Meeks

Strategies for Using Detailed Kinetics in Engine Simulations Ellen Meeks ERC Symposium: Fuels for Future IC Engines June 6-7, 2007 Madison, Wisconsin

Outline ● Role of simulation in design ● Importance of chemical kinetics ● Challenges of using detailed kinetics ● Strategies that bridge the gap between design simulation and kinetics

2

Cost

Simulation reduces cost of development Testing

● There is a growing opportunity to: Simulation Complexity, Capability, Time

3

– – – –

Reduce risk Improve use of testing Speed development Facilitate innovation

How effective can simulation be? ● According to our customers, it depends on the accuracy: – Needs to be predictive to Ñ Allow

exploration and discovery Ñ Guide design and testing

– Must provide deep understanding of dependencies to Ñ Manage

complexity of tradeoffs Ñ Facilitate innovative solutions

4

Computational fluid dynamics (CFD) is already widely used for engine design Geometry A

Example Goal: Maximize uniformity in piston bowl

10º BTDC

TDC

20º ATDC

STAR-CD simulations with simplified chemistry

5

Geometry B

10º BTDC

CFD Simulation: Determine effects of port flow distribution, injector design, injector timing, piston bowl geometry

TDC

20º ATDC

An engine system is a chemical plant Fuel

Air (O2 + N2)

Horsepower CO2 + H2O CO, NOX ,PM Unburned HC Pt / Rh CO2 + H2O

Air (O2+N2)

6

There is a gap between CFD simulation and testing requirements Goal: Minimize emissions while maximizing efficiency

Workflow is incomplete CFD users cannot predict chemistry effects adequately

Ignition delay, CO, UH, PM, etc. 7

CFD Simulation: Determine effects of port flow distribution, injector design, injector timing, piston bowl geometry

10º BTDC

TDC

20º ATDC

Test: Determine emissions, efficiency, performance

Detailed chemistry is required to address many important issues ● Ignition timing – Diesel – HCCI/CAI

● Knock reduction ● Emissions control – NOx, CO, UHC, PM

● Fuel flexibility ● Exhaust-gas recirculation ● System control 8

Challenges for using detailed kinetics ● Real fuels are too complex ● Validated detailed mechanisms are scarce ● Simulations are too time-consuming ● Companies often lack chemistry expertise

9

Addressing the challenges… ● Real fuels are too complex ● Validated detailed mechanisms are scarce ● Simulations are too time-consuming ● Companies often lack chemistry expertise

10

Real fuel components can be categorized 50

Example: Gasoline*

● Classes of chemical compounds have

*http://www.atsdr.cdc.gov/toxprofiles/tp72-c3.pdf

40

– Common molecular structures – Similar chemical behavior

30 20 10

11

miscellaneous

aromatics

cycloalkenes

cycloalkanes

isoalkanes

alkenes

alkanes

0

● Biofuels have different compositions – Methyl-ester structures – Contain oxygen

Surrogate fuel mixtures can be used to represent real fuels in simulations ● 1 or 2 molecules represent each significant chemical class, e.g.: Real Fuel Component iso-paraffins normal paraffins single ring aromatics cyclo-paraffins olefinic species multi-ring aromatics oxygenates

Surrogate Fuel Candidate iso-octane, hepta-methyl nonane n-heptane, n-hexadecane toluene methylcyclohexane 1-pentene alpha-methyl napthalene methyl stearate, methyl linoleate

● Detailed chemistry models are built for each molecule – Use elementary reactions – Allow merging to form mixtures 12

Different model fuels can be assembled for different applications ● Tailor to prediction of desired combustion and physical properties: – – – – –

13

Ignition delay Knocking tendency Flame speeds Pollutant emissions Sooting tendency & particle size distributions

Example shows prediction of ignition delay with 5-component mixture ● Surrogate mixture compositions defined to match class composition or RON #* – All do reasonably well in capturing ignition behavior with idealized model of engine, detailed kinetics

