Dynamic Heterogeneous Multiscale Filtration Model - ACS Publications

Aug 31, 2017 - This dynamic HMF model is based on a probability density function (PDF) description of the pore size distribution and classical filtrat...
0 downloads 0 Views 986KB Size
Subscriber access provided by UNIVERSITY OF CONNECTICUT

Article

Dynamic Heterogeneous Multiscale Filtration Model: Probing Micro- and Macro-scopic Filtration Characteristics of Gasoline Particulate Filters Jian Gong, Sandeep Viswanathan, David Rothamer, David E Foster, and Christopher J. Rutland Environ. Sci. Technol., Just Accepted Manuscript • DOI: 10.1021/acs.est.7b02535 • Publication Date (Web): 31 Aug 2017 Downloaded from http://pubs.acs.org on September 2, 2017

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

Environmental Science & Technology is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 29

Environmental Science & Technology

1

Dynamic Heterogeneous Multiscale Filtration

2

Model: Probing Micro- and Macro-scopic Filtration

3

Characteristics of Gasoline Particulate Filters

4

Jian Gong*, †, Sandeep Viswanathan†, David A. Rothamer†, David E. Foster†, Christopher J.

5

Rutland† †

6 7

Engine Research Center, University of Wisconsin-Madison,1008 Engineering Research Building,1500 Engineering Drive, Madison, WI 53706, United States

8

*

Corresponding author e-mail: [email protected]

Present address: 1900 McKinley Avenue, MC 50183, Columbus, IN 47201, USA

ACS Paragon Plus Environment

1

Environmental Science & Technology

Page 2 of 29

9

ABSTRACT – Motivated by high filtration efficiency (mass- and number-based) and low

10

pressure drop requirements for gasoline particulate filters (GPFs), a previously developed

11

heterogeneous multiscale filtration (HMF) model is extended to simulate dynamic filtration

12

characteristics of GPFs. This dynamic HMF model is based on a probability density function

13

(PDF) description of the pore size distribution and classical filtration theory. The microstructure

14

of the porous substrate in a GPF is resolved and included in the model. Fundamental particulate

15

filtration experiments were conducted using an exhaust filtration analysis (EFA) system for

16

model validation. The particulate in the filtration experiments was sampled from a spark-ignition

17

direct-injection (SIDI) gasoline engine. With the dynamic HMF model, evolution of the

18

microscopic characteristics of the substrate (pore size distribution, porosity, permeability, and

19

deposited particulate inside the porous substrate) during filtration can be probed. Also, predicted

20

macroscopic filtration characteristics including particle number concentration and normalized

21

pressure drop show good agreement with the experimental data. The resulting dynamic HMF

22

model can be used to study the dynamic particulate filtration process in GPFs with distinct

23

microstructures, serving as a powerful tool for GPF design and optimization.

24

TOC/Abstract art

25

ACS Paragon Plus Environment

2

Page 3 of 29

26

Environmental Science & Technology

Introduction

27

Diesel particulate filters (DPFs) have been widely and successfully applied on diesel engines

28

as standard aftertreatment devices to control particulate emission since 2007.1 In the last decade,

29

there has been significant progress on the development of spark-ignition direct-injection (SIDI)

30

gasoline engines, which show benefits for fuel economy and CO2 emission reduction.2,3

31

However, SIDI engines were found to generate more ultrafine particulate (90%).13,14 Furthermore, fuel economy and power of SIDI engines are more sensitive to the

42

exhaust backpressure introduced by GPFs13. The need to achieve high filtration efficiency and

43

low pressure drop poses a significant challenge for the development of advanced GPFs for

44

particulate emission control. On the other hand, in addition to mass-based particulate emission

45

regulations throughout the world, a particle number limit of 6 x 1011#/km is being enforced in

46

Europe for all SIDI engines from 2017.15 These challenges and stringent particulate emission

47

regulations require a fundamental understanding of particulate filtration process in GPFs, as well

48

as, detailed mathematic models that assist GPF development.

ACS Paragon Plus Environment

3

Environmental Science & Technology

Page 4 of 29

49

Particulate filtration is a dynamic process, in which the microstructure of the porous GPF

50

substrate varies continuously as filtration proceeds. In the literature, there are limited

51

experimental and modeling studies on the dynamic filtration process in GPFs. It is critical to

52

understand how the microscopic (e.g., collector size, local porosity, and local permeability) and

53

macroscopic filtration characteristics (e.g., number-based filtration efficiency and pressure drop)

54

vary with time. One widely adopted approach for simulating the dynamic filtration process is

55

“unit-cell” method.16,17,18 This approach was developed on the basis of the flow field solutions

56

initially derived by Kuwabara16 and Happel17 for low-speed laminar flow in a randomly packed

57

bed of spheres. Lee et al.19 theoretically derived the equations of filtration efficiency and

58

pressure drop for a clean filter on the basis of Kuwabara flow.16 Later, these equations were

59

extended to soot loaded DPFs and used to simulate the filtration process.20,21,22,23 In the “unit-

60

cell” approach, a representative mean pore size and a mean porosity were selected to represent

61

the porous substrate by assuming a homogeneous porous substrate24. This assumption works well

62

for filtration modeling in DPFs, which are less sensitive to the microstructure of the porous

63

substrate. GPFs are exposed to a larger fraction of ultrafine particulate that can penetrate through

64

the filter due to the absence of a “soot cake” layer as the primary filtration medium.25,26,27 As a

65

result, the microstructure of the porous substrate is critical and the interaction between the

66

particulate and the porous substrate has a significant impact on the filtration performance.

