John E. Decand

Magnus SjöbergSandia National Laboratories

Sponsor: U.S. Dept. of Energy, OTT, OAAT and OHVTProgram Managers: Kathi Epping and Gurpreet Singh

HCCI Combustion: the Sources of Emissions atLow Loads and the Effects of GDI Fuel Injection

8th Diesel Engine Emissions Reduction WorkshopAugust 25-29, 2002

Introduction

HCCI engines are a low-emissions alternative to diesel engines.

– Provide diesel-like or higher efficiencies.

– Very low engine-out NOX and PM emissions.

Research is required in many areas to resolve technical barriers tothe development of HCCI engines by industry.

– The objective of our work is to help provide this understanding.

Establish a laboratory to investigate HCCI combustion fundamentals.

– All-metal engine: fully operational – result are subject of presentation.

– Optically accessible engine: examine in-cylinder processes (end of 02).

CHEMKIN kinetic-rate computations

– Guide experiments

– Assist in data analysis

– Show limiting behavior

Engine and Operating Conditions

Six-cylinder diesel engine converted forbalanced, single-cyl., HCCI operation.

Versatile facility to investigate variousoperational and control strategies.– Compression ratios from 13 - 21 (18)*

– Swirl ratios from 0.9 - 3.2; 7.2 (0.9)*– Speeds to 3600 rpm (600 - 1800 rpm)*

– Multiple fueling systems.> Fully premixed (curr.)*> Port fuel injection (PFI)> Direct injection, gasoline-type (curr.)*> Direct injection, diesel-type

– Liquid or gas-phase fuels (iso-octane)*

Complete intake charge conditioning.– Intake temperatures to 180° C. (varies)*

– Intake pressures to 4 bars. (varies)*

– Simulated or real EGR. (none)*

HCCI All-Metal Engine

* Values in (red) are used for current work.Based on Cummins B, 0.98 ltr./cyl.

Engine Appears to be Working Well

88

90

92

94

96

98

100

90 100 110 120 130 140 150Intake Temperature [°C]

Co

mb

ust

ion

effi

cien

cy[%

]

36

38

40

42

44

46

48

Th

erm

alef

fici

ency

[%]

Combustion efficiency

Thermal efficiency

CR = 18; φφφφ = 0.24; Pin = 120 kPa;1200 rpm; Well-Mixed Charge

– Large squish clearance.– Ring-land crevice 1% of TDC vol.– Various compression ratios

0

25

50

75

100

125

150

175

200

90 100 110 120 130 140 150Intake Temperature [°C]

CO

&H

C[g

/kg

fuel

]

0

100

200

300

400

500

600

700

800

NO

x[m

g/k

gfu

el]

CO

HC

NOx

0

50

100

150

200

250

300

350

400

90 100 110 120 130 140 150Intake Temperature [°C]

IME

Pg

[kP

a]

0

0.5

1

1.5

2

2.5

3

3.5

4S

td.D

ev.o

fIM

EP

g[%

]

IMEPg

Std. Dev. IMEPg

Computational Approach

Senkin application of the CHEMKIN-III kinetics rate code.

– Single-zone model with uniform properties and no heat transfer.

– Allows compression and expansion with slider-crank relationship.

– Full chemistry for iso-octane (Westbrook et al., LLNL).

Great oversimplification of a real engine. Model cannot reproduceall real-engine behavior.

Model is well suited for investigating certain fundamental aspects ofHCCI combustion.

– Allows the effects of kinetics and thermodynamics to be isolated andevaluated without complexities of walls, crevices, and inhomogeneities.> Assists in analysis of experimental data by separating chemical-kinetic

and physical effects.

– Represents the adiabatic limit for bulk-gas behavior in real engines.

– Guide experiments by showing approximate trends in ignition timing &temperature compensation with changes in operating conditions.

0.0 0.1 0.2 0.30

20

40

60

80

100

120

Equivalence Ratio (Phi)

Com

bust

ion

Effi

cien

cy(%

)

0.0 0.1 0.2 0.3 0.4 0.51e-8

1e-7

1e-6

1e-5

1e-4

1e-3

1e-2

1e-1

Equivalence Ratio (Phi)

Mol

eF

ract

ion/

Phi

HCOHCHC + OHCCO

100 ppm for φφφφ = 1.0

Below φ = 0.2, emissions rise followed by a drop in combustion efficiency.– Temperatures are too low to complete reactions, especially CO → CO2.

Indicates high emissions of OHC as well as CO and HC.– OHC not well-detected by standard FID HC detector, and they can be harmful.

Results for bulk-gas alone, in the absence of heat transfer.– Occurs in range of interest (typical diesel idle conditions are φ = 0.10 - 0.12).

– In real engine, heat transfer will shift onset of incomplete reactions to higher φ.

– Walls & crevices also add to emissions.

CHEMKIN predicts incomplete bulk-gas reactions at low loads.Iso-Octane; 1800 rpm; CR = 21; Tin = 380 K; Pin = 1 atm.

