experiments attributed only to O2 in the inlet
`feed stream: The experimental data provided information on
`thermal soot oxidation and low temperature soot oxidation in
`the presence of Cu−zeolite catalytic material.
`Soot oxidation experiments in the presence of O2 and NO2 in
`the feed stream: Both NO2 and O2 work as a soot oxidizer. The
`soot oxidation rate by NO2 can be quantified in this step.
`Experiments comparing SCR performance between clean
`and soot loaded 2-way DPF/SCR cores: Both clean catalyst and
`soot loaded catalyst samples are examined in terms of DeNOx
`performance under temperature programmed conditions to
`isolate soot-NOx interactions.
`
`4. RESULTS
`4.1. NO Oxidation and NO2 Dissociation. As NO2 is an
`active soot oxidizer, the model should capture the concen-
`tration of NO2 inside the 2-way device for better prediction of
`soot oxidation characteristics. The NO2 can exist in exhaust
`gases and it also can be formed from NO oxidation over the
`Cu−zeolite catalyst. The purpose of
`the first
`test
`is to
`investigate the NO oxidation characteristics to NO2 over
`Cu−zeolite catalyst using experimental results and to calibrate
`the model using the experimental data. The test conditions are
`listed in Supporting Information, Table S4.
`
`Figure 3 shows the NO to NO2 formation by oxidation as a
`function of temperature. The feed gas does not contain NO2, so
`
`Article
`
`Figure 3. NO to NO2 conversion efficiency with respect
`to
`temperature. Conversion efficiency is calculated by 1 − Ci,IN/Ci,OUT.
`
`the NO2 in the outlet gases is the result of NO oxidation by O2.
`As expected, greater NO oxidation occurs with 10% O2
`compared to 5% O2. The results from the experiment and
`model agree within 2%. The NO conversion becomes higher as
`temperature goes up but it is still less than 7%. This is low
`compared to the NO oxidation over a DOC. One study shows
`the conversion of NO by oxidation to NO2 within a DOC is up
`to 40% over a fresh catalyst at 35 000 1/h space velocity.20 The
`1% difference between experiment and simulation at 200 °C is
`due to drift of experimental measurement because, in general,
`200 °C is too low to cause a meaningful NO oxidation reaction.
`In Figure 4, NO2 dissociation under
`the programmed
`temperature condition is plotted. The inlet feed gas does not
`
`Figure 4. Concentration of NO and NO2 in the outlet of 2-way
`blended DPF/SCR during NO2 dissociation test using the temperature
`programmed oxidation (TPO) procedure.
`
`include NO. Hence, the entire NO in the outlet feed gas is from
`NO2 dissociation. Significant NO2 dissociation is observed over
`400 °C. It is observed that the NO2 dissociation is much higher
`than NO oxidation which is shown in Figure 3. At 600 °C, the
`NO2 dissociation is almost 77%, whereas the NO oxidation is
`only 7%. This implies that NO oxidation is significantly limited
`by NO2 dissociation at high temperature due to thermody-
`namic equilibrium. The NO oxidation and NO2 dissociation
`
`15585
`
`dx.doi.org/10.1021/ie3020796 | Ind. Eng. Chem. Res. 2012, 51, 15582−15592
`
`BASF-2011.004
`
`

`
`Article
`
`Industrial & Engineering Chemistry Research
`
`tests confirm that the model can capture the NO2 conversion
`with good accuracy.
`4.2. Soot Oxidation. The NO oxidation and NO2
`dissociation test were conducted with soot-free 2-way DPF/
`SCR samples. For the soot oxidation tests, the samples were
`prepared with an initial loading of soot. Two test cases are
`examined for
`the soot oxidation study, and they are
`summarized in Supporting Information, Table S5 under labels
`“test 1” and “test 2”. The feed gas composition in test 1
`contains 500 ppm of NO and 5% O2, which is a soot oxidizer at
`temperatures above 550 °C. For test 2, the feed gas includes 5%
`O2 and 100 ppm of NO2, which is known to oxidize soot at
`temperatures below 550 °C.
