nox abatement - cleers€¦ · model vs brack (2016) urea decomposition urea mass decreases over...

32
Public AVL List GmbH (Headquarters) NOX ABATEMENT —1D/3D simulation of urea dosing and selective catalytic reduction J. C. Wurzenberger, A. Nahtigal, T. Mitterfellner, K. Pachler, S. Kutschi

Upload: others

Post on 21-Oct-2020

1 views

Category:

Documents


0 download

TRANSCRIPT

  • Public

    AVL List GmbH (Headquarters)

    NOX ABATEMENT

    —1D/3D simulation of urea dosing and

    selective catalytic reduction

    J. C. Wurzenberger, A. Nahtigal, T. Mitterfellner, K. Pachler, S. Kutschi

  • J.C. Wurzenberger | CDS | 14 September 2017 | 2Public

    WHY SIMULATION

    ▪ Urea SCR is the technology to reduce NOx emissions in HD applications

    ▪ NOx aftertreatment needs to deal with strongly transient operating conditions

    ▪ Deposit formation is a key aspect in the design if DEF dosing systems

    ▪ 3D phenomenon influenced by geometry, flow, control Candidate for 3D CFD

    ▪ Transient phenomenon influenced by the course of the operating conditions Candidate for Real-time system level simulation

    BACKGROUND

  • J.C. Wurzenberger | CDS | 14 September 2017 | 3Public

    DEPOSITS DEPEND ON:

    1. DESIGN

    2. INJECTOR

    3. BOUNDARY CONDITIONS

    All physical phenomena can

    be covered by 3D FIRE™

    in detail - but long

    simulation time is required.

    SOLUTION: Provide CFD-

    3D FIRE™ simulation

    results to Real – Time

    capable 1D BOOST™.

  • J.C. Wurzenberger | CDS | 14 September 2017 | 4Public

    CONTENT

    1. Model / Model Validation

    i. DEF Dosing (CFD)

    ii. Drive Cycle Performance (SysEng)

    2. Tools and Workflows

    3. Use Case

    4. Summary

  • J.C. Wurzenberger | CDS | 14 September 2017 | 5Public

    SPRAY PREPARATION MODELS

    ▪ Injection of urea-water solution▪ Urea-water properties: f(T,wi,g)

    ▪ Nozzle modeling

    DEF DOSINGINJECTION OF UREA-WATER SOLUTION

  • J.C. Wurzenberger | CDS | 14 September 2017 | 6Public

    SPRAY PREPARATION MODELS

    ▪ Injection of urea-water solution▪ Urea-water properties: f(T,wi,g)

    ▪ Nozzle modeling

    ▪ Spray/gas interaction▪ Multicomponent evaporation

    ▪ Thermolysis: (NH2)2CO NH3 + HNCO

    ▪ Hydrolysis: HNCO + H2O NH3 + CO2

    DEF DOSINGSPRAY/GAS INTERACTION

    H 0 (l)2

    H 0 (g)2

    H 0 (l)2

    H 0 (g)2

    (NH ) CO (s or l)2 2

    (NH ) CO (g)2 2

    NH (g) + HNCO (g)3

    H 02 (NH ) CO2 2

    I. II. III.

    (NH ) CO (s or l)2 2

    (NH ) CO (s or l)2 2

    (NH ) CO (g)2 2

    (NH ) CO (g)2 2

    H 0 (l)2

    H 0 (g)2

    H 0 (l)2

    H 0 (g)2

    (NH ) CO (s or l)2 2

    (NH ) CO (g)2 2

    NH (g) + HNCO (g)3

    H 02 (NH ) CO2 2

    I. II. III.

    (NH ) CO (s or l)2 2

    (NH ) CO (s or l)2 2

    (NH ) CO (g)2 2

    (NH ) CO (g)2 2

    H2O

    (NH2)2CO NH3

    HNCO

    NH3

    CO2

  • J.C. Wurzenberger | CDS | 14 September 2017 | 7Public

    SPRAY PREPARATION MODELS

    ▪ Injection of urea-water solution▪ Urea-water properties: f(T,wi,g)

