Bridging the Gap between Fundamental Physics and Chemistry ... · Use GT-Power as platform for system modeling Use experiments and CFD models to provide simplified combustion models

Bridging the Gap betweenBridging the Gap betweenFundamental Physics and ChemistryFundamental Physics and Chemistry

and Applied Modelsand Applied Modelsfor HCCI Enginesfor HCCI Engines

Dennis N. Assanis

DEER Conference August 20-25, 2005

Chicago, IL

Bridging the Gap Between HCCI Fundamentals and Applied Models

HCCI Engine Grand Challenges

Physics Models

Enabling Technologies

Experiments ChemicalKinetics

Sensors, Actuators,

Control

HCCI HCCI CombustionCombustion

University of MichiganUniversity of MichiganDennis Assanis (PI), Arvind Atreya, Zoran Filipi, Hong Im,Dennis Assanis (PI), Arvind Atreya, Zoran Filipi, Hong Im,

George Lavoie, Volker Sick, Margaret WooldridgeGeorge Lavoie, Volker Sick, Margaret Wooldridge

Massachusetts Institute of TechnologyMassachusetts Institute of TechnologyWai Cheng, William Green Jr., John HeywoodWai Cheng, William Green Jr., John Heywood

Stanford UniversityStanford UniversityTom Bowman, David Davidson, David Golden, Ronald Hanson, Chris ETom Bowman, David Davidson, David Golden, Ronald Hanson, Chris Edwardsdwards

University of California, BerkeleyUniversity of California, BerkeleyJyhJyh--Yuan Chen, Robert Dibble, Karl HedrickYuan Chen, Robert Dibble, Karl Hedrick

A University Consortium onHomogeneous Charge Compression Ignition

Engine Research - Gasoline

HCCI University ConsortiumHCCI University Consortium


Acknowledgements


Objectives

Bridge the gap between fundamental HCCI physics and chemistry considerations and applied models─ Balance model fidelity with computational expediency and ultimate goal in

mind

Demonstrate a cascade sequence where high fidelity models and experiments feed phenomenogical models appropriate for systems analysis

Apply system models to assess candidate control schemesMulti-Variable

Engine Controller

SIGNALS IN:1. Coolant temperature sensor2. "Combustion" sensor : pressure transducer; or novel SOC probe; or ionization probe3. Air mass flow sensor4. Intake manifold temperature sensor5. Intake manifold pressure sensor6. Exhaust manifold temperature sensor7. Exhaust manifold pressure sensor8. Wide range oxygen sensor

RPM

Requested Torque

SIGNALS OUT:1. Camless intake valve(s) timing2. Spark plug for DGI mode3. Fuel injector timing4. Camless exhaust valve(s) timing5. EGR control valve6. Controllable water pump

IN OUT


Approach


UM Optical Engines UM Heat Transfer Engine Stanford Camless

MIT VW Engine with VVT UCB VW TDI engine UCB CAT Single Cylinder

Engine University Consortium Set-Ups


Local Heat Fluxes

-0.5

0.0

0.5

1.0

1.5

2.0

2.5

-120 -90 -60 -30 0 30 60 90 120

p2

Hea

t Flu

x [M

W/m

2 ]

deg crank angle

-0.5

0.0

0.5

1.0

1.5

2.0

-120 -90 -60 -30 0 30 60 90 120

p3

Hea

t Flu

x [M

W/m

2 ]

crank angle (deg)

-0.5

0.0

0.5

1.0

1.5

2.0

-120 -90 -60 -30 0 30 60 90 120

h1

Hea

t Flu

x [M

W/m

2 ]

crank angle (deg)

spark

P7

HEAD

PISTONinjector

H1

TELEMETRY

P6 P2P1

P3P4

P5

H2 spark

P7

HEAD

PISTONinjector

H1

TELEMETRY

P6 P2P1

P3P4

P5

H2


Concept of Modified Woschni Correlation

Models need to be calibrated to provide accurate total heat loss Classic Woschni underpredicts heat

transfer during compression and leads to unrealistic ignition predictions

Modified Woschni heat transfer model:

─Original: A2 = 1

─Modified: A2 = 1/6

-0.5

0.0

0.5

1.0

1.5

2.0

2.5

-90 -60 -30 0 30 60 90

Spatially AveragedWoschniReduced Woschni

crank angle

Hea

t Flu

x [M

W/m

2 ]

1 21 2 ( )d rp m

r r

V Tw C S C PV

A A Pp

= + −

1 22.28 0.308 , 0.00324p

p

B wC C

Sπ

= + =

0.2 0.8 0.53 0.Woschni

83.26h B p T wχ − −= ⋅ ⋅ ⋅ ⋅


Experimental Burn Rate Data

Varying RPM, load, Tcool, Tinlet, A/F ratio

Generate correlations as function of ignition timing (one variable)