`Figure 5 represents the COx (carbon monoxide and carbon
`dioxide) release rate due to soot oxidation for test 1. The
`
`Figure 6. Comparison of CO2 release with and without catalytic soot
`oxidation for test 1.
`
`400 °C. Therefore,
`it can be inferred that most of the low
`temperature soot oxidation reaction in test 1 is caused by
`catalytic oxidation.
`Figure 7 shows the COx release rate due to soot oxidation for
`test 2 described in Supporting Information, Table S5. For this
`
`Figure 5. Concentration of COx in the outlet of 2-way blended DPF/
`SCR during the soot oxidation test 1.
`
`the soot oxidation rate.
`simulation result overestimates
`However, based on the integration of the COx release curve
`from the experiment, the estimated weight of the soot-loaded 2-
`way DPF sample is less than the actual measured weight of the
`soot-loaded 2-way DPF-SCR sample by approximately 21%.
`This could be one of the reasons for discrepancy in comparing
`experimental and simulation results. It is also observed that
`there is some soot oxidation for test 1 at lower temperatures
`(300−400 °C). Possible mechanisms for this low-temperature
`soot oxidation are NO2-assisted soot oxidation and catalytic
`soot oxidation by the Cu−zeolite SCR catalyst.
`The nature of
`low temperature soot oxidation observed
`during test 1 is investigated further by using the model. The
`calibrated values of activation energy for soot oxidation are
`listed in Supporting Information, Table S6 and are consistent
`with previous research work.21−25 To investigate low temper-
`ature oxidation, test 1 was recalculated by the validated model
`cat = 0 in
`with the catalytic soot oxidation option turned off (RRO2
`eq 9). In Figure 6, the results with and without catalytic
`oxidation of soot are compared. Without catalytic oxidation, the
`low temperature soot oxidation rate underpredicts the actual
`oxidation rate. The difference between the two curves in Figure
`6 implies soot oxidation occurs by catalytic reaction. As such
`there is no NO2 in the inlet feed gas and consistent with steady
`state results on NO oxidation (Figure 3), a small amount of
`NO2 is formed by the NO oxidation reaction between 300 and
`
`Figure 7. COx concentration in the outlet of 2-way blended DPF/SCR
`during soot oxidation test 2.
`
`test, the inlet feed gas consists of NO2 only as the NOx
`constituent. The soot oxidation reaction starts at 300 °C
`because of both catalytic and NO2-assisted oxidation. Figure 8
`shows the NO2 and NO concentrations in the outlet feed
`stream. It should be noted that in the absence of ammonia in
`the inlet feed stream, there is no DeNOx reaction and the total
`amount of NOx is unchanged. The relative concentrations of
`NO2 and NO change due to NO2-assisted soot oxidation and
`NO2 dissociation. Good agreement between simulation and
`experiment in Figure 8 represents that soot oxidation by NO2 is
`well captured by simulation with the activation energy in
`Supporting Information, Table S6.
`The transient results from the NO2 dissociation test (clean 2-
`way DPF/SCR sample) and test 2 (soot-loaded 2-way DPF/
`SCR sample) are plotted in terms of NO2 conversion efficiency
`with respect to temperature in Figure 9. This plot shows how
`much NO2 is decomposed to NO for each test case. More NO2
`
`15586
`
`dx.doi.org/10.1021/ie3020796 | Ind. Eng. Chem. Res. 2012, 51, 15582−15592
`
`BASF-2011.005
`
`

`
`Industrial & Engineering Chemistry Research
`
`Article
`
`the inner surface of the porous wall and a deposit layer grows to
`form pore bridging. In the case of a clean porous wall, an
`exhaust gas stream can directly contact the catalyst which is
`coated inside the porous wall so the mass transfer mechanism
`from the exhaust gas stream to the inner catalyst washcoat is a
`boundary layer diffusion as represented in the left-hand side of
`eq 3. However,
`in the soot deposited case, the exhaust gas
`stream tries to find its pathway to minimize flow resistance
`inside the filter wall so there is little chance for the exhaust gas
`stream to contact directly the inner catalyst washcoat. Hence,
`the diffusion through the soot deposit to the catalyst washcoat
`is added to the mass transfer mechanism. This soot deposit
`layer has very small inner length scale so diffusion mechanism
`might be Knudsen diffusion and it might significantly limit the
`overall
`reaction rate as compared with a clean case as
`summarized in Supporting Information, Table S7. Thus, the
`actual concentration on the catalyst washcoat (Cws,i) becomes
`lower as compared to the clean case, and the reaction rate
`denoted by RRi(Cws,i) will decrease. A schematic diagram for
`this case is presented in Figure 10.