    ▪ Nozzle modeling

    ▪ Spray/gas interaction▪ Multicomponent evaporation

    ▪ Thermolysis: (NH2)2CO NH3 + HNCO

    ▪ Hydrolysis: HNCO + H2O NH3 + CO2

    DEF DOSINGSPRAY/GAS INTERACTION

    H2O

    (NH2)2CO NH3

    HNCO

    NH3

    CO2

    VelocityNH3

    Uni=0.84 =0.94

  • J.C. Wurzenberger | CDS | 14 September 2017 | 8Public

    SPRAY PREPARATION MODELS

    ▪ Injection of urea-water solution▪ Urea-water properties: f(T,wi,g)

    ▪ Nozzle modeling

    ▪ Spray/gas interaction▪ Multicomponent evaporation

    ▪ Thermolysis: (NH2)2CO NH3 + HNCO

    ▪ Hydrolysis: HNCO + H2O NH3 + CO2

    ▪ Spray/wall interaction▪ Heat transfer between spray and wall

    ▪ Wallfilm formation

    ▪ Multicomponent film evaporation & thermolysis

    DEF DOSINGSPRAY/WALL INTERACTION

    H2O

    (NH2)2CO NH3

    HNCO

    NH3

    CO2

  • J.C. Wurzenberger | CDS | 14 September 2017 | 9Public

    DEF DOSINGSPRAY/WALL INTERACTION -- VALIDATION

    Nahtigal et al. “SCR recent development and method”, AVL International Simulation Conference, 2017

    Test bench at Graz University of Technology

  • J.C. Wurzenberger | CDS | 14 September 2017 | 11Public

    SPRAY PREPARATION MODELS

    ▪ Injection of urea-water solution▪ Urea-water properties: f(T,wi,g)

    ▪ Nozzle modeling

    ▪ Spray/gas interaction▪ Multicomponent evaporation

    ▪ Thermolysis: (NH2)2CO NH3 + HNCO

    ▪ Hydrolysis: HNCO + H2O NH3 + CO2

    ▪ Spray/wall interaction▪ Heat transfer between spray and wall

    ▪ Wallfilm formation

    ▪ Multicomponent film evaporation & thermolysis

    ▪ Cooling of walls▪ Radial and lateral heat transfer (walls, mixers,…)

    DEF DOSINGCOOLING OF WALLS

  • J.C. Wurzenberger | CDS | 14 September 2017 | 12Public

    SPRAY PREPARATION MODELS

    ▪ Injection of urea-water solution▪ Urea-water properties: f(T,wi,g)

    ▪ Nozzle modeling

    ▪ Spray/gas interaction▪ Multicomponent evaporation

    ▪ Thermolysis: (NH2)2CO NH3 + HNCO

    ▪ Hydrolysis: HNCO + H2O NH3 + CO2

    ▪ Spray/wall interaction▪ Heat transfer between spray and wall

    ▪ Wallfilm formation

    ▪ Multicomponent film evaporation & thermolysis

    ▪ Cooling of walls▪ Radial and lateral heat transfer (walls, mixers,…)

    ▪ Catalytic conversion▪ Ad- and desorption, fast, standard, slow SCR,…

    DEF DOSINGCATALYTIC CONVERSION

    H2O

    (NH2)2CO NH3

    HNCO

    NH3

    CO2

  • J.C. Wurzenberger | CDS | 14 September 2017 | 13Public

    CONTENT

    1. Model / Model Validation

    i. DEF Dosing (CFD)

    ii. Drive Cycle Performance (SysEng)