67

Moreover, in a realistic filter, the porous substrate has an inherently heterogeneous

68

microstructure, which spans a wide range of length scales. Recently, a PDF-based heterogeneous

69

multiscale filtration (HMF) model was developed to calculate the filtration efficiency of clean

70

GPFs by considering the heterogeneous microstructure of the substrate.28 Rather than using the

71

mean pore size and mean porosity, a pore size probability density function (PDF) established

ACS Paragon Plus Environment

4

Page 5 of 29

Environmental Science & Technology

72

from experimental observations29,30 was introduced to represent the heterogeneous multiscale

73

porous substrate. The HMF model was validated on three particulate filters with distinct mean

74

pore sizes at various flow rates28. In the present work, the PDF-based HMF model is extended to

75

model the dynamic filtration characteristic of a GPF, which allows us to probe the dynamic

76

evolution of the microstructure and the macroscopic filtration characteristics.

77

Experiments

78

Particulate emissions. Particulates in the filtration study were generated under steady-state

79

operation of a single-cylinder SIDI gasoline engine fueled with EPA Tier II EEE gasoline

80

(Haltermann Solutions). The SIDI engine specifications are summarized in Table S1 in the

81

Supporting Information. The engine was operated at a fuel-rich condition to generate a particle-

82

size distribution (PSD) representative of emissions during cold-start operation where emissions

83

are expected to be at their highest.31 The specific engine operating parameters for the fuel-rich

84

condition are defined in Table S2. A TSI scanning mobility particle sizer (SMPS) with a TSI

85

3081 long differential mobility analyzer (DMA) and a TSI 3010 condensation particle counter,

86

were used to measure the PSD in the diluted exhaust stream during the initial characterization.

87

The dilution was achieved using a two-stage partial flow dilution system, with a heated primary

88

porous tube dilution stage (350°C) followed by a room temperature secondary ejector dilution

89

stage. Once the engine operating conditions were established, micro-scale filtration experiments

90

on a GPF sample were performed using the EFA system.

91

Exhaust Filtration Analysis (EFA) System. The EFA system was initially developed at the

92

University of Wisconsin-Madison Engine Research Center (ERC) for diesel particulate filtration

93

studies32,33. It was later modified by Viswanathan et al.34,35 for low particulate mass

94

concentration filtration studies using SIDI engine particulate. A schematic view of the EFA

95

system is depicted in Figure S1. A thin, rectangular section of the filter wall, referred to as a

ACS Paragon Plus Environment

5

Environmental Science & Technology

Page 6 of 29

96

wafer, was placed in a stainless steel holder and sealed using silicone O-rings capable of

97

handling temperatures up to 200°C. The wafer holder was designed to have near uniform

98

filtration velocities across the inlet face of the wafer.32

99

Raw exhaust was sampled downstream of the exhaust surge tank and sent through the wafer

100

holder which was placed in an oven to enable precise control of the filtration temperature. A

101

dilution system including an ejector diluter (Dekati® DI 1000), which was supplied with heated

102

dilution air at 175°C, was adopted in the EFA system as shown in Figure S1. The flow rate (or

103

filtration velocity) through the wafer was controlled by adjusting the dilution air flow at the

104

ejector diluter downstream of the oven with respect to the total flow into the SMPS, CO2

105

analyzer and flow orifice. Real time dilution ratio was determined using the measured CO2

106

concentrations in the raw exhaust and downstream of the EFA.

107

The specifications of the wafer sample used in this study are shown in Table S3. The pressure

108

drop across the wafer holder was measured using a differential pressure transducer (Omega, PX-

109

138) with a range of 5 psi and an accuracy of ±1% of full scale. In the EFA system, the absence

110

of alternatively plugged channels allows us to study the fundamental filtration process within the

111

filter walls without considering the effects of flow contraction and expansion at the entrances and

112

exits of filter channels as in a typical full-scale filter. As a result, only the particulate filtration

113

process, in which the exhaust gas with particulate passed through the wafer, was simulated. A

114

constant face or wall velocity (reported in Table S3) calculated from measured volumetric flow

115

rate was used in the simulation.

116

A TSI engine exhaust particle sizer (EEPS) was used alongside the SMPS to monitor the PSD

117

upstream and downstream of the EFA system. The EEPS is capable of measuring real-time PSD

118

in the range of 5.6 to 560 nm with 1-s resolution. More details about the EFA system and

ACS Paragon Plus Environment

6

Page 7 of 29

Environmental Science & Technology

119

experimental procedures can be found in Viswanathan et al.35 It should be noted that the gas

120

temperature was held at 175 °C in order to prevent any potential particulate oxidation. It is

121

widely reported that soot oxidation typically occurs at relatively high temperatures (passive NO2

122

soot oxidation can be initiated around 250 °C36,37 and active thermal soot oxidation occurs

123

around 550 °C38,39). During the filtration experiments, size-resolved particle number

124

concentrations and pressure drop across of the EFA system were continuously recorded.

125

Multiscale Filtration Modeling

126

The state-of-art particulate filtration modeling involves several coexisting length scales:40

127

entire filter brick scale (dm), single channel scale (cm), wall scale (mm) and pore scale (µm). At

128

the filter brick scale, component and system level modeling studies such as particulate

129

loading,41,42,43 regeneration control,44,45 and interactions between particulate filter and other

130

emission control devices,46,47,48,49 have been extensively conducted. A complete mathematical

131

description of the filter inlet and outlet channels can be found in Bissett50. In modeling the

132

dynamic filtration process in the EFA system, in which filter inlet and outlet channels do not

133

exist, description of the channel scales in a filter can be neglected and replaced by a single face

134

velocity. In this section, a dynamic filtration model at the wall and pore scale is presented. A

135

graphical depiction of the dynamic HMF model is shown in Figure 1. There are several

136

assumptions in modeling the filtration process:1) axisymmetric geometry of the porous substrate;

137

2) all particles with different mobility diameters sampled from the engine exhaust are spherical;

138

3) a constant effective density (1000 kg/m3) for all particulate51.

