Integrating Filtration Mechanism with a 3D Diesel Particulate Filter - - PowerPoint PPT Presentation

Integrating Filtration Mechanism with a 3D Diesel Particulate Filter (DPF) Model using

Hoon Lee Center for Transportation Research Argonne National Laboratory

Orlando, Florida, USA March 19, 2013

Integrating Filtration Mechanism with a 3D Diesel Particulate Filter (DPF) Model using STAR-CCM+

 Objective  Background

 Diesel Engine & DPF  Two Approaches in Filtration Modeling

 Theoretical Analysis

 Pressure Drop Model  Soot Filtration Model

 Experiment  Model Setup

 Domain Setup & Meshing  Physical Assumptions & Boundary Conditions

 Filtration Algorithm

 Cell Value Localization  Built-in Function Utilization  Recursive Operation

 Computing Environment  Model Results

 Channel-flow Profiles & Pressure Drop  Wall-flow Rearrangements  Local Soot Mass Deposited  Local Collection Efficiency  Soot Cake Layer Properties

 Summary  Future Work  Acknowledgement

Outline

Integrating Filtration Mechanism with a 3D Diesel Particulate Filter (DPF) Model using STAR-CCM+

Objective

 To study quantitative analysis of soot filtration processes in DPF (diesel particulate filter) systems by developing a three dimensional model using a commercial CFD package, STAR-CCM+.  To analyze the time evolution and spatial distributions of local filtration parameters – e.g. porosity, soot mass, collection efficiency, soot cake profile - for each filtration period, along with evaluations of flow properties and pressure drop characteristics across the DPF.

Objective

Background Theoretical Analysis Experiment Computing Environment Model Setup Filtration Algorithm Model Results Summary Future Work Acknowledgement

Integrating Filtration Mechanism with a 3D Diesel Particulate Filter (DPF) Model using STAR-CCM+

4

Background (1/2)

Objective

Background

Theoretical Analysis Experiment Computing Environment Model Setup Model Results Summary Future Work Acknowledgement Filtration Algorithm

 Diesel Particulate Filter (DPF)

 Highway diesel vehicles are required to meet the stringent PM emission standards.  Physically trap (Filtration), and chemically

• xidize PM (Regeneration) periodically.

 Uncontrolled regeneration may occur which causes system failure due to highly exothermic reaction. Prediction of particulate deposition in the porous filter wall is important.

Plug Porous wall

 Diesel Engine

Pros: High Thermal Efficiency, Fuel Economy, Torque, Low Emission (CO, UHC) Cons: Noise, Vibration, $, Emission (PM, NOX)  NOX reduction by SCR and/or EGR  PM needs to be reduced in both mass and number... ☞ Arising issue for GDI engines, too! PM and NOX are the major emissions

• regulated. (USA: EPA Tier 4, EU: Euro 5 / 6)

Integrating Filtration Mechanism with a 3D Diesel Particulate Filter (DPF) Model using STAR-CCM+

 Two Approaches in Filtration Modeling

• H. Lee et al. 2012 DEER Conference
• H. Lee et al. SAE 2013-01-1583

Background (2/2)

Objective

Background

Theoretical Analysis Experiment Computing Environment Model Setup Model Results Summary Future Work Acknowledgement Filtration Algorithm

 Lagrangian: Qualitative analysis by tracking particle trajectories with appropriate B.C.s.

• S. Bensaid et al. Chem. Eng. J., 2009.
• Chem. Eng. Sci., 2010.
• P. Tandon et al. Chem. Eng. Sci., 2010.
• H. Kato et al. Int. J. Engine. Res., 2011.

 Eulerian: Quantitative analysis of soot filtration process by coupling specific filtration algorithms.