    2. Tools and Workflows

    3. Use Case

    4. Summary

  • J.C. Wurzenberger | CDS | 14 September 2017 | 14Public

    SYSTEM ENGINEERING MODELDRIVER, VEHICLE, DRIVELINE ENGINE, COOLING, CONTROL …

    Multi-physics model

    Multi-rate numeric

    Diesel Exhaust System

    Dedicated coupling technique for engine thermodynamics and EAS modeling

  • J.C. Wurzenberger | CDS | 14 September 2017 | 15Public

    0D/1D DOSING MODEL

    ▪ Arbitrary liquid/gas mixtures

    ▪ Instantaneous evaporation gas phase▪ Break-up model (DEF2NH3+CO2+6H2O)

    ▪ Spraying liquid droplets

    ▪ Split is empirical parameterized or from CFD

    ▪ Droplets▪ Passive transport

    ▪ Deposition following “adsorption chemistry”

    ▪ Wall film▪ Heat-transfer: wall, wall film and gas phase

    ▪ Multi-component evaporation f(Tw, pi,g, Re)

    ▪ Arbitrary film reaction chemistry (i.e. decomposition of urea)

    SYSTEM ENGINEERING MODELDOSING, EVAPORATION, WALL FILM …

    Qz

    vw

    t

    w

    gpassi,gpassi,g

    emptys,passi,depdep Zwkr

    injection

    surfaceu w

    wutransport

    wu

    u b t c adecomposition

    HNCO + NH3H2O

    Instantaneous evaporation

    transport

    deposition

    H2O, HNCO, NH3

  • J.C. Wurzenberger | CDS | 14 September 2017 | 16Public

    MODEL BRACK (2014*, 2016†)

    ▪ Rates are parameterized using experimental data from TGA measurements

    ▪ Translated into open User-Coding models

    ▪ General rate:

    ▪ CYA decomp. rate:

    ▪ HNCO evap. rate:

    ▪ AR, VR: estimated to match published data

    ▪ Validated using published data

    SYSTEM ENGINEERING MODELWALL FILM – UREA DECOMPOSITION MODEL

    1) CYA(s) 3 HNCO(g)2) biuret(m) urea(m) +HNCO(l)3) urea(m) +HNCO(l) biuret(m)4) urea(m) HNCO(l) +NH3(g)5) 2 biuret(m) ammelide(s) +HNCO(l) +NH36) biuret(m) +HNCO(l) CYA(s) +NH3(g)7) biuret(m) +HNCO(l) triuret(s)8) triuret(s) CYA(s) +NH3(g)9) urea(m) +2HNCO(l) ammelide(s) +H2O(g)10) biuret(m) biuret(matrix)11) biuret(matrix) biuret(m)12) biuret(matrix) 2 HNCO(g) +NH313) urea(s) urea(m)14) ammelide(s) ammelide(g)15) HNCO(l) HNCO(g)

    * Brack, W.; Heine, B.; Birkhold, F.; Kruse, M.; Schoch, G.; Tischer, S. & Deutschmann, O., Chemical Engineering Science, 2014, 106, 1-8

    † Brack, W.; Heine, B.; Birkhold, F.; Kruse, M. & Deutschmann, O., Emission Control Science and Technology, 2016, 1-9

    biuret

    triuret

    cyanuric acid ammelideurea

  • J.C. Wurzenberger | CDS | 14 September 2017 | 17Public

    MODEL VS BRACK (2014)

    ▪ CYA decomposition▪ CYA(s) 3 HNCO(g) (0th order rate!)