139 140

Wall Scale. At the substrate wall scale, a porous substrate with a thickness of ℎ was

discretized into a number of slabs with a uniform thickness of ∆ along the depth of the substrate

141

wall. Each slab was discretized into a number of cells with a constant width of ∆ along the

142

channel direction as shown in Figure 1. Moreover, a porosity distribution profile was imposed to

ACS Paragon Plus Environment

7

Environmental Science & Technology

Page 8 of 29

143

resolve the heterogeneous wall structure. From recent TEM studies30,52 of the microstructure of

144

particulate filters, porosity is inhomogeneous across the filter wall. The porosity increases

145

drastically from a nominal mean value at the center of a filter to a higher value near the filter

146

surfaces. Furthermore, the porosity distribution across the filter wall was found to have a

147

significant effect on local flow distribution, filtration efficiency and pressure drop26,53,54. Our

148

previous simulation studies25 showed that most of particles were deposited at the top slab with a

149

constant or homogeneous porosity, resulting in a high particle concentration gradient near the top

150

surface of the substrate. With an inhomogeneous porosity distribution, the highest concentration

151

of particles was observed at the transition region of the porosity distribution rather than at the top

152

slab. Also, particles are readily to penetrate deeper into the substrate due to the high porosities of

153

the top slabs. Therefore, it is critical and necessary to resolve the porosity distribution to capture

154

the dynamic interactions between particulate and the local microstructure of the porous substrate.

155

As shown in Figure 1, each slab at different locations has a specific porosity. Due to lack of

156

experimental porosity distribution data for this filter sample, a porosity distribution was imposed

157

with a high porosity of 0.68 at the top and bottom surfaces and a nominal mean porosity of 0.48

158

at the center of the filter wall on the basis of experimental porosity distributions of particulate

159

filters30. An exponential function was imposed to model the transition between the high surface

160

porosity and the nominal wall porosity, which is shown in eq S1. It should be noted that it is

161

possible to have a porosity distribution, in which the porosity decreases near the substrate surface

162

for coated particulate filters. The impact of various porosity distributions in GPF’s filtration

163

performance will be studied in our future work55.

ACS Paragon Plus Environment

8

Page 9 of 29

Environmental Science & Technology

164 165

Figure 1. Schematics of multiscale modeling of particulate filtration in a GPF.

166

Mass conservation of particulate at each slab can be achieved by realizing that the mass of

167

particulate flowing into the slab is equal to the sum of the mass of particulate flowing out of the

168

slab and the mass of particulate deposited in the slab. By assuming that all the particles are

169

spherical with a constant density, mass conservation equations for the particulate can be

170

translated into number-based conservation equations. In other words, the number of particles

171

breaking through the jth slab ( ) can be calculated from the number of particles exiting

172

the (j-1)th slab ( ) and the filtration efficiency of the jth slab. PNd   1  ηd  ∙ PNd 

(1)

173

At the first slab, a partition coefficient  was employed to determine the fraction of upstream

174

particulate deposited on the surface of the first slab, which has been widely used to model the

175

partition fraction between surface “cake” filtration and “deep-bed” filtration.20,56,57 With the

176

partition coefficient, the number of particles flowing through the first slab can be calculated as PNd

 1  ϕ ∙ PNd 

(2)

ACS Paragon Plus Environment

9

Environmental Science & Technology

177

where 

178

defined as



is the number of particles upstream of the filter. The partition coefficient is

ϕ 

179

Page 10 of 29

d#!,  d#!$ ψb #  d#!$

(3)

where ' is a percolation factor with a value of 0.95. It determines the maximum collector size

180

that can be reached before the end of “deep-bed” filtration. Here ( is the maximum diameter of a

181

collector, which can be calculated from the initial collector size ()$ ) and mean porosity (*), i.e.,

182

,(   .

//1 .

183 184

+

Eventually, the number of particles exiting the filter (at the last slab 2  3456 ) can be calculated. The total particle number filtration efficiency is given by eq 4. ηd  1 

185 186

PNd 789:;< PNd 

Pressure drop is another critical parameter of particulate filtration. According to Darcy’s law58, the total pressure drop across the entire filter wall can be calculated as ∆P=>?  ∆P!>@A + ∆PC>?? 

187

(4)

μvC wG + ∆PC>?? kG

(5)

where I3 represents the permeability of a “soot cake” with a thickness of J3 . The wall velocity

188

KL along the channel can be calculated from one-dimensional incompressible Navier Stokes

189

equations.50,56 In simulating the EFA system, the wall velocity is a constant. The “soot cake”

190

accumulates only above the substrate surface (the first slab), therefore there is only one pressure

191

drop term on the right side of eq 5. On the other hand, the pressure drop across the entire wall is

ACS Paragon Plus Environment

10

Page 11 of 29

Environmental Science & Technology

192

a summation of the pressure drop across each slab. The total pressure across the entire wall is

193

then calculated as 89:;
40

270

µm) does not change significantly. The initial population of large collectors is smaller and

271

filtration efficiency of the large collectors (>40 µm) is also lower compared to the smaller

272

collectors (10-20 µm). Therefore, there is less particulate depositing on large collectors and the

273

increase in the collector diameter of the large collectors is negligible. Consequently, the collector

274

size PDF shifts to the right with a sharp peak of 25 µm, which is slightly larger than the initial

275

mean collector diameter of 18 µm as filtration proceeds. At the middle slab, as a result of the

276

reduced amount of particulate being trapped, the increase in mean diameter of the collector

277

cluster is milder and the collector size PDFs shows only slight differences at 20 min compared to

278

that at 10 min. At the last slab, there is negligible increase in the mean diameter of the collector

279

cluster. Detailed progressive evolutions of the collector size PDFs at the first slab and middle

280

slab are shown in Figure S2.