*Reported by: C. V. Naik, et al SAE 2005-01-3742

14

With a database of molecules, we can assemble an appropriate surrogate Real Fuel Characteristics • Class composition • Heat-release rate • Octane / Cetane # • H/C ratio, O content

U.S. E15 Gasoline

aromatics olefins

Match Properties • Select Molecules • Set Composition

c-paraffins

n-paraffins

e an pt he n-

i-paraffins

n-heptane Iso-octane 19% 1-pentene mchexane 45% m-xylene

15% 3% 1% 15%

15

ethanol

Merge Mechanisms • Species • Reactions Surrogate Fuel Composition

Chemical Model for Simulation

Addressing the challenges… ● Real fuels are too complex ● Validated detailed mechanisms are scarce ● Simulations are too time-consuming ● Companies often lack chemistry expertise

16

Mechanism generation for a molecule can be fully automated Reaction Patterns

Determine Reactions and Molecules

Group Contribution

Determine Properties of Molecules

Properties

Stored Data

Quantum Chemistry

Determine Kinetic Rate Parameters

Groups

Kinetic Rates

Experimental Data

Simulate Target Experiments Rules

Based on work in collaboration with the Dow Chemical Company

17

Surrogate Fuel & Oxidizer

No

aromatics olefins

OK? Yes

Chemical Chemical Mechanism Mechanism

e n-h

xad

eca

ne

c-paraffins i-paraffins

n-paraffins

Validation focused on mechanismgeneration rules is key to building database ● Well studied molecules used to refine rules ● Assures consistency between surrogate fuel component mechanisms ● Allows easy assembly of mixtures

18

Get full benefit from molecules that have been well studied experimentally ● For example: n-heptane – Shock-tube, JSR, Flames, RCM, and Engine data Ignition delay time (sec)

1

Shock-tube and RCM Data Pressure Range: 0.07 to 60 atm Stoichiometry: ϕ = 0.1 to 4

0.1 0.01

Shock-tube data RCM data

0.001 0.0001 0.00001 0.5

0.7

0.9

1.1

1.3

1000/Temperature (1/K)

19

1.5

Validation involves simulations of well controlled experiments with full chemistry ● CHEMKIN simulation of shock-tube show detailed kinetics ability to predict ignition Ignition delay time (sec)

0.1 0.01 0.001

Ciezki et al., 3 atm Simulation, 3 atm Ciezki et al., 13.3 atm Simulation, 13.3 atm

0.0001

Ciezki et al., 40 atm Simulation, 40 atm

0.00001 0.6

0.8

1

1.2

1.4

1000/Temperature (1/K)

20

1.6

Improvement of rules creates solid foundation for mechanism generation ● Enables full power of mechanism generation ● Assures consistency in mixtures Example: Example: Improvement Improvementof ofMCH MCH mechanism mechanisminvolved involvedupdated updated reaction reactionrate raterules rulesfor forRO RO2 chemistry chemistry Ignition time (sec)

2

0.10

Model, 10 atm, Year10 1 mec Original model, atm

Original model, atm Model, 15 atm, Year15 1 mec Original model, atm Model, 20 atm, Year20 1 mec Data, 10 10 atmatm Data, Data, 15 atmatm Data, 15 Data, 20 20 atmatm Data, Model, 10 atm, updated Revised model, 10 m atm

0.01

Revised model, 15 m atm Model, 15 atm, updated Revised model, 20 m atm Model, 20 atm, updated

650

750

850 Temperature (K)

950

1050

Work performed in collaboration with C. Westbrook and W. Pitz, et al., LLNL 21

Addressing the challenges… ● Real fuels are too complex ● Validated detailed mechanisms are scarce ● Simulations are too time-consuming ● Companies often lack chemistry expertise

22

Several methods allow linking to CFD with limited impact on simulation time CFD Model of cylinder geometry

+ global chemistry

+ progress variables

Automated AutomatedZone Zone Mapping Mapping Table Tablelook-up look-upvs. vs. Progress Progressvariable variable Multi-zone model