Integrating Filtration Mechanism with a 3D Diesel Particulate Filter (DPF) Model using STAR-CCM+

 Pressure Drop Model

Δ𝑄 = Δ𝑄𝑞𝑝𝑠𝑝𝑣𝑡 𝑥𝑏𝑚𝑚 + Δ𝑄𝑡𝑝𝑝𝑢 𝑑𝑏𝑙𝑓 + Δ𝑄

𝑔𝑠𝑗𝑑𝑢𝑗𝑝𝑜 + Δ𝑄𝑑𝑝𝑜𝑢/𝑓𝑦𝑞𝑏𝑜𝑡

= 𝜈 𝑙𝑝 𝑣𝑥𝑥𝑡 + 𝛾𝜍𝑣𝑥2𝑥𝑡 + 𝜈 𝑙𝑡𝑝𝑝𝑢 𝑣(𝑦)

𝑥

𝑒𝑦 + 𝜈𝐺 3𝑏2 𝑉𝑝,𝑗𝑜𝑀𝜊 + 𝜈𝐺 3𝑏2 𝑉𝑝,𝑝𝑣𝑢𝑀𝜊 + ζ𝑑𝑝𝑜𝑢 𝜍𝑣2 2 + ζ𝑓𝑦𝑞 𝜍𝑣2 2  Each velocity term is defined as,

𝑣𝑥 = 𝑅𝑝 𝐵𝑔𝑗𝑚𝑢 = 𝑉𝑝𝐵𝑝 4𝑏𝑀 = 𝑉𝑝𝑏2 4𝑏𝑀 = 𝑉𝑝𝑏 4𝑀 𝑣(𝑦)𝑒𝑦

𝑥

= 𝑅𝑝 𝐵𝑔𝑗𝑚𝑢(𝑦)

𝑥

𝑒𝑦 = 𝑅𝑝 4( 𝑏 − 2(𝑥 − 𝑦 )𝑀

𝑥

𝑒𝑦 = 𝑅𝑝 8𝑀 ln 𝑏 𝑏 − 2𝑥 𝑉𝑝,𝑗𝑜 = 𝑅 𝐵𝑝

𝑗𝑜𝑚𝑓𝑢

= 𝑅 𝜌𝐸2 4 1 2 1 2 (𝑏 − 2𝑥)2 (𝑏 + 𝑥𝑡)2 = 16𝑅 𝜌𝐸2σ(𝑏 − 2𝑥)2 𝑅𝑝 = 𝑅𝐵𝑝 𝐵𝑝

𝑗𝑜𝑚𝑓𝑢

= 𝑅(𝑏 − 2𝑥)2 𝜌𝐸2 4 1 2 1 2 (𝑏 − 2𝑥)2 (𝑏 + 𝑥𝑡)2 = 16𝑅(𝑏 + 𝑥𝑡)2 𝜌𝐸2 𝑣 = 𝑅 𝑂𝑏2

: Clean filter condition for Ao

𝑅𝑝, 𝑉𝑝, 𝑣𝑥, 𝑣(𝑦) 𝑅, 𝑉

Half-cut sample for experiment

a w ws a - 2w

Theoretical Analysis (1/2)

: Soot cake thickness (w) for Ao

Objective Background

Theoretical Analysis

Experiment Computing Environment Model Setup Model Results Summary Future Work Acknowledgement Filtration Algorithm

Darcy-Forchheimer’s Law

Integrating Filtration Mechanism with a 3D Diesel Particulate Filter (DPF) Model using STAR-CCM+

 Soot Filtration Model

 Unit Collector Mechanism  Collection Efficiency

𝒆𝒅𝟏 = 𝟒(𝟐 − 𝜻𝟏) 𝟑𝜻𝟏 𝒆𝒒𝒑𝒔𝒇 𝒆𝒅𝟏

𝟒

𝒄𝟒 = 𝟐 − 𝜻𝟏

Diffusional Deposition (𝜽𝑬) Flow-line Interception (𝜽𝑺)

𝑒𝑑(𝑗, 𝑢) = 2 3 4𝜌 𝑛𝑚𝑝𝑑𝑏𝑚(𝑗, 𝑢) 𝜍𝑡𝑝𝑝𝑢,𝑥𝑏𝑚𝑚 + 𝑒𝑑0 2

3 1 3

𝜁 𝑗, 𝑢 = 1 − 𝑒𝑑(𝑗, 𝑢) 𝑒𝑑0

3

(1 − 𝜁0) 𝑙 𝑗, 𝑢 = 𝑙0 𝑒𝑑(𝑗, 𝑢) 𝑒𝑑0

2 𝑔(𝜁 𝑗, 𝑢 )