281

ACS Paragon Plus Environment

15

Environmental Science & Technology

Page 16 of 29

282

Figure 2. Evolution of mean collector sizes and PDFs (inset graphs) at first, middle, and last slab

283

of porous substrate.

284

(b) Particulate mass, porosity and permeability distributions. Snapshots of the particulate

285

mass distributions at 4 min, 8 min, 12 min, and 20 min are shown in Figure 3. At the beginning of

286

filtration (4 min), the filter is almost clean as there is only a small amount of particulate trapped

287

in the filter. At 8 min, the amount of particulate in the filter starts to increase and there is a

288

gradient of particulate mass trapped across the filter. Even though a high concentration of

289

particulate is observed at the top 1/4 of the filter, there is considerable particulate penetrating into

290

the middle of the filter as is evident from the higher particulate concentrations in the bottom 1/2

291

of the filter at 8 min compared to 4 min. At 12 min, a clear gradient of particulate mass across

292

the top 1/4 of the filter can be observed. A larger gradient of particulate mass is found at 20 min

293

with a higher concentration of particulate in the top portion of the wall. There is little increase in

294

the particulate in the bottom 1/2 of the filter. The local porosity distributions of the filter at

295

various times are described in Figure 4 with a local minimum in porosity at the 3rd slab of the

296

filter. The location corresponding to the minimum porosity actually correlates to a transition

297

position in the imposed porosity distribution as shown in Figure 1. At this transition position, the

298

porosity is neither too high nor too low resulting in a relatively high particulate trapping capacity

299

as well as a high filtration efficiency. These observations indicate that the porosity distribution

300

plays a significant role in altering the local microstructure and performance of the filter. Similar

301

to the local porosity distribution of the filter wall, the local permeability is shown in Figure S3

302

with a local minimum permeability of 0.8E-13 m2 at the 3rd slab, which is about 22.9% of the

303

initial permeability of the clean filter.

ACS Paragon Plus Environment

16

Page 17 of 29

Environmental Science & Technology

304

The progressive particulate mass, local porosity, and permeability distributions in the filter

305

are consistent with each other, as well as, the mean cluster diameter evolutions in Figure 2. A

306

majority of the incoming particulate is trapped in the very top portion of the filter, where a higher

307

gradient of particulate mass, and a lower porosity and permeability are observed. There is a small

308

fraction of particulate breaking through the top half of the filter and depositing in the bottom half

309

of the filter. The particulate mass and porosity distributions in the bottom region of the filter are

310

very similar between 20 min and 4 min. This indicates that the bottom 1/3 of the filter makes

311

little contribution to the overall filtration performance for this loading case.

312 313

Figure 3. Local particulate mass distribution across the filter wall at t=4 min, 8 min, 12 min and

314

20 min.

ACS Paragon Plus Environment

17

Environmental Science & Technology

Page 18 of 29

315 316

Figure 4. Local porosity across the filter wall at t=4 min, 8 min, 12 min and 20 min.

317

Macroscopic filtration characteristics

318

Pressure drop and filtration efficiency (mass- and number-based) are widely used to represent

319

the macroscopic filtration characteristics of a GPF. An ideal GPF should have a low pressure

320

drop and high filtration efficiency. During filtration, the macroscopic state of a filter can be

321

characterized by the apparent permeability of the filter, which is directly correlated to the total

322

pressure drop. On the other hand, contributions from local and individual collectors at various

323

scales on trapping particulate lead to an overall filtration efficiency, which can be experimentally

324

measured.

325

(a) Normalized pressure drop.

326

In the EFA system, the face velocity towards the filter slightly decreased with time during

327

filtration. This decreased face velocity is assigned to the increased backpressure as a result of

328

particulate deposition. To isolate the effect of face velocity in evaluating the apparent

329

permeability, a normalized pressure drop is defined in eq 13 according to Darcy’s law. On the

330

right side of eq 12, gas viscosity | and wall thickness ℎ are constants. Accordingly, the

331

normalized pressure drop was employed to represent the apparent permeability.

ACS Paragon Plus Environment

18

Page 19 of 29

Environmental Science & Technology

∆ | ∙ ℎ  } IL,~

(12)

332

The particulate packing density vw in eq 9 was calibrated to be 40 kg/m3 in order to match the

333

experimental normalized pressure drop. With the calibration, the model is capable of describing

334

the dynamic macroscopic state of the filter. The normalized pressure drops of individual slabs

335

are shown in Figure 5. It is interesting to see that slabs at distinct positions behave differently.

336

Even though the face velocity slightly decreases with time, the normalized pressure drops of the

337

slabs at the top 1/3 of the filter increase exponentially with time. In contrast, the normalized

338

pressure drops of the slabs at the bottom 1/2 of the filter increase marginally after 20 min. These

339

different behaviors are the consequences of distinct contributions of various slabs during

340

particulate filtration. Overall, the normalized total pressure drop of the filter in Figure 5 increases

341

approximately linearly with time. This linear relationship is quite different compared to the non-

342

linear relationship in traditional DPFs, in which normalized pressure drop shows two distinct

343

slopes with time in the early stage of filtration.

344

absence of complete transition to “soot cake” filtration regime in the current experiment.

20,60

The linear correlation is attributed to the

345

ACS Paragon Plus Environment

19

Environmental Science & Technology

Page 20 of 29

346

Figure 5. Normalized individual pressure drop across each slab (1 to 36) and the total normalized

347

pressure drop (inset graph).

