the pennsylvania state university...the pennsylvania state university may 2001 gm 6.6l duramax ohv...

Fuchs & Rutland,SAE 980508

Dan HaworthThe Pennsylvania State University

May 2001

GM 6.6L DuramaxOHV V-8

Flynn et al.,SAE 1999-01-0509

Scope Device-scale 3D time-dependent CFD Emphasis on direct-injection diesel engines

Caveat Mainly others work

Acknowledgements Profs. Rolf Reitz and Chris Rutland

Engine Research Center, University of Wisconsin-Madison

Prof. Norbert Peters ITM/RWTH, Aachen, Germany (Stanford University)

Homogeneous Charge Spark Ignition (HCSI) Stratified Charge Compression Ignition (SCCI) Stratified Charge Spark Ignition (SCSI) Homogeneous Charge Compression Ignition (HCCI)

T. Baritaud (Ed.), Multi-Dimensional Simulation of Engine Internal Flows,Oil & Gas Science and Technology: Revue de lInstitut Français du Pétrole,

Vol. 54, No. 2 (1999)

HCSI, SCCI, and SCSI Engines Numerical Methodology Physical Modeling Emissions Prediction

NOX and soot

Turbulent Hydrodynamics Gas-Phase Mixing (Fuel, Air, Residual) Two-Phase Flows (Fuel Sprays) Turbulent Combustion

Heat release Emissions

Other Ignition Heat transfer . . .

Formulation Ensemble averaged equations Two-equation models (mainly k-ε and variants) Wall functions Effective turbulent Prandtl and Schmidt numbers

Strengths Single cycle yields ensemble average (in principle)

Limitations All fluctuations about ensemble mean modeled Cycle-to-cycle variations, transients problematic

Alternatives Higher-order conventional closures (e.g., RSM) Large-eddy simulation (LES)

Formulation Lagrangian particle methods Droplet-distribution-function-based models Physical processes:

Injection/breakup/coalescence/vaporization (single-/multi-component) Gas/liquid coupling (mass, momentum, energy, species) Spray-wall interactions/liquid wall films

Strengths Concentrates particles where needed (adaptive) Enables implementation of comprehensive physical models

Limitations Physics often poorly understood Current models require extensive tuning/calibration to engine, fuel

injector, operating conditions, numerics (e.g., mesh size), . . . Alternatives

Eulerian moment methods

References Veynante & Vervisch, Turbulent combustion

modeling, submitted to Prog. Energy Combust. Sci., 2001. Also in von Karman Institute Lecture Series: Turbulence and Combustion, 2001; to appear.

Haworth, Applications of turbulent combustion modeling, von Karman Institute Lecture Series: Turbulence and Combustion, 2001; to appear.

Saturn 2.2LL-4

If: Steady-state operation Warmed-up engine Homogeneous mixture,

Then: Principal issue is low-speed part-load combustion

efficiency vs. high-speed wide-open-throttle power Turbulent hydrodynamics dominates Intake-port/in-cylinder region can be considered in

isolation

1≈Φ

Discharge Coefficient Flow losses

Swirl and Tumble Large-scale flow structure

Spark-Gap Velocity Flame advection

Late-Compression Turbulence Burn rate

Mixing Fuel/air/residual

Burn-Rate Curves

CD

Swirl orTumble

RealEngines

Inaccessible

Optimum

fasterburn

higher peak power

Scaling (IC Engine) Burn rate is proportional to crankshaft rotational speed

Empirical Evidence Mantzaras, Felton & Bracco SAE 881635 Ziegler et al. SAE 881634

LCSLPLLTLT SSVSuAASS //~/1~// Ω∝′+∝

Unburned Burned

Tu

Tb

<T(x)>

Tb

Tu

T(x)

TDCTmin = 600 K

Tmax = 2000 K

Flame front

Crank Angle Degrees (360=TDC Intake)

Tum

ble

Rat

io

Flamelet Models Represent the Correct Physics for Flame Propagation and Heat Release

Models Are Reasonably Pre-dictive Volumetric efficiency, global flow structure, burn rate Robust nearly homogeneous mixtures