𝑔(𝜁0) 𝛸 𝑢 = 𝑒𝑑(𝑗, 𝑢) 2 − 𝑒𝑑0

2

𝛺 𝑐 2 − 𝑒𝑑0

2

𝜃𝐸 = 3.5 𝑕 𝜁 𝑄𝑓−2

3 = 3.5 𝑕 𝜁

𝑉𝑗𝑒𝑑 𝐸

−2 3

𝜃𝑆 = 1.5 𝑂𝑆

2

𝑕 𝜁

3

1 + 𝑂𝑆

3−2𝜁 3𝜁

𝜃𝐸𝑆 = 𝜃𝐸 + 𝜃𝑆 − 𝜃𝐸𝜃𝑆 𝐹 𝑗, 𝑢 = 1 − 𝑓𝑦𝑞 − 3𝜃𝐸𝑆 1 − 𝜁 𝑗, 𝑢 Δ𝑧 2𝜁(𝑗, 𝑢) 𝑒𝑑(𝑗, 𝑢)

Key parameters for CFD code (UDF) to specify region properties

collector collector

Particulates

Key parameters for User Code to

• btain local soot

mass (mw)

Theoretical Analysis (2/2)

Objective Background

Theoretical Analysis

Experiment Computing Environment Model Setup Model Results Summary Future Work Acknowledgement Filtration Algorithm

Integrating Filtration Mechanism with a 3D Diesel Particulate Filter (DPF) Model using STAR-CCM+

 Clean Filter Test  Soot Loading Test

 2” x 6” cordierite DPF Test Results

ko=2.30E-13 ko=1.77E-13

PM Size Distribution PM Mass Concentration

SMPS TEOM

Experiment

 Clean filter permeability (ko) and particle-laden flow properties are directly measured.  Soot cake permeability (ks,cake), particle density (ρs), and soot cake porosity (εs,cake) can be estimated.  Packing densities (ρs,wall, ρs,cake) are assumed.

Ready to model

Objective Background Theoretical Analysis

Experiment

Computing Environment Model Setup Model Results Summary Future Work Acknowledgement Filtration Algorithm 0.0E+00 1.0E-13 2.0E-13 3.0E-13 4.0E-13 5.0E-13 6.0E-13 7.0E-13 8.0E-13 0.0E+00 2.0E-03 4.0E-03 6.0E-03

Permeability, ko [m2]

0.0 0.3 0.6 0.9 1.2 1.5 1.8 2.1 2.4 0.0E+00 2.0E-03 4.0E-03 6.0E-03

Pressure Drop [kPa]

• Vol. Flow Rate [m3/s]

0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0 10.0 10 20 30 40 50 60

Pressure Drop [kPa] Time [min.]

• Low flow rate (7.4 SCFM): 200CPSI
• High flow rate (9.0 SCFM): 200CPSI
• 100 CPSI (ws=17 mils)
• 200 CPSI (ws=12 mils)

◆ Analytical Solution ◆ 100 CPSI (ws=17 mils) ◆ 200 CPSI (ws=12 mils)

Integrating Filtration Mechanism with a 3D Diesel Particulate Filter (DPF) Model using STAR-CCM+

 Domain Setup

 Geometry is based on a 200CPSI, lab-scaled (2”x 6”) cordierite filter with regions of upstream flow and soot cake formation.

 Meshing

 Volume meshes are generated by using Trimmer for porous regions (filter wall, soot cake), and Polyhedral for fluid and solid regions (channels, plugs).

Model Setup (1/2)

a w ws

CFD domain

Total 2,013,762 cells Upstream=L 0.78”, Plug= L 0.39”, ws= 12.0 mils

Objective Background Theoretical Analysis Experiment Computing Environment

Model Setup

Model Results Summary Future Work Acknowledgement

POROUS REGION FLUID REGION SOLID REGION Filter wall Soot cake Upstream Plugs I/O Channels

10 9 8 7 6 5 4 3 2 1

 Trimmer is exclusively used for filter wall, consisting of 10 separate porous regions, to represent soot filtration. (Growth rate = 1, Cell size = thickness of each region) All cells in wall regions are regular hexahedrons

Filtration Algorithm

z y x

Integrating Filtration Mechanism with a 3D Diesel Particulate Filter (DPF) Model using STAR-CCM+

Model Setup (2/2)