-10

0

10

20

30

40

-8 -6 -4 -2 0 2

BURN DATA

CA

(Deg

ATC

)

CA0 (Deg ATC)

EXPERIMENTALDATA POINTS

CURVE FITS

CA0

CA10

CA50

CA90

])/)((exp[1 10

+Δ−−−= wNORMx θθθ

85

90

95

100

-8 -6 -4 -2 0 2

COMBUSTION EFFICIENCY

CEF

(%)

CA0 (deg ATC)

DATA POINTSCURVE FIT

0

5

10

15

20

25

30

35

40

-20 -10 0 10 20 30 40

115C105C95C85C75C

Net

Hea

t Rel

ease

Rat

e [J

/deg

]

Crank Angle (deg)

Intake TemperatureIncrease


Modeling Approaches for HCCI Engines

Zero-D

Multi-D

Quasi-DMulti-zone

DetailedChemistry

ReducedChemistry

Predictions of ignition

Predictions of rate of burning

Predictions of emissions

Computational Efficiency

DetailedChemistry

?

?


Initial Modeling Approach:Sequential CFD + Multi-zone Model

1

2

3a

3b

Open-cycle calculation using Kiva-3V

Temperature / Equivalence Ratiodistribution obtained before TDC

Calculation of combustion event usingMulti-zone model with Temperature Zones only

Calculation of combustion event usingMulti-zone model with T/Φ Zones

TRANSITIONPOINT


Sensitivity to Transition Timing

Calculations at LLNL showed that the point of transition from KIVA to the multi-zone code can affect the results (SAE 2005-01-0115)─ Imposed Φ distribution on a 2D coarse grid (Fuel CH4)─ Solution obtained using sequential multi-zone approach compared against

detailed solution (KIVA-3V linked with Chemkin)─ Implication: Temperature and composition stratification is important

40

50

60

70

80

90

100

110

120

-20 -10 0 10 20

Pres

sure

[bar

]

Crank Angle Degree

Full chemistryin each cell

Transition @ 10 BTDC



Centerline


Naturally Occurring Thermal StratificationCollaboration with Sjöberg and Dec (Sandia) - SAE 2004-01-2994

Temperature [K]

Below Above

700

750

800

850

900

950

1000

Centerline

0

0.05

0.1

0.15

700 750 800 850 900 950 1000 1050

Mas

s D

ensi

ty D

istr

ibut

ion

Temperature [K]

330o

340o

360o

370o

350o

380o

Temperature and composition distributions continue to evolve until late in the compression stroke. Decoupling of flow and chemistry is problematic.

Temperature field at TDC


Coupling of KIVA-3V with Multi-Zone Model (KMZ)

Instead of solving for detailed chemistry in each cell, at each timestep divide into zones of cells with similar properties (i.e. T and ϕ), and assume they behave in a similar manner

1

2

3

Estimate composition change and heat released in each zone, and re-distribute to corresponding KIVA-3V cells

4

Temperature zonedivided into ϕ zones

Shaded part:3rd Temperature zone (15-35%)

ϕHigh

Low

TempHigh

Low

Temperature and ϕ distributions

Head

PistonHead

Piston


Validation of Fully Coupled Kiva-3V withMulti-Zone Model

The new model gives results that match very well the “exact” solution obtained by solving for detailed chemistry in every cellComputational time is reduced significantly (~90% for 10,000 cells)

60

65

70

75

80

85

90

-5 0 5 10 15 20 25


~100 zones

Pres

sure

[bar

]

Crank Angle [degree]

1000

1100

1200

1300

1400

1500

1600

1700

1800

-5 0 5 10 15 20 25


~100 zones

Mea

n Te

mpe

ratu

re [K

]


1000

1200

1400

1600

1800

2000

2200

-5 0 5 10 15 20 25


~100 zones

Max

imum

Tem

pera

ture

[K]


MaximumTemperature

Pressure MeanTemperature


Numerical Experiments: KMZ generated “data”

-10

0

10

20

30

40

-8 -6 -4 -2 0 2

ALL BURN RESULTS (KIVA-CHEMKIN)

1% of fuel10% burned50% burned90% burned

Bur

n A

ngle

s (d

eg A

TC)

CAD (deg ATC)

CA0

CA10

CA50

CA90

Variable Sweeps

CR 16, 9

RPM 2000, 1200

Load (fuel) 9, 13 mg

-10

0

10

20

30

40

-8 -6 -4 -2 0 2

BURN DATA

CA

(Deg

ATC

)

CA0 (Deg ATC)


CURVE FITS

CA0

CA10

CA50

CA90

ExperimentalData

Run T sweeps of KMZ model for various operating parametersGenerate burn angles as function of ignition timing, as well as other variables


System Studies Approach

Use GT-Power as platform for system modelingUse experiments and CFD models to provide simplified combustion models with correct trends for ignition, burn rates and combustion efficiency Develop single-cylinder engine model with manifolds including thermal system submodelExplore implementation issues at multi-cylinder and vehicle level, especially related to thermal transients

IGN DELAY

INTAKE COMPRESSION COMBUSTION EXHAUST

GT-POWER

DLL-ENGHEATTRUSER- ENGCOMBUSR

Heat transfer coefficient



Heat release rate

COMB. CORR.