348

(b) Particle number filtration efficiency. In SIDI engines, there is a significant amount of

349

ultrafine particulate below 100 nm. The filtration efficiencies for these ultrafine particles are

350

critical for an effective GPF. During the rich operating mode, more than 95% of the particles in

351

the SIDI engine exhaust are below 200 nm in mobility diameter as shown in Table S2. On the

352

other hand, current particle number regulations count all particles with a mobility diameter above

353

23 nm15. Moreover, evolution of the number concentrations at specific diameters is important to

354

understand the complicated filtration process. Thus, the changes in number concentrations of

355

various particle diameters (30, 50, 100, and 200 nm) are displayed in Figure 6. Also,

356

experimental and simulated number-based filtration efficiencies for each particle size at 20 min

357

are given in Figure 6.

358

As can be seen from Figure 6, both experiments and simulations show very high number

359

filtration efficiency close to 100% for particulate with diameters below 50 nm. The dynamic

360

evolutions of particle number concentrations over the full spectrum of the particle size in 20 min

361

are shown in Figure S4, in which the filtration efficiency decreases as particle size increases. The

362

most penetrating particle size (MPS), which is defined as the particle size with the lowest capture

363

efficiency, is the signature characteristic of a specific substrate under a specific operating

364

condition. From Figure S4, the most penetrating particle size for this filter substrate is found to

365

be around 100 nm with a filtration efficiency of 45.7%. This observation is the consequence of

366

interactions between diffusion collection and interception collection mechanisms in particle

367

filtration. Diffusion collection efficiency decreases as particle size increases. In contrast,

368

interception collection efficiency increases with particle size. For small particles, diffusion

ACS Paragon Plus Environment

20

Page 21 of 29

Environmental Science & Technology

369

collection is dominant. As the particle size increases, interception collection starts to play a role

370

in total particle filtration. The resulting consequence is the total filtration efficiency shows a “V”

371

shape in the particle size domain.

372

The filtration efficiency of 100 nm particle is under-predicted by the model. For large

373

particulate with a diameter of 200 nm, the filtration efficiency is above 97%, which is in a good

374

agreement with the model. Although there are some differences in the filtration efficiencies

375

between the experiment (45.71%) and simulation (33.31%) for 100 nm particle as shown Figure

376

6, it should be noted that the experimental and simulated filtration efficiencies for comparison

377

were evaluated at 20 min. What is more important from Figure 6 is the dynamic evolution of the

378

filtration efficiency with time. From this perspective, the experimental and simulated filtration

379

efficiencies are quite comparable over the full 20 min. Also, this discrepancy could be resulted

380

from the sensitivity of the particulate measurement system on the increased backpressure across

381

the wafer due to particulate deposition. As it can be seen from Figure 6, the particle

382

concentration starts to oscillate after 15 min. Overall, the dynamic evolution of the particle

383

concentration is correctly predicted by the model across the whole range of particle sizes.

Conc. [#/ cm 3]

10 10 10 10

Conc. [#/ cm 3]

10 10 10 10

384

6

30 nm 4

2

0

0

10 ηexp=100 %

10

ηsim=99.94 % 10

20

8

100 nm 6

4

2

0

10

10

6

50 nm 4

2

10

10

ηsim=33.31 % 10 Time [min]

20

ηsim=93.91 %

0

10 ηexp=45.71 %

ηexp=100 %

10

0

10

20

6

200 nm 4

2

0

0

ηexp=97.21 % ηsim=97.61% 10 Time [min]

20

ACS Paragon Plus Environment

21

Environmental Science & Technology

Page 22 of 29

385

Figure 6. Evolution of particle number concentrations of mono-sized particles penetrating

386

through the filter sample (dash: exp; solid: sim).

387

The role of GPF’s microstructure in filtration modeling. The microstructure of a GPF’s

388

substrate is critical to the filtration performance. The heterogeneous porosity across the substrate

389

has a direct effect on the local particulate distribution61. As shown in Figure 3, the highest

390

concentration of particulate is observed at the top 1/4 of the filter, which corresponds to the

391

transition position of the porosity profile across the substrate (between the high porosity at the

392

surfaces and the nominal low porosity at the center of the substrate). The resulting low porosity

393

at this transition position prevents particulate from penetrating further into the substrate.

394

Furthermore, the pore size distribution of the GPF’s substrate is critical and determines the

395

filter’s initial filtration efficiency28. A filter with a narrow pore size distribution was found to

396

have higher filtration efficiency62. During the filtration, the dynamic change of the pore size

397

distribution is directly related to the microscopic filtration characteristics as described in Figure 2

398

and Table 1. It is therefore necessary to resolve the microstructure of the GPF’s substrate in the

399

filtration model.

400

With the GPF’s microstructure resolved in the dynamic HMF model, the dynamic interaction

401

between the substrate’s microstructure and particulate can be studied. Specifically, the evolution

402

of microscopic characteristics of the porous substrate such as porosity, permeability, and the

403

amount of deposited particulate can be probed. The dynamic changes of the substrate’s

404

microscopic properties lead to the change of macroscopic filtration characteristics such as

405

pressure drop and filtration efficiency. As a result, the gap between the microscopic filter

406

properties and the macroscopic filtration characteristics during particulate filtration in GPFs is

ACS Paragon Plus Environment

22

Page 23 of 29

Environmental Science & Technology

407

bridged, which is beneficial to better comprehend the filtration dynamics in GPFs. Moreover,

408

this dynamic HMF model can be applicable to explore optimum or ideal structure for

409

microscopic filter design.

410

Acknowledgement

411

The authors also would like to thank General Motors Research and Development for their

412

funding and support of this research through the GM-UW collaborative research laboratory

413

(CRL) program.

414

Supporting Information Available

415

Details of SIDI engine specifications and filtration experiments. Derivations of porosity,

416

permeability in the dynamic HMF model. Progressive evolutions of collector size PDF and

417

porosity distribution during filtration. This material is available free of charge via the Internet at

418

http://pubs.acs.org.