    ▪ TGA Experiment: 6mg CYA heated at different heating rates

    ▪ End-of-decomposition temperature increases with increasing heating rate

    ▪ Good match with published data

    ▪ Biruet decomposition▪ All 15 reactions

    ▪ TGA Experiment: ~50mg biuret heated at 2K/min

    ▪ Initial decomposition to major amounts of CYA, minor amounts of ammelide, which, in turn decompose at higher temperatures

    ▪ Good match with published data

    SYSTEM ENGINEERING MODELVALIDATION – CYA & BIURET DECOMPOSITION

    CYA decomposition

    Biuret decomposition

  • J.C. Wurzenberger | CDS | 14 September 2017 | 18Public

    MODEL VS BRACK (2016)

    ▪ Urea Decomposition

    ▪ All 15 reactions

    ▪ Simulation: initial urea decomposed at different constant temperatures for 180 min assuming two different film thicknesses

    ▪ Brack presents several contour plots gained from simulation data like the own shown here

    ▪ To compare them cuts at 5 temperatures were made to gather data points for comparison

    SYSTEM ENGINEERING MODELVALIDATION – UREA DECOMPOSITION (SETUP)

  • J.C. Wurzenberger | CDS | 14 September 2017 | 19Public

    MODEL VS BRACK (2016)

    ▪ Urea Decomposition

    ▪ Urea mass decreases over time

    ▪ Fasted decomposition at highest temperature

    ▪ Incomplete decomposition for all temperatures, in the given time span, is in line with the reference data

    ▪ Effect of changing film thickness (model input parameter) is reflected accurately

    ▪ Reasonable qualitative agreement between simulated data from Brack and BOOST, for both cases

    SYSTEM ENGINEERING MODELVALIDATION – UREA DECOMPOSITION (TOTAL MASS)

    Total mass decrease

    173 µm film thickness

    750 µm film thickness

    Points: data from contour plots x-cutsLines: BOOST simulation

  • J.C. Wurzenberger | CDS | 14 September 2017 | 20Public

    MODEL VS BRACK (2016)

    SYSTEM ENGINEERING MODELVALIDATION – UREA DECOMPOSITION (SPECIES F1)

    Species mass fractions

    173 µm film thickness

    Points: data from contour plots x-cutsLines: BOOST simulation

  • J.C. Wurzenberger | CDS | 14 September 2017 | 21Public

    MODEL VS BRACK (2016)

    SYSTEM ENGINEERING MODELVALIDATION – UREA DECOMPOSITION (SPECIES F2)

    Species mass fractions

    750 µm film thickness

    Points: data from contour plots x-cutsLines: BOOST simulation

  • J.C. Wurzenberger | CDS | 14 September 2017 | 22Public

    CONTENT

    1. Model / Model Validation

    i. DEF Dosing (CFD)

    ii. Drive Cycle Performance (SysEng)