-10

0

10

20

30

40

-8 -6 -4 -2 0 2

BURN DATA

CA

(Deg

ATC

)

CA0 (Deg ATC)


CURVE FITS

CA0

CA10

CA50

CA90

)/33700exp(103.1)( ]//[41.177.005.14

2TRPms KmolcalOign ⋅⋅⋅⋅×= −−−− χφτ


Thermal Transients – Simulating the UM Engine

HCCI mode,SAE 2002-01-0420

SI mode

1

3

2

4

0

2

4

6

8

10

12

-270 -180 -90 0 90 180 270

LIFT

(mm

)

CA (deg ATC)

EXH IN T

380

400

420

440

460

480

0 20 40 60time (sec)

T (K)

AVERAGE WALL TEMPERATURE

1

4

3

2


Thermal Transients – Challenges

Stable operation possible at steady state thermal points

Transient operation unsatisfactory

380

400

420

440

460

480

0 20 40 60time (sec)

T (K)

AVERAGE WALL TEMPERATURE

1

4

3

2

0

10

20

30

40

50

60

-20 -10 0 10 20 30 40

Point 3 at Steady State T3Point 3 at T4 (hot - knocking)

P(bar)

CA (deg ATC)

P3T4

P3T3

0

10

20

30

40

-20 -10 0 10 20 30 40

Point 2 at Steady State T2Point 2 at T1 (cold - misfire)

P

(bar)

CA (deg ATC)

P2T1

P2T2

Advanced / knocking

Retarded / misfire


Transient Compensation

Hot-to-cold compensation possible by reducing rebreathing lift or increasing Pinlet (both decrease EGR)

Cold-to-hot compensation not achievable (not enough EGR heat is available)

-10

-5

0

5

10

15

20

0

10

20

30

40

50

60

2.0 3.0 4.0 5.0

P3T4 LIFT

SO

C[C

AD

ATC

]

Lift[mm]

EGR [%]

BMEP x 10 [bar]

SOC [CAD]

ORIGINALCALIBRATION

ADJUSTEDCALIBRATION

-10

-5

0

5

10

15

20

0

10

20

30

40

50

60

0.940.950.960.970.980.99

P3T4 Pin

SO

C[C

AD

ATC

]

Pin [bar]

EGR [%]

BMEP x 10 [bar]

SOC [CAD]

ORIGINALCALIBRATION

ADJUSTEDCALIBRATION

0

10

20

30

40

50

60

-20 -10 0 10 20 30 40

P3T3,4mmP3T4,4mmP3T4,2.5mmP3T4,0.99bar

P(bar)

CA (deg ATC)

P3T4

P3T3

P3T4 (2.5 mm lift)P3T4 (0.99 bar Pin)


Thermal Transients - Implications

HCCI regime is shifted depending on direction of transient

Hot-to-cold will extend the low load HCCI region

Cold-to-hot will require SI operation until walls warm up

Challenge is to maximize HCCI operating regime

Cold-to-Hot

Hot-to-Cold

Steady State


Future Work

Refine modeling approaches and validate them against optical and metal engine measurements – emphasize DI, stratified operationExtract knowledge developed from detailed CFD + comprehensive chemistry models and capture it into practical correlations compatible with “smart” phenomenological single-zone modelsProvide a single-cylinder module to multi-cylinder system level, controls-oriented simulationsUse models to develop strategies for HCCI engine in-vehicle operation with alternative fuels

Mirror

Filter and Splitting Optics

ICCD Camera

Mirror

Filter and Splitting Optics

ICCD Camera


Acknowledgements

University of MichiganDr. Aris Babajimopoulos, Dr. George Lavoie, Dr. Zoran Filipi, Dr. Junseok Chang, Dr. Kukwon Cho, Orgun Guralp, Mark Hoffman,Yanbin Mo, Kyoung Joon Chang

General Motors Research LaboratoriesRod Rask, Paul Najt, Tang Wei Kuo, Jim Eng

Lawrence Livermore National LaboratoriesSalvador Aceves, Dan Flowers

Sandia National LaboratoriesJohn Dec, Magnus Sjoberg

Documents

Bridging the Gap between Fundamental Physics and Chemistry ... · Use GT-Power as platform for system modeling Use experiments and CFD models to provide simplified combustion models