 Physical Assumptions

• 1. Fluid: 3D, Ideal gas, Laminar, Incompressible
• 2. Implicit unsteady method (2nd order temporal discretization, Δ𝑢 = 0.05 sec)
• 3. Segregated flow & energy solver (2nd order convection scheme, URF= 0.5P, 0.2V)
• 4. Convective heat loss
• 5. No flow in the axial(z) direction in wall regions
• 6. Homogeneous distribution of particulates in the flow
• 7. Particle properties (dp=54.5 [nm], ρp=2.87 [g/cm3]) evaluated by experiments

 Boundary Conditions

Objective Background Theoretical Analysis Experiment Computing Environment

Model Setup

Model Results Summary Future Work Acknowledgement

Pressure Outlet 125.44 [kPa] (= 18.2 psi) Mass flow Inlet 3.05E-6 [kg/s] (= 7.4 SCFM) Convection Specify convection flux across the boundary to environment (ambient) Adiabatic Neglect heat loss for thermal condition at channel inlet Slip Define geometrically symmetry planes/surfaces ℎ 𝑑𝑝𝑜𝑤 = 𝑅 𝐵𝑓𝑟𝛦𝑈 = 𝑛 𝑑𝑞 𝑈𝑗 − 𝑈

𝑝

𝐵𝑓𝑟 𝑈 − 𝑈𝑏𝑛𝑐 𝑆𝑢,𝑔 = 1 𝜆 𝑥𝑠

Neglected Potential energy Chemical reactions Compression effect Expansion effect Plugging effect Soot Cake Transport Ash formation

Total Temperature 195.0 [C]

Filtration Algorithm

z y x

Integrating Filtration Mechanism with a 3D Diesel Particulate Filter (DPF) Model using STAR-CCM+

 Cell Value Localization

Filtration Algorithm (1/3)

Objective Background Theoretical Analysis Experiment Computing Environment Model Setup

Filtration Algorithm

Model Results Summary Future Work Acknowledgement

 Problem : To make each CFD cell acting as a unit collector, cell index must be ordered, so that the cell values can be transferred in certain direction.

Structured meshing is NOT allowed in standard CFD tools mlocal,1 (t) = min,1(t) E1(t) mlocal,2 (t) = min,2(t) E2(t) mlocal,i (t) = min,i(t) Ei(t) Inlet Outlet mcake (t)= min (t) 𝜲(t)

min,1 = min (1-Ф) min,2 = min,1 - mlocal,1 = min,1 (1-E1) min,i = min,i-1 (1-Ei-1)

Cake layer Layer 1 Layer 2 Layer i . . .

min (t) = χs Qs 𝜠t mout (t) = min,N(t) (1-EN(t)) z y x

0 ≤ ≤ 1 Soot Cake Mass Fraction

 Solution : Having the same cell indices in y direction by meshing each wall layers, separately.

Localiz ized d paramete ters

Integrating Filtration Mechanism with a 3D Diesel Particulate Filter (DPF) Model using STAR-CCM+

 Built-in Function Utilization

Filtration Algorithm (2/3)

Objective Background Theoretical Analysis Experiment Computing Environment Model Setup Model Results Summary Future Work Acknowledgement

Filtration Algorithm

 Problem : Classic unit collector mechanism causes a circulation error during initialization. Thus, Eq.(3) needs to be modified to account 𝑭 𝒋, 𝒖 − 𝟐 . ☞ …but, time array can NOT be handled through UDFs.

𝒆𝒅(𝒋, 𝒖) 𝜻 𝒋, 𝒖 𝑭 𝒋, 𝒖 𝒏𝒎𝒑𝒅𝒃𝒎 𝒋, 𝒖

Recall

𝑒𝑑(𝑗, 𝑢) = 2 3 4𝜌 𝑛𝑚𝑝𝑑𝑏𝑚(𝑗, 𝑢) 𝜍𝑡𝑝𝑝𝑢,𝑥𝑏𝑚𝑚 + 𝑒𝑑0 2

3 1/3

𝜁 𝑗, 𝑢 = 1 − 𝑒𝑑(𝑗, 𝑢) 𝑒𝑑0

3

(1 − 𝜁0) 𝐹 𝑗, 𝑢 = 1 − 𝑓𝑦𝑞 − 3𝜃𝐸𝑆 1 − 𝜁 𝑗, 𝑢 Δ𝑧 2ε(𝑗, 𝑢) 𝑒𝑑(𝑗, 𝑢) 𝑛𝑚𝑝𝑑𝑏𝑚 𝑗, 𝑢 = 𝑛𝑗𝑜 𝑗, 𝑢 𝐹(𝑗, 𝑢)