Emissions Usually Are Not Modeled Emissions Models Are Available

CO: equilibrium at specified freeze-out temperature NOX: thermal (Zeldovich, extended Zeldovich) UHC: crevice and oil film adsorption/desorption (non-

combustion processes) PM: negligible

High Dilution (EGR) and/or Highly Lean Detailed Chemistry

UHC speciation, alternative fuels, fuel additives Cycle to Cycle Variations Transient Operating Conditions Ignition Flame/Wall Interactions Crevice Flows Knock

GM 6.6L DuramaxOHV V-8

Fuchs & Rutland,SAE 980508

Flynn et al.,SAE 1999-01-0509

Flynn et al., Diesel combustion: an integrated view combining laser diagnostics, chemical kinetics, and empirical validation, SAE 1999-01-0509 (1999)

T. Baritaud (Ed.), Multi-Dimensional Simulation of Engine Internal Flows,Oil & Gas Science and Technology: Revue de lInstitut Français du Pétrole,

Vol. 54, No. 2 (1999) Barths, Pitsch & Peters (pp. 233-244)

Unsteady non-premixed flamelet model 118 species, 557 reactions for autoignition, heat release, NOX, soot precursors

Belardini & Bertoli (pp. 251-257) Ignition delay correlation, one-step global fuel oxidation (Arrhenius/EBU) Extended Zeldovich NO (Arrhenius/equilibrium/steady-state), six-step soot

(Arrhenius/EBU) Magnusson (pp. 293-296)

EBU heat release Detailed soot model, Zeldovich NO with source terms from flamelet library

Taklanti & Delhaye (pp. 271-277) Shell model autoignition (Arrhenius), EBU heat release Zeldovich NO, simple soot formation/oxidation

Configurations Various engines, sector models, computations begin post-IVC Variations in engine speed, load, SOI, injection history, EGR, fuel type, . . .

Barths, Pitsch & Peters, in T. Baritaud (Ed.), Multi-Dimensional Simulation of Engine Internal Flows, Oil & Gas Science and Technology: Revue de lInstitut Français du Pétrole, Vol. 54, No. 2, pp. 233-244 (1999).

Note: computed NOX values

multiplied by 6.5

Belardini & Bertoli, in T. Baritaud (Ed.), Multi-Dimensional Simulation of Engine Internal Flows, Oil & Gas Science and

Technology: Revue de lInstitut Français du Pétrole, Vol. 54, No. 2, pp. 251-257 (1999).

Magnusson, in T. Baritaud (Ed.), Multi-Dimensional Simulation of Engine Internal Flows, Oil & Gas Science and

Technology: Revue de lInstitut Français du Pétrole, Vol. 54, No. 2, pp. 293-296 (1999).

Taklanti & Delhaye, in T. Baritaud (Ed.), Multi-Dimensional Simulation of Engine Internal Flows, Oil & Gas

Science and Technology: Revue de lInstitut Français du Pétrole, Vol. 54, No. 2, pp. 271-277 (1999).

Reference Yi, Hessel, Zhu & Reitz, SAE 2000-01-1178, 2000

Combustion Modeling Shell autoignition (Arrhenius) Characteristic time combustion (Arrhenius/EBU) Extended Zeldovich NO (Arrhenius/equilibrium/steady state) Two equation soot formation/oxidation

Configuration Heavy-duty DI diesel (Cat 3401) Sector model, calculations begin post-IVC

Approach Model calibration over range of conditions Sensitivity analysis to variations in key engine parameters

SAE 2000-01-1178

SAE 2000-01-1178

SAE 2000-01-1178

Reference Fuchs & Rutland, SAE 980508, 1998

Combustion Modeling Shell autoignition (Arrhenius) Characteristic time combustion (Arrhenius/EBU) Extended Zeldovich NO (Arrhenius/equilibrium/steady state) Two equation soot formation/oxidation

Configuration Heavy-duty DI diesel (Cat 3406) Intake and complete in-cylinder region represented

Approach Variations in valve lift profile, intake-valve shrouds, injection scheme

Fuchs & Rutland, SAE 980508

Case Name Lift Profile ShroudSpeed(rpm)

SOI(deg. ATDC)