`
`Figure 8. The NOx concentration in the outlet of 2-way blended
`DPF/SCR during soot oxidation test 2.
`
`Figure 9. Comparison of NO2 to NO conversion for the soot-loaded
`case (test 2) and clean case (NO2 dissociation test).
`
`Figure 10. Schematic diagram for soot deposit as a deep bed filtration
`and related variables for eqs 13 to 15.
`
`is converted to NO for test 2 than for the NO2 dissociation test.
`This is because NO2 is also consumed by the soot oxidation
`reaction in test 2. The difference in NO2 conversion efficiency
`between the two tests implies the consumption of NO2 due to
`the soot oxidation reaction. At 400 °C, the difference is at its
`maximum and further decreases with an increase in temper-
`ature. At 600 °C,
`it is observed that all the NO2 to NO
`conversion is from dissociation as opposed to NO formation
`from soot oxidation.
`4.3. Effect of Soot Deposit on SCR Reaction.
`4.3.1. Effect of Soot Deposit
`Inside the Filter Wall.
`Deterioration of catalyst performance by coking is very
`common in a device which uses hydrocarbon fuel. This
`performance deterioration can be recovered to the original state
`if the soot particles are removed. This is different from catalyst
`deactivation by sintering or poisoning. Several empirical
`relations have been suggested to model
`the impact of
`coking.26,27 In this study, an analytical approach was developed
`to model the effect of soot deposits inside the filter wall and the
`impact on SCR reactions.
`The assumption is that soot deposits inside the filter wall
`work as a barrier for mass transfer from the gas stream to the
`catalytic sites, and this reduces the overall catalytic reaction rate
`for the SCR reactions facilitating NOx reduction. Usually,
`during a deep bed filtration process, soot particles deposit on
`
`The concept of effectiveness factor and a parameter called
`the Thiele modulus has been used to model the effect of
`diffusion inside the catalyst washcoat. This idea can be
`extended to the soot loaded filter wall to capture the reaction
`mechanism between the exhaust gas stream and the catalyst
`washcoat inside the porous filter wall. As discussed, diffusion
`through the soot deposit layer is a deterministic mass transfer
`mechanism for the soot deposit case. Therefore, the diffusion-
`reaction equation can be used to model the balance between
`mass transfer and reaction for the soot deposited case as it is
`expressed by eq 3 for the clean case.
`On the basis of the coordinate system shown in Figure 10,
`the diffusion reaction equations are
`∂
`2
`C
`ξ
`∂ ̅
`2
`∂
`2
`C
`ξ
`∂ ̅
`2
`(13b)
`where C̅s a nondimensional concentration and defined as C̅ =
`C/CI and ξ̅ is a nondimensional length scale and defined as ξ̅ =
`ξ/δ
`w. The symbol CI is the concentration at the interface of the
`soot deposit and the washcoat. The thickness of the soot
`− δ
`deposit layer, δ
`w can be estimated using the unit collector
`s
`model8 which is incorporated in the deep-bed filtration model.
`
`− Φ ̅ =
`2
`C
`
`0
`
`for 0
`
`ξ
`≤ ̅ ≤
`
`δ
`w
`
`(13a)
`
`=
`
`0
`
`for 0
`
`ξ
`≤ ̅ ≤
`
`δ
`s
`
`15587
`
`dx.doi.org/10.1021/ie3020796 | Ind. Eng. Chem. Res. 2012, 51, 15582−15592
`
`BASF-2011.006
`
`̅
`̅
`

`
`Article
`
`(20)
`
`(21)
`
`⎞⎠⎟
`
`−
`δ
`s
`
`ws
`
`
`C C/
`I
`δ
`−
`w
`
`1
`
`⎛⎝⎜
`
`RR ideal
`
`= −k C
`R,w ws
`
`η
`s
`
`=
`
`RR
`RR
`
`act
`
`=
`
`ideal
`
`D
`eff,s
`δ
`k
`R,w w
`
`where Cws/CI can be acquired from eq 17.