    2. Tools and Workflows

    3. Use Case

    4. Summary

  • J.C. Wurzenberger | CDS | 14 September 2017 | 23Public

    EAS TOOLS –REACTION MODELING

    User Coding Interface

    Reaction/Transfer/Diffusion

    RT-Solver /1D, 2D, 1D+1D/ for catalysts, filters, pipes, dosers,… EAS systems

    𝑉 ⋅𝜕𝜌

    𝜕𝑡= −

    𝜕 ሶ𝑚

    𝜕𝑧⋅ d𝑧 +MG𝑖 ⋅𝜈𝑖,𝑗 ⋅ ሶ𝑟𝑗 | 0 =

    d𝑝

    d𝑧− 𝜁 ⋅

    1

    2⋅ 𝜌 ⋅ 𝑣2

    𝑉 ⋅𝜕𝜌 ⋅ 𝑤𝑖𝜕𝑡

    = −𝜕 ሶ𝑚 ⋅ 𝑤𝑖𝜕𝑧

    ⋅ d𝑧 + D ⋅ 𝜌 ⋅ AC ⋅𝜕2𝑤𝑖𝜕𝑧2

    ⋅ d𝑧 − 𝛽 ⋅ 𝜌 ⋅ AW ⋅ 𝑤𝑖 − 𝑤𝑖,L

    𝑉 ⋅𝜕𝜌 ⋅ 𝑢

    𝜕𝑡= −

    𝜕 ሶ𝑚 ⋅ ℎ

    𝜕𝑧⋅ d𝑧 + 𝜆 ⋅ AC ⋅

    𝜕2𝑇

    𝜕𝑧2⋅ d𝑧 − 𝛼 ⋅ AW ⋅ 𝑇 − 𝑇W

    𝑚L ⋅𝜕𝑤𝑖,L𝜕𝑡

    = 𝛽 ⋅ 𝜌 ⋅ AW ⋅ 𝑤𝑖 −𝑤𝑖,L +MG𝑖 ⋅𝜈𝑖,𝑗 ⋅ ሶ𝑟𝑗

    𝑚W ⋅ 𝑐p,W ⋅𝜕𝑇W𝜕𝑡

    = 𝜆W ⋅ AW ⋅𝜕2𝑇𝑊𝜕𝑧2

    ⋅ d𝑧 + AW ⋅ 𝛼 ⋅ 𝑇 − 𝑇W +Δℎ𝑗 ⋅ ሶ𝑟𝑗 | 𝑍𝑛d𝑤𝑘,𝑛,Sd𝑡

    = MG𝑘 ⋅𝜈𝑘,𝑗 ⋅ ሶ𝑟𝑚

    Execution Environments

    MyReac.ucp

    MyTrans.ucp

    MyDiff.ucp

    MyModel.fmu

    MyModel.zip

    Kinetics/Transfer

    Library• ASC: Scheuer

    • LNT: Olsson

    • SCR: Olsson

    • SCR: Ebrahimian/Brack

    • TWC: Brinkmeier

    • DPF: Konstandopoulos

    • CSF: Premchand

    • XXX: Surface Chemkin

    • …

    This

    study

  • J.C. Wurzenberger | CDS | 14 September 2017 | 24Public

    1D / 3D SIMULATION WORKFLOW

    Tasks

    Tools

    Results

    Component Design

    3D CFD

    EAS

    Species uniformity, wall

    film mass and

    position,…

    Concept Layout

    Virtual Testing

    1D EAS

    Optimization

    EAS Layout (drive cycle

    emissions, component

    performance,

    durability….)

    Reaction modeling/

    parameterization

    1D EAS

    Coding Interface

    Optimization

    Model

    (kinetic, transfer,

    diffusion…)

    Control Develepment

    /Calibration

    1D EAS

    Control strategy

    Control params.

  • J.C. Wurzenberger | CDS | 14 September 2017 | 25Public

    CONTENT

    1. Model / Model Validation

    i. DEF Dosing (CFD)

    ii. Drive Cycle Performance (SysEng)

    2. Tools and Workflows

    3. Use Case

    4. Summary

  • J.C. Wurzenberger | CDS | 14 September 2017 | 26Public

    ▪ HD exhaust line

    ▪ Simplified Geometry

    ▪ Artificial mixer geometry

    ▪ Fame Poly Mesh

    ▪ 4.200.000 cells

    ▪ w/o wall film reactions

    ▪ w/ wall film evaporation

    COMPONENT DESIGNDEF DOSING – MODEL SETUP

    Mixer

    Tg

    (degC)

    mg(kg/h)

    mDEF(g/s)

    Nr. pulses

    (-)

    Case 220 220 2000 2 10

    Case 270 270 2000 2 10

    Case 370 370 2000 2 10

  • J.C. Wurzenberger | CDS | 14 September 2017 | 27Public

    COMPONENT DESIGNDEF DOSING – 3D SIMULATION RESULTS

    Urea dosing – snapshot of animation

    Wall FilmCase 370

    Wall FilmCase 220

    Wall FilmCase 270

  • J.C. Wurzenberger | CDS | 14 September 2017 | 28Public

    DISCUSSION

    ▪ Most of the injected DEF vaporizes in the gas phase

    ▪ Wall film formation declines with increasing temperature

    ▪ Simulation of 10 pulse gives a trend, full steady-state is not reached

    ▪ Results for 1D▪ Deposition split ratio

    ▪ Film thickness

    ▪ Frozen flow field* technology enables a significant speed-up to conventional CFD