InjectionScheme

Standard standard none 1600 351.5 singleSlow slow none 1600 351.5 singleFlat flat none 1600 351.5 single2500 rpm standard none 2500 349.5 singleSwirl standard 1x180o 1600 351.5 singleAnti-Swirl standard 1x180o 1600 351.5 singleTumble standard 2x180o 1600 351.5 singleStandard Split standard none 1600 351.5 doubleSwirl Split standard 1x180o 1600 351.5 double

Swirl TumbleAnti-Swirl

0

10

20

30

40

50

60

355

NO

x (g

/kg

fuel

)

365 375 385 395 405 415 425Crank Angle (deg. ATDC)

StandardSlowFlat2500 rpmSwirlAnti-SwirlTumble

2500 rpm case lacks time to form NOx

Fuchs & Rutland, SAE 980508

X

Standard case Split injection reduces soot emissions

Swirl case Split injection increases soot emissions Corresponds to degraded combustion after 2nd injection

caused by stratification

0

0.5

1

1.5

2

2.5

3

3.5

340

Soot

(g/

kg fu

el)

360 380 400 420 440 460Crank Angle (deg. ATDC)

Swirl SplitStandardStandard SplitSwirl

Fuchs & Rutland, SAE 980508Soot Emissions and Swirl Split Injection

Fuchs & Rutland, SAE 980508Soot Emissions and Swirl Split Injection

Swirl case Soot is stratified 2nd injection aligns with soot cloud

from neighbor plume

fuel - bluesoot - gray

Soot - 15.4%Oxygen - 58.49%

Soot - 44.67%Oxygen - 45.43%

Standard Split Injection Case

Swirl Split Injection Case

373o ATDC: 3o after start of 2nd injection

Reference Peters & Hasse, BMBF Workshop, Germany, 2001

Combustion Modeling Unsteady non-premixed flamelet model (RIF model) >100 species, 500 reactions for autoignition, heat release, NOX,

soot precursors

Configurations Automotive DI diesels: VW 1.9L, Audi 2.5L Sector model, calculations begin post-IVC

Approach Variations in fueling (load) and EGR

Current Models Require Extensive Calibration Current Turbulent Combustion Models Are Not

Literal Representations of the Physics Have proper scaling Have reasonable limiting behaviors

Different Approaches Yield Comparable Levels of Agreement with Engine Measurements

Dominant Combustion Modeling Approaches Equilibrium/Arrhenius/EBU combinations and variants Flamelet models

Current Models Can Be Used Effectively By Experts in Close Concert with Experiments Including NOX and soot emissions

Temporally (and Spatially?) Resolved Engine-Out Velocity, Temperature, Major Species, and Minor Species (Including Pollutants)

Steady-State and Transient Operating Conditions Variations in:

Engine configuration (ports, combustion chamber, valve-lift profiles, fuel injectors, . . .)

Engine operating conditions (load and speed) Fuel composition (alternative fuels, additives, blended

fuels) EGR, injection schedule, . . .

First-Generation Ensemble-AveragedModel Formulations

Fuel-Sprays Diesel Combustion Autoignition, Heat

Release, NOX, and Soot Homogeneous/Nearly Homogeneous

Robust Premixed Flame Propagation Flow Structure and Gas-Phase Mixing

increasinglypre-dictive

Elucidate Physics (with Experiments) Interpolate/Extrapolate (Not Too Far!) from

Established Calibration Points Establish Trends Identify Figures of Merit Develop Correlations for Reduced Models Compute Total and Time-Resolved Engine-Out

Emissions (Steady-State Engine Operation)

Improved Fuel Spray Models Physics Numerical methodology

Next-Generation Turbulent Combustion Models Correct physics of SCCI combustion processes Turbulence/chemistry interactions Flamelet models, PDF methods, and hybrids

Accommodation for Detailed Chemical Kinetics Flamelet models Formal reduction techniques Storage/retrieval schemes

Beyond the Ensemble Average Cycle-to-cycle variations, transient operating conditions Large-eddy simulation

Spatial Filtering Replaces Ensemble Averaging Filtered Moment/Density Function PDEs Have

Essentially Same Form as Ensemble-Averaged Counterparts

Differences Arise From: Filtering being non-commutative with differentiation Filtered fluctuations being non-zero

Differences Often Are Neglected, in Practice Key Physics to be Modeled is at Sub-Filter Scale

Spray and combustion models remain essentially the same Principal difference is specification of turbulence scales