419

Nomenclature

420

Abbreviations

PDF

probability density function

HMF

heterogeneous multi-scale filtration

GPF

gasoline particulate filter

EFA

exhaust filtration analysis

SIDI

Spark-ignition direct-injection

DPF

diesel particulate filter

PFI

port fuel injection

DEFA

diesel exhaust filtration analysis

ACS Paragon Plus Environment

23

Environmental Science & Technology

421

SEM

scanning election microscopy

LBM

lattice boltzmann method

Page 24 of 29

Symbols

PN

U

 

' (

G?> U)c , 

€,c ∆ |



KL J3

JL I3

IL

)c

particle number number filtration efficiency mobility particulate diameter partition coefficient percolation factor maximum diameter of a collector number of slabs number filtration efficiency of a collector with diameter of )c at particulate diameter of  total single collector efficiency pressure drop across a filter gas viscosity wall thickness wall velocity cake thickness slab thickness soot cake permeability slab permeability individual collector diameter in a collector cluster

ACS Paragon Plus Environment

24

Page 25 of 29

Environmental Science & Technology

`a+,b * ℎ

^+) v u

),

a*

pdf of a collectors cluster porosity of the jth slab thickness of filter wall length scale of the filter collector particulate density mass of deposited particulate mean diameter of collector cluster hydrodynamic term

422

Literature Cited

423

(1)

Johnson, T. Vehicular Emissions in Review. SAE Int J Engines 2013, 6 (2), 699–715.

424 425

(2)

Zhao, F.; Lai, M.-C.; Harrington, D. L. Automotive spark-ignited direct-injection gasoline engines. Prog Energy Combust Sci 1999, 25 (5), 437–562.

426 427 428

(3)

Gong, J.; Rutland, C. J. Three Way Catalyst Modeling with Ammonia and Nitrous Oxide Kinetics for a Lean Burn Spark Ignition Direct Injection (SIDI) Gasoline Engine. SAE Tech Pap 2013.

429 430 431

(4)

Storey, J. M.; Barone, T.; Norman, K.; Lewis, S. Ethanol Blend Effects On Direct Injection Spark-Ignition Gasoline Vehicle Particulate Matter Emissions. SAE Int J Fuels Lubr 2010, 3 (2), 650–659.

432 433

(5)

Myung, C. L.; Park, S. Exhaust nanoparticle emissions from internal combustion engines: A review. Int J Automot Technol 2011, 13 (1), 9.

434 435

(6)

Gong, J.; Rutland, C. J. A Quasi-Dimensional NOx Emission Model for Spark Ignition Direct Injection (SIDI) Gasoline Engines. SAE Tech Pap 2013.

436 437 438 439

(7)

Zheng, Z.; Durbin, T. D.; Xue, J.; Johnson, K. C.; Li, Y.; Hu, S.; Huai, T.; Ayala, A.; Kittelson, D. B.; Jung, H. S. Comparison of Particle Mass and Solid Particle Number (SPN) Emissions from a Heavy-Duty Diesel Vehicle under On-Road Driving Conditions and a Standard Testing Cycle. Environ Sci Technol 2014, 48 (3), 1779–1786.

440 441

(8)

Gaddam, C. K.; Vander Wal, R. L. Physical and chemical characterization of SIDI engine particulates. Combust Flame 2013, 160 (11), 2517–2528.

442 443 444

(9)

Wang-Hansen, C.; Ericsson, P.; Lundberg, B.; Skoglundh, M.; Carlsson, P.-A.; Andersson, B. Characterization of Particulate Matter from Direct Injected Gasoline Engines. Top Catal 2013, 56 (1), 446–451.

ACS Paragon Plus Environment

25

Environmental Science & Technology

Page 26 of 29

445 446 447

(10)

Maricq, M. M.; Podsiadlik, D. H.; Chase, R. E. Gasoline Vehicle Particle Size Distributions:  Comparison of Steady State, FTP, and US06 Measurements. Environ Sci Technol 1999, 33 (12), 2007–2015.

448 449 450

(11)

Liu, Z. G.; Ford, D. C.; Vasys, V. N.; Chen, D.-R.; Johnson, T. R. Influence of Engine Operating Conditions on Diesel Particulate Matter Emissions in Relation to Transient and Steady-State Conditions. Environ Sci Technol 2007, 41 (13), 4593–4599.

451 452 453 454

(12)

Saffaripour, M.; Chan, T. W.; Liu, F.; Thomson, K. A.; Smallwood, G. J.; Kubsh, J.; Brezny, R. Effect of Drive Cycle and Gasoline Particulate Filter on the Size and Morphology of Soot Particles Emitted from a Gasoline-Direct-Injection Vehicle. Environ Sci Technol 2015, 49 (19), 11950–11958.

455 456

(13)

Richter, J. M.; Klingmann, R.; Spiess, S.; Wong, K.-F. Application of Catalyzed Gasoline Particulate Filters to GDI Vehicles. SAE Int J Engines 2012, 5 (3), 1361–1370.

457 458 459

(14)

Chan, T. W.; Meloche, E.; Kubsh, J.; Rosenblatt, D.; Brezny, R.; Rideout, G. Evaluation of a Gasoline Particulate Filter to Reduce Particle Emissions from a Gasoline Direct Injection Vehicle. SAE Int J Fuels Lubr 2012, 5, 1277–1290.

460 461

(15)

Johnson, T. V. Review of Vehicular Emissions Trends. SAE Int J Engines 2015, 8 (3), 2015-01–0993.

462 463 464

(16)

Kuwabara, S. The Forces Experienced by Randomly Distributed Parallel Circular Cylinders or Spheres in a Viscous Flow at Small Reynolds Number. J Phys Soc Japan 1959, 14 (4), 527–532.