Detailed Chemistry Model 23

+ reduced chemistry

Automated Automated Chemistry Chemistry Reduction Reduction

Single-zone or 1-D flame model

Multi-zone models allow detailed kinetics ● Multiple, homogeneous regions – – – –

Core, boundary layer, crevice Connected through work/heat Can be auto-extracted from CFD Can include full particle-formation chemistry

● Approach shows good prediction of heat-release, pressure, ignition, emissions From: S. M. Aceves, et al., SAE 2001-01-1027

24

Automated mechanism reduction can be very effective for “skeletalization” Equiv Ratio 0.5 and 20 atm

● Ignition delay example:

1.00E+00

Ñ Ñ Ñ

Ignition Delay (sec)

– 5-component gas. surrogate – 60% reduction, ~ 5X Faster Equivalence ratio 0.1 - 2.0 T= 600 -1800K P= 0.5 - 60atm

1.00E-01

Skeletal

1.00E-02

Master

1.00E-03 1.00E-04 1.00E-05 1.00E-06 0.0000

– < 4% error

120

Skeletal

100

(cm/sec)

– < 2% error

140

Flame Speed

Ñ

1.5000

Flame speed at 500K

– n-heptane diesel surrogate – 80% reduction, ~ 30X Faster Equivalence ratio 0.7 - 1.7 T= 300 - 700K

1.0000

1000/T (1/K)

● Flame-speed example:

Ñ

0.5000

Master

80 60 40 20 0 0.4

0.8

1.2

1.6

Equivalence Ratio 25

2

More severe reduction can also be achieved ● Lumping of species or reactions – Remove requirement of elementary reactions

● Use of partial equilibrium assumptions – Remove some species from flow equations

● Identification of “manifolds” – Separate out timescales

26

Addressing the challenges… ● Real fuels are too complex ● Validated detailed mechanisms are scarce ● Simulations are too time-consuming ● Companies often lack chemistry expertise

27

The Model Fuels Consortium is addressing many of these issues ● Industry funded and Directed ● Built on commercial offerings ● Advised by leading science experts

28

Overall goal is to bridge gap between chemistry and engine design simulation

Cost per Test / Simulation

● Make better use of expensive tests TEST

CFD MFC Kinetics Simulation

● Improve accuracy of design simulations ● Provide bridge between detailed chemistry and design tools ● Build fuel-chemistry knowledge base

Number of Tests / Simulations 29

Focus is on building a validated database and engineering analysis tools Knowledge-base

Fuel Component Mechanisms

Fuel Mixture Representation Master Mechanism Assembly Mechanism Evaluation and Validation

Reduced Mechanisms 30

Mechanism Reduction

Software Tools

Development of Mechanism Analysis and Engine Simulation Tools

Development of Mechanism Reduction Tools

MFC Technical Advisory Team ● Dr. Charles Westbrook – Chief Technical Advisor – A pioneer in combustion modeling while at the Lawrence Livermore National Laboratory

● Prof. Mitsuo Koshi, Tokyo University – Expert in combustion kinetics and mechanism generation

● Prof. Anthony Dean, Colorado School of Mines – Expert kineticist; formerly lead scientist at Exxon

● Prof. William Green, Massachusetts Inst. of Technology – Expert in numerical methods for model reduction and mechanism generation techniques

● Prof. Ulrich Maas, Universität Karlsuhe – Expert in engine combustion simulation and numerical methods

● Prof. Hiromitsu Ando, Fukui University – Former deputy general manager of engine research at Mitsubishi Motors

31

Summary and Conclusions ● There is increasing need for detailed chemistry simulation for engine design ● Advances in technology facilitate use of kinetics in simulation – – – –

32

Automation of mechanism generation Automation of mechanism reduction Advanced overlay techniques with CFD More powerful computers

Contributors C. V. Naik K. V. Puduppakkam C. Wang

Thank You

33