… 𝐅𝐫. (𝟐) … 𝐅𝐫. (𝟑) … 𝐅𝐫. (𝟒) … 𝐅𝐫. (𝟓)

𝑛𝑚𝑝𝑑𝑏𝑚 𝑗, 𝑢 = 𝑛𝑗𝑜 𝑗, 𝑢 𝑭(𝒋, 𝒖 − 𝟐)

diameter void fraction mass characteristic

 Solution : Store current (t) cell values using Table function, then access the data by interpolating the table as fields (UDF) at the next time step (t+1).

𝜻𝟏 𝒆𝒅𝟏 𝒏𝒎𝒑𝒅𝒃𝒎 = 𝟏 𝜻 𝒋, 𝟐 𝒆𝒅(𝒋, 𝟐) 𝒏𝒎𝒑𝒅𝒃𝒎 𝒋, 𝟐 𝑭 𝟏

Store Table

t=0 (Initialize) t=1

Access via UDF & User Code Access via UDF & User Code

𝒏𝒎𝒑𝒅𝒃𝒎 𝒋, 𝟑 𝑭 𝒋, 𝟐

Store

t=2

. . . . . .

Integrating Filtration Mechanism with a 3D Diesel Particulate Filter (DPF) Model using STAR-CCM+

Filtration Algorithm (3/3)

Objective Background Theoretical Analysis Experiment Computing Environment Model Setup Model Results Summary Future Work Acknowledgement

Filtration Algorithm

 Recursive Operation

 Problem : Flow changes with engine operating condition.  Local soot mass must be accumulated through time integral, considering collection efficiency. …but standard CFD code do NOT have ability to allow mathematical recursiveness through UDFs.

𝑛in = χs Qs 𝛦t 𝑛𝑗𝑜, 1= 𝑛𝑗𝑜 (1-Φ) 𝑛𝑑𝑏𝑙𝑓= 𝑛𝑗𝑜 Φ 𝑛𝑚𝑝𝑑𝑏𝑚, 1= (𝑛𝑗𝑜,1𝐹1)

𝑂 𝑢=1

𝑛𝑑𝑏𝑙𝑓 < 𝑛𝑑𝑏𝑙𝑓, 𝑛𝑏𝑦 𝑛𝑚𝑝𝑑𝑏𝑚, 1 < 𝑛𝑚𝑝𝑑𝑏𝑚, 𝑛𝑏𝑦 Φ = 1 No Yes 𝑛𝑚𝑝𝑑𝑏𝑚,𝑗 = (𝑛𝑗𝑜,𝑗𝐹𝑗 (1 − 𝐹𝑙)

𝑗−1 𝑙=1

)

𝑂 𝑢=1

𝑥 = 𝑛𝑑𝑏𝑙𝑓 1 𝐵𝑡𝜍𝑡,𝑑𝑏𝑙𝑓

𝑂 𝑢=1

Soot mass conc. Flow rate Partition factor (0 ≤ Φ ≤ 1) 𝝇𝒕,𝒙𝒃𝒎𝒎 𝜌 𝑂𝑒𝑑,𝐷𝐺𝐸 𝑒𝑑,𝑛𝑏𝑦

3 − 𝑒𝑑0 3

6 𝜍𝑡,𝑑𝑏𝑙𝑓𝑊

𝑑𝑏𝑙𝑓,𝐷𝐺𝐸

No (𝒖 + 𝜠t) Iteration

Get

𝜁, 𝐹, 𝑒𝑑

End of depth filtration

𝑛𝑑𝑏𝑙𝑓= 0

End of cake filtration

𝛺𝑐

Yes

 Solution : Couple User Code and Monitor function.