    COMPONENT DESIGNDEF DOSING – 2D SIMULATION RESULTS

    Wall Film Mass

    0

    5

    10

    15

    20

    Liq

    uid

    ma

    ss in

    jecte

    d (

    g)

    0 2 4 6 8 10

    Time (s)

    9.95

    20

    3.8755

    2.48917

    0.741679

    Transient_220 - Injected mass

    Transient_220 - WF mass

    Transient_270 - WF mass

    Transient_370 - WF mass

    Gas

    temp.

    (degC)

    Mass

    flow

    (kg/h)

    Dosing

    rate

    (g/s)

    Nr.

    pulses

    Film

    mass

    (%)

    CPU

    time**

    [h]

    Case 220 220 2000 2 10 19.3 243

    Case 270 270 2000 2 10 12.45 151

    Case 370 370 2000 2 10 3.7 106

    * Schellander D., Pachler K., Schmalhorst C., Nahtigal A., “Predictive Numerical Models and Methods for Selective Catalytic Reactor Applications in Diesel Powered Vehicles”, COMODIA 2017

    ** CPU time for 10s physical time on Linux cluster on 64 cores (less WF accumulation -> faster sim.)

  • J.C. Wurzenberger | CDS | 14 September 2017 | 29Public

    SUMMARY

    ▪ AT Model: DOC, injector, pipes, SCR, AMOX

    ▪ Engine out data from test bed (measured drive cycle)

    ▪ Injection mass flux: Controlled =1.1▪ Injection shut off: TSD_SCR ≥ 180°C▪ DEF split is taken from CFD▪ ~75% instantaneous evaporation

    ▪ ~25% wall film

    ▪ Passive species are deposited(= converted to surface species) in dedicated pipe

    CONCEPT SIMULATIONHD EXHAUST LINE – MODEL SETUP

    Start of injection@Ts(SCR) = 180°C

  • J.C. Wurzenberger | CDS | 14 September 2017 | 30Public

    SUMMARY

    ▪ SCR operating conditions are reached after ~120s

    ▪ Tailpipe NOx levels out after cold start and shows slight increase for the remaining simulation time

    ▪ NH3 conversion▪ SCR: 85%

    ▪ AMOX: 100%

    ▪ NOx conversation: 93%

    CONCEPT SIMULATIONHD EXHAUST LINE – TAILPIPE EMISSIONS

  • J.C. Wurzenberger | CDS | 14 September 2017 | 31Public

    SUMMARY

    ▪ Results depend on assumed wall film thickness60 µm: avg. film thickness from CFD173 µm: moving film (Brack2016)

    ▪ after 60 min:

    ▪ thin film: less deposits, formed CYA decomposes again fully

    CONCEPT SIMULATIONHD EXHAUST LINE – DEPOSIT FORMATION

    dfilm / µm CYA / mg ammelide / mg Total / mg

    60 - 150 150

    173 320 380 700

  • J.C. Wurzenberger | CDS | 14 September 2017 | 32Public

    SUMMARY

    ▪ 1/3 pipe insulation lower decomposition temperatures (~25°C)

    ▪ after 60 min:

    ▪ higher total deposit mass

    CONCEPT SIMULATIONHD EXHAUST LINE – DEPOSIT FORMATION

    dfilm / µm CYA / mg ammelide / mg Total / mg

    60 - 200 200

    173 710 410 1120

  • J.C. Wurzenberger | CDS | 14 September 2017 | 33Public

    ▪ 3D DEF dosing model validated in various sub-models

    ▪ 1D real-time, system engineering model urea decomposition model from Brack

    ▪ 3D-1D workflow

    ▪ 3D simulations provide dosing split ratio and film thickness

    ▪ 1D simulations show strong impact of film thickness (film surface area) on urea decomposition

    SUMMARY