`Equation 21 can be rearranged to eq 22 by replacing Cws/CI
`with eq 17.
`
`⎞ ⎠⎟⎟
`
`=
`
`η
`s
`
`lim
`δ
`δ→
`s
`w
`
`=
`
`tanh(
`Φ
`w
`
`)
`
`=
`
`η
`c
`
`(23)
`
`⎛ ⎝⎜⎜
`
`η
`s
`
`=
`
`D
`eff,s
`δ
`k
`R,w w
`
`×
`
`D k
`eff,w R,w
`D k
`eff,w R,w
`
`Φ
`tanh(
`)
`w
`Φ +
`tanh(
`)
`w
`
`δ
`s
`
`(
`
`−
`
`δ
`w
`
`)
`
`D
`eff,s
`(22)
`s approaches δ
`Equation 22 shows that if δ
`w (e.g., the soot layer
`thickness goes to zero), the effectiveness factor converges to
`that of a soot-free catalyst as given by eq 23.
`Φ
`Φ
`D k
`tanh(
`)
`w
`eff,w R,w
`w
`δ
`k
`R,w w
`
`Industrial & Engineering Chemistry Research
`The symbol Φ
`w represents the Thiele Modulus which is defined
`as
`
`Φ = k
`
`w
`
`2
`
`δ
`R,w w
`D
`eff,w
`
`(14)
`
`This represents the ratio of diffusion and reaction inside the
`washcoat. The symbol kR,w is a representative first-order
`reaction constant.
`The boundary conditions for eqs 13a and 13b are
`∂ ̅
`=C
`ξ
`̅ =
`ξ
`∂ ̅
`̅ =C
`
`at
`
`0,
`
`ξ̅ =
`
`at
`
`1,
`
`0
`
`1
`
`(15a)
`
`(15b)
`
`(15c)
`
`C C
`
`ws
`
`̅ =C
`
`,
`
`w s
`δδ
`
`ξ
`
`̅ =
`
`at
`
`for 0
`
`ξ
`≤ ̅ ≤
`
`(24)
`
`act
`
`clean
`
`In the case of a wall-flow-type SCR, the washcoat layer is much
`thinner than that of a flow-through type SCR. Hence, the η
`c is
`close to unity as seen in Figure 13 and RRclean is almost equal to
`RRideal.
`Finally, the apparent reaction rate for the soot-loaded case
`can be adjusted using the effectiveness factor giving the
`following expression:
`η=
`RR
`RR
`s
`Even though this effectiveness factor is derived based on the
`assumption of a first order reaction, it can be applied to the
`SCR reactions, since most of the SCR reactions used in this
`study are of first order.
`Equations 22 and 24 are implemented into the 2-way DPF/
`SCR solver and the model is applied to simulate a numerical
`exercise. Test conditions are listed in Supporting Information,
`Table S8, and the simulations are conducted for a clean and
`soot loaded 2-way DPF/SCR. The catalyst is initially free from
`adsorbed ammonia until time t = 0 min. In Figure 15, it is
`observed that NOx conversion is significantly decreased in the
`case of a soot-loaded filter at 200 and 300 °C. There is also no
`significant soot oxidation in this temperature range. However,
`at 400 °C or higher, DeNOx performance for the soot-loaded
`filter is recovered to the clean case performance because these
`temperatures are high enough to burn out the soot deposited
`inside the filter wall.
`the amount of soot
`Figure 12 shows the variation of
`deposited inside the filter wall and the thickness of the soot
`cake layer. Around 50 min into the test, indicated by “(A)” in
`Figure 11, there is still soot remaining inside the filter wall.
`Inhibition by soot present inside the filter wall contributes to
`poor NOx conversion compared to the clean sample. At around
`200 min into the test, the soot deposited inside the filter wall is
`burned out, but soot still remains in the cake layer as shown in
`Figure 12. At this moment, the filter is in a pa