Link Compile Monitor Run User Code Field Sum

Object file (.obj, .so) Library (.dll) STAR-CCM+ Linux Linux Win Win

Load Load

Integrating Filtration Mechanism with a 3D Diesel Particulate Filter (DPF) Model using STAR-CCM+

Computing Environment

 Argonne TRACC (Transportation Research and Analysis Computing Center)

 A national user facility to meet US DOT advanced computation needs.  A focal point for computational research for transportation applications.  Linked to federal and non-federal R&D facilities, regional, state and city departments of transportation, and university research centers.

 High Performance Clusters

 Total 3,968 cores in 220 compute nodes Zephyr : 16 AMD 6273 (cores/CPU) * 2 (CPUs/node) * 92 (nodes) Phoenix : 4 AMD 2378 (cores/CPU) * 2 (CPUs/node) * 128 (nodes)

Objective Background Theoretical Analysis Experiment

Computing Environment

Model Setup Model Results Summary Future Work Acknowledgement Filtration Algorithm

Total solver elapsed time [s]

100 200 300 400 500 1 2 3 4 5 6 7 8 9 10

Iteration [#]

10 Iterations Benchmark Test

1 core: Local (3.4GHz, 16GB) 2 cores: Local 4 cores: Local 8 cores: Cluster (2.3GHz, 32GB)

16 cores: Cluster

Integrating Filtration Mechanism with a 3D Diesel Particulate Filter (DPF) Model using STAR-CCM+

Model Results (1/5)

Objective Background Theoretical Analysis Experiment Computing Environment Model Setup

Model Results

Summary Future Work Acknowledgement Filtration Algorithm

 Channel-flow Profiles

P L U G P L U G P L U G P L U G

 Pressure Drop Characteristics

 Pressure  Velocity

300s 600s 1000s

 Effect of percolation factor (ψ)

[kPa] [m/s]

125.0 126.0 127.0 128.0 129.0 130.0 131.0 0.2 0.4 0.6 0.8 1 0.0 10.0 20.0 30.0 0.0 10.0 20.0 30.0 0.2 0.4 0.6 0.8 1 Normalized Channel Length [-]

2.0 2.5 3.0 3.5 4.0 4.5 5.0 5.5 6.0 300 600 900 1200 1500 1800 Pressure Drop [kPa] Time [sec]

Experiment Model (ψ=0.9) Model (ψ=0.86)

Depth filtration Cake filtration Transition

Integrating Filtration Mechanism with a 3D Diesel Particulate Filter (DPF) Model using STAR-CCM+

Model Results (2/5)

Objective Background Theoretical Analysis Experiment Computing Environment Model Setup

Model Results

Summary Future Work Acknowledgement Filtration Algorithm

 Wall-flow Rearrangements

 Pre-transition regime  Post-transition regime

[m/s] [m/s] 30s 90s 120s Depth filtration 300s: Depth filtration end 600s: Transition 1000s: Cake filtration

0.020 0.028 0.036 0.044 0.052 0.060 0.2 0.4 0.6 0.8 1 Wall-through Velocity [m/s] Normalized Channel Length [-] 0.035 0.038 0.041 0.044 0.047 0.050 0.2 0.4 0.6 0.8 1 Normalized Channel Length [-]

0.000 0.020 0.040 0.060 30s 90s 120s

Velocity Avg.

1.0E-03 2.0E-03 3.0E-03 4.0E-03 300s 600s 1000s

• Std. Deviation

UNIFORMIZED ACCELERATED

Integrating Filtration Mechanism with a 3D Diesel Particulate Filter (DPF) Model using STAR-CCM+

Model Results (3/5)

Objective Background Theoretical Analysis Experiment Computing Environment Model Setup

Model Results

Summary Future Work Acknowledgement Filtration Algorithm

 Local Soot Mass Deposited

 y-z plane (@ x = 1/4a)  Deposited soot profiles  x-y plane (@ z = 1/2L)

0.0E+00 1.0E-14 2.0E-14 3.0E-14 4.0E-14 5.0E-14 6.0E-14 7.0E-14

50 100 150 200 250 300 Local Soot Mass [kg] Wall Penetration [μm]

Inlet Mid Outlet

0.0E+00 1.0E-14 2.0E-14 3.0E-14 4.0E-14 5.0E-14 6.0E-14 7.0E-14

50 100 150 200 250 300 Local Soot Mass [kg] Wall Penetration [μm]