465 466

(17)

Happel, J. Viscous flow in multiparticle systems: Slow motion of fluids relative to beds of spherical particles. AIChE J 1958, 4 (2), 197–201.

467 468

(18)

Rajagopalan, R.; Tien, C. Trajectory analysis of deep-bed filtration with the sphere-in-cell porous media model. AIChE J 1976, 22 (3), 523–533.

469 470

(19)

Lee, K. W.; Gieseke, J. A. Collection of aerosol particles by packed beds. Environ Sci Technol 1979, 13 (4), 466–470.

471 472

(20)

Konstandopoulos, A. G.; Johnson, J. H. Wall-Flow Diesel Particulate Filters-Their Pressure Drop and Collection Efficiency. SAE Tech Pap 1989.

473 474 475

(21)

Konstandopoulos, A. G.; Kostoglou, M.; Skaperdas, E.; Papaioannou, E.; Zarval, D.; Kladopoulou, E. Fundamental Studies of Diesel Particulate Filters: Transient Loading, Regeneration and Aging. SAE Tech Pap 2000.

476 477

(22)

Koltsakis, G. C.; Stamatelos, A. M. Modeling Thermal Regeneration of Wall-Flow Diesel Particulate Traps. AIChE J 1996, 42 (6), 1662–1672.

478 479

(23)

Bensaid, S.; Marchisio, D. L.; Fino, D.; Saracco, G.; Specchia, V. Modelling of diesel particulate filtration in wall-flow traps. Chem Eng J 2009, 154 (1–3), 211–218.

480

(24)

Tien, C.; Payatakes, A. C. Advances in deep bed filtration. AIChE J 1979, 25 (5), 737–

ACS Paragon Plus Environment

26

Page 27 of 29

Environmental Science & Technology

481

759.

482 483

(25)

Gong, J.; Rutland, C. J. Filtration Characteristics of Fuel Neutral Particulates Using a Heterogeneous Multiscale Filtration Model. J Eng Gas Turbines Power 2015, 137, 1–8.

484 485 486

(26)

Meng, Z.; Fang, J.; Pu, Y.; Yan, Y.; Wu, Y.; Wang, Y.; Song, Q. Experimental Study on the Influence of DPF Micropore Structure and Particle Property on Its Filtration Process. J Combust 2016, 2016.

487 488 489

(27)

Saito, C.; Nakatani, T.; Miyairi, Y.; Yuuki, K.; Makino, M.; Kurachi, H.; Heuss, W.; Kuki, T.; Furuta, Y.; Kattouah, P.; et al. New Particulate Filter Concept to Reduce Particle Number Emissions. SAE Tech Pap 2011.

490 491

(28)

Gong, J.; Rutland, C. J. PDF-Based Heterogeneous Multiscale Filtration Model. Environ Sci Technol 2015, 49 (4963), 4970.

492 493 494

(29)

Ogyu, K.; Ogasawara, T.; Sato, H.; Yamada, K.; Ohno, K. Development of High Porosity SiC-DPF Which is Compatible with High Robustness and Catalyst Coating Capability for SCR Coated DPF Application. SAE Tech Pap 2013.

495 496 497

(30)

Karin, P.; Cui, L.; Rubio, P.; Tsuruta, T.; Hanamura, K. Microscopic Visualization of PM Trapping and Regeneration in Micro-Structural Pores of a DPF Wall. SAE Int J Fuels Lubr 2009, 2 (1), 661–669.

498 499 500

(31)

Storey, J. M.; Lewis, S.; Szybist, J.; Thomas, J.; Barone, T.; Eibl, M.; Nafziger, E.; Kaul, B. Novel Characterization of GDI Engine Exhaust for Gasoline and Mid-Level GasolineAlcohol Blends. SAE Int J Fuels Lubr 2014, 7 (2), 571–579.

501 502

(32)

Wirojsakunchai, E.; Kolodziej, C.; Yapaulo, R.; Foster, D. E. Development of the Diesel Exhaust Filtration Analysis System (DEFA). SAE Int J Fuels Lubr 2008, 1 (1), 265–273.

503 504 505

(33)

Wirojsakunchai, E.; Schroeder, E.; Kolodziej, C.; Foster, D. E.; Schmidt, N.; Root, T. Characterization and Real-Time DPF Filtration Efficiency Measurements During PM Filling Process. SAE Tech Pap 2007, No. 724, 776–790.

506 507

(34)

Viswanathan, S.; Sakai, S.; Rothamer, D. Design & Evaluation of an Exhaust Filtration Analysis (EFA) System. SAE Tech Pap 2014.

508 509 510

(35)

Viswanathan, S.; Rothamer, D.; Sakai, S.; Hageman, M.; Foster, D.; Fansler, T.; Andrie, M. Effect of Particle Size Distribution on the Deep-Bed Capture Efficiency of an Exhaust Particulate Filter. J Eng Gas Turbines Power 2015, 137 (10), 101504.

511 512

(36)

Shrivastava, M.; Nguyen, A.; Zheng, Z.; Wu, H.-W.; Jung, H. S. Kinetics of Soot Oxidation by NO2. Environ Sci Technol 2010, 44 (12), 4796–4801.

513 514

(37)

Kandylas, I. P.; Haralampous, O. A.; Koltsakis, G. C. Diesel Soot Oxidation with NO2:  Engine Experiments and Simulations. Ind Eng Chem Res 2002, 41 (22), 5372–5384.

515 516

(38)

Stanmore, B. R.; Brilhac, J. F.; Gilot, P. The oxidation of soot: a review of experiments, mechanisms and models. Carbon N Y 2001, 39, 2247–2268, 2001.