Inlet Mid Outlet

0.0E+00 1.0E-14 2.0E-14 3.0E-14 4.0E-14 5.0E-14 6.0E-14 7.0E-14

50 100 150 200 250 300 Local Soot Mass [kg] Wall Penetration [μm]

Inlet Mid Outlet

90s: Depth filtration 600s: Transition 1000s: Cake filtration Filter wall Soot cake

Integrating Filtration Mechanism with a 3D Diesel Particulate Filter (DPF) Model using STAR-CCM+

Model Results (4/5)

Objective Background Theoretical Analysis Experiment Computing Environment Model Setup

Model Results

Summary Future Work Acknowledgement Filtration Algorithm

 Local Collection Efficiency (dP=54.5 nm)

 y-z plane (@ x = 1/4a)  x-y plane (@ z = 1/2L)

Streamlines 40% 50% 60% 70% 80% 90% 100% 200 400 600 800 1000 Local Collection Efficiency Time [sec]

Wall layer 1 Wall layer 3 Wall layer 5 Wall layer 7 Wall layer 9

Filter wall

Integrating Filtration Mechanism with a 3D Diesel Particulate Filter (DPF) Model using STAR-CCM+

Model Results (5/5)

Objective Background Theoretical Analysis Experiment Computing Environment Model Setup

Model Results

Summary Future Work Acknowledgement Filtration Algorithm

 Soot Cake Layer Properties

 Porosity (ρs,cake=120 kg/m3)  Thickness (@ 1000 s) Maintain 0.99↑ during first 1000 seconds of filtration

Integrating Filtration Mechanism with a 3D Diesel Particulate Filter (DPF) Model using STAR-CCM+

Summary

 A 3D CFD model was successfully developed for quantitative analysis

• f transient soot filtration processes in a wall-flow type DPF.

 The local value and rearrangement behaviors of each filtration parameter are well predicted within isotropically discretized meshes in the multi-layered porous wall regions.  Self-developed user subroutines, developed on basis of the unit collector mechanism, are integrated with the CFD code.  Built-in functions – Table, Monitor, UDF – were combined and fully coupled with algorithm to calculate the local value of soot mass and collection efficiency in the wall layer at each time step.  Results were visually demonstrated at the channel length scale in 3D, representing correlations among wall flow pattern, soot mass distribution, and soot cake profile.

Objective Background Theoretical Analysis Experiment Computing Environment Model Setup Model Results

Summary

Future Work Acknowledgement Filtration Algorithm

Integrating Filtration Mechanism with a 3D Diesel Particulate Filter (DPF) Model using STAR-CCM+

Future Work

 Modeling additional porous and fluid regions

 Create additional soot cake regions near the surface of the plugs to take into account plugging effects.  Create the downstream region to consider flow-expansion effects.

Objective Background Theoretical Analysis Experiment Computing Environment Model Setup Model Results Summary

Future Work

Acknowledgement

 Integrating PM oxidation reaction mechanisms

 Utilize soot filtration simulation results (soot mass distribution and soot cake profile) for the initial state of regeneration simulation.  Apply chemical kinetics of soot oxidation in consideration of the effects of O2, CO, NO2 and HCs (additional user subroutines need to be developed).

Filtration Algorithm

Integrating Filtration Mechanism with a 3D Diesel Particulate Filter (DPF) Model using STAR-CCM+

Acknowledgement

 Financial Support by US DOE - Office of Vehicle Technologies  Argonne TRACC

 Hubert Ley hley@anl.gov TRACC Director  Steven Lottes slottes@anl.gov Simulation, Modeling, Analysis Leader  Cezary Bojanowski cbojanowski@anl.gov Computational Mechanics Engineer  Waldemar Nowakowski wnowakowski@anl.gov System Administrator

Objective Background Theoretical Analysis Experiment Computing Environment Model Setup Model Results Summary Future Work

Acknowledgement

 CD-adapco

 Scott Wilensky scott.wilensky@cd-adapco.com East Region Technical Support Team Lead

Filtration Algorithm

Integrating Filtration Mechanism with a 3D Diesel Particulate Filter (DPF) Model using STAR-CCM+

For more information, please contact Hoon Lee at hoonlee@anl.gov