ACS Paragon Plus Environment

27

Environmental Science & Technology

Page 28 of 29

517 518

(39)

Setiabudi, A.; Makkee, M.; Moulijn, J. A. The role of NO2 and O2 in the accelerated combustion of soot in diesel exhaust gases. Appl Catal B Environ 2004, 50 (3), 185–194.

519 520

(40)

Konstandopoulos, A. G.; Kostoglou, M.; Vlachos, N. The multiscale nature of diesel particulate filter simulation. Int J Vechicle Des 2006, 41, 256–284, 2006.

521 522

(41)

Hayashi, H.; Kubo, S. Computer simulation study on filtration of soot particles in diesel particulate filter. Comput Math with Appl 2008, 55 (7), 1450–1460.

523 524

(42)

Rakovec, N.; Viswanathan, S.; Foster, D. E. Micro-scale Study of DPF Permeability as a Function of PM Loading. SAE Int J Engines 2011, 4 (1), 913–921.

525 526

(43)

Serrano, J. R.; Climent, H.; Piqueras, P.; Angiolini, E. Filtration modelling in wall-flow particulate filters of low soot penetration thickness. Energy 2016, 112, 883–898.

527 528

(44)

Zhan, R.; Huang, Y.; Khair, M. Methodologies to Control DPF Uncontrolled Regenerations. SAE International 2006.

529 530

(45)

Gong, J.; Rutland, C. J. Pulsed Regeneration for DPF Aftertreatment Devices. SAE Tech Pap 2011.

531 532 533

(46)

Singh, N.; Rutland, C. J.; Foster, D. E. Investigation into Different DPF Regeneration Strategies Based on Fuel Economy Using Integrated System Simulation Kushal Narayanaswamy and Yongsheng He. SAE Tech Pap 2009.

534 535 536

(47)

Park, S. Y.; Narayanaswamy, K.; Schmieg, S. J.; Rutland, C. J. A model development for evaluating soot-NOx interactions in a blended 2-way diesel particulate filter/selective catalytic reduction. Ind Eng Chem Res 2012, 51 (48), 15582–15592.

537 538

(48)

Rutland, C. J.; England, S. B.; Foster, D. E. Integrated Engine, Emissions, and Exhaust Aftertreatment System Level Models to Simulate DPF Regeneration. SAE Tech Pap 2007.

539 540

(49)

Gong, J.; Rutland, C. J. Study the DPF Regeneration at Transient Operating Conditions Using Integrated System-Level Model. SAE Tech Pap 2010.

541 542

(50)

Bissett, E. J. Mathematical model of the thermal regeneration of a wall-flow monolith diesel particulate filter. Chem Eng Sci 1984, 39, 1233–1244, 1984.

543 544

(51)

Maricq, M. M.; Xu, N. The effective density and fractal dimension of soot particles from premixed flames and motor vehicle exhaust. J Aerosol Sci 2004, 35 (10), 1251–1274.

545 546 547

(52)

Dillon, H.; Stewart, M.; Maupin, G.; Gallant, T.; Li, C.; Mao, F.; Pyzik, A.; Ramanathan, R. Optimizing the Advanced Ceramic Material for Diesel Particulate Filter Applications. SAE Tech Pap 2007.

548 549

(53)

Bollerhoff, T.; Markomanolakis, I.; Koltsakis, G. Filtration and regeneration modeling for particulate filters with inhomogeneous wall structure. Catal Today 2012, 188 (1), 24–31.

550 551

(54)

Yamamoto, K.; Sakai, T. Effect of Pore Structure on Soot Deposition in Diesel Particulate Filter. Computation 2016, 4 (4), 46.

ACS Paragon Plus Environment

28

Page 29 of 29

Environmental Science & Technology

552 553 554 555

(55)

Gong, J.; Stewart, M.; Zelenyuk, A.; Viswanathan, S.; Rothamer, D. A.; Foster, D. E.; Rutland, C. J.; Narayanaswamy, K. Importance of Filter’s Microstructure in Dynamic Filtration Modeling of Gasoline Particulate Filters (GPFs): Inhomogeneous Porosity and Pore Size Distribution. submitted 2017.

556 557 558

(56)

Huynh, C. T.; Johnson, J. H.; Yang, S. L.; Bagley, S. T.; Warner, J. R. A OneDimensional Computational Model for Studying the Filtration and Regeneration Characteristics of a Catalyzed Wall-Flow Diesel Particulate Filter. SAE Tech Pap 2003.

559 560 561 562

(57)

Kato, H.; Ito, K.; Suda, H.; Kusaka, J.; Mori, T.; Tsurumi, F.; Masaki, N.; Hirata, K.; Akagawa, H. Development of a Quasi-Two-Dimensional Model for Analysing Continuous Regeneration-Diesel Particulate Filter States during Continuous and Active Regeneration. Int J Engine Res 2011, 12 (1), 1–13.

563 564

(58)

Opris, 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.

565 566

(59)

León y León, C. a. New perspectives in mercury porosimetry. Adv Colloid Interface Sci 1998, 76–77, 341–372.

567 568 569

(60)

Haralampous, O. A.; Kandylas, I. P.; Koltsakis, G. C.; Samaras, Z. C. Diesel particulate filter pressure drop Part 1 : modelling and experimental validation. Int J Engine Res 2004, 5 (2), 149–162.

570 571 572

(61)

Gong, J.; Rutland, C. J. Filtration Characteristics of Fuel Neutral Particulates Using a Heterogeneous Multiscale Filtration Model. Proc ASME 2014 Intern Combust Engine Div Fall Tech Conf 2014, 1–9.

573 574 575

(62)

Merkel, G. A.; Beall, D. M.; Hickman, D. L.; Vernacotola, M. J. Effects of Microstructure and Cell Geometry on Performance of Cordierite Diesel Particulate Filters. SAE Tech Pap 2001.

576

ACS Paragon Plus Environment

29