SAE Paper 2002-01-1309

Vary φφφφ: Experiment and CHEMKIN

Iso-Octane ; CR = 18; 1200 rpm;Pin = 120 kPa; Pre-Mixed

Intake temperature adjusted for50% burn at TDC for φ = 0.14.

– Experiment: Tin = 140° C

– CHEMKIN: Tin = 117° C

Experiment shows greatervariation in combustion phasing.

– Heat transfer and residuals.

– Advanced timing at higherloads has little effect onemissions.

– Timing retard at low loads issmall and has little effect onemissions.

40

60

80

100

120

Pre

ssur

e(B

ars)

MotoredPhi = 0.06Phi = 0.10Phi = 0.14Phi = 0.18Phi = 0.22Phi = 0.26

CHEMKINTin = 117° C

340 350 360 370 380 390 40020

30

40

50

60

70

80

90

100

Crank Angle (Degrees)

Pre

ssur

e(B

ars)

MotoredPhi = 0.06Phi = 0.10Phi = 0.14Phi = 0.18Phi = 0.22Phi = 0.26

ExperimentTin = 140° C

Emissions: Experiment and CHEMKIN

1200 rpm; CR = 18; Pin = 120 kPa;Pre-Mixed Fueling

Experimental Tin was increasedto maintain near-TDC ignition.– Compensate for heat transfer.

Experimental emissions matchmodel closely.

– CO levels match closely ~65%> Bulk-gas source.

– HC, rise for φ < 0.2 is similar tomodel indicates bulk-gassource at low φ.> Near-constant baseline

level for φ > 0.2 suggestcrevice source.

– “Missing carbon” in experimentindicates presence of OHCs.> Similar to model, but lower

due to FID response.

0

10

20

30

40

50

60

70

80

90

100

0.02 0.06 0.1 0.14 0.18 0.22 0.26

Equivalence Ratio [φφφφ]

Fu

elC

arb

on

into

Em

iss

ion

s[%

]

CO

CO2

HC

OHC

CHEMKINTin = 117° C

0

10

20

30

40

50

60

70

80

90

100

0.02 0.06 0.1 0.14 0.18 0.22 0.26

Equivalence Ratio [φφφφ]

Fu

el

Ca

rbo

nin

toE

mis

sio

ns

[%]

CO

CO2

HC

OHC

ExperimentTin = 140° C

0

10

20

30

40

50

60

70

0.02 0.06 0.1 0.14 0.18 0.22 0.26 0.3Equivalence Ratio [φφφφ]

Fu

elC

arb

on

into

CO

[%]

CHEMKIN, Tin = 117° CExperiment, Tin = 118° CExperiment, Tin = 131°CExperiment, Tin = 142° CExperiment, Tin = 159° CInterpolated 50% HR@TDC

Comparison of CO Emissions with φφφφ at Various Tin

Magnitude of increase in CO with decreasing φ agrees well with theCHEMKIN results. Shows incomplete bulk-gas reactions are the cause.

– Onset of rise in CO levels is shifted to higher φ in engine due to heat transfer.

– Rise in CO shifts to progressively higher φ as Tin is reduced (lower Tcombustion.).

– Engine data also show a large drop in combustion efficiency at low loads,corresponding to the increase in CO.

-50

0

50

100

150

200

250

300

350

0.02 0.06 0.1 0.14 0.18 0.22 0.26Equivalence Ratio [φφφφ]

IME

Pg

[kP

a]

0

2

4

6

8

10

12

14

16

Std

.Dev

.IM

EP

g[k

Pa

&%

]IMEPg [kPa]

Std. Dev. IMEPg [kPa]

Std. Dev. IMEPg [%]

Efficiencies and Combustion Stability

Combustion efficiency dropsfrom 95% to 60% as fuel isreduced to low-idle, φ = 0.1.– Similar drop in pressure-

indicated thermal efficiency.

– Commensurate with therapid rise in CO.

Std. Dev. of IMEP is 2 – 4kPa for all fueling rates (φ).– Increase at φ = 0.16 due this

being in the middle of therapid rise in CO.

Normalized σIMEPincreases below φ = 0.1because IMEP is near zero.– Std. Dev. of IMEPg ≤ 2.6%

from φ = 0.1 to φ = 0.26.

0

10

20

30

40

50

60

70

80

90

100

0.02 0.06 0.1 0.14 0.18 0.22 0.26Equivalence Ratio [φφφφ]

Co

mb

ust

ion

effi

cien

cy[%

]

0

5

10

15

20

25

30

35

40

45

50

Th

erm

alE

ffic

ien

cy[%

]

Combustion efficiency [%]

Thermal efficiency [%]

CR = 18; Pin = 120 kPa;Tin = 140° C; Pre-Mixed

Effect of Intake Pressure

1200 rpm; CR = 18; Pre-Mixed

Changing Pin from 101 to 120kPa has little effect on onset ofincomplete bulk-gas reactionswhen combustion phasing ismaintained.

– Experiment and CHEMKINboth show slightly lower COvalues during rapid rise.

Tin adjusted for 50% burn atTDC for φ = 0.14.

– 101 kPa: Tin = 157° C

– 120 kPa: Tin = 140° C

0

10

20

30

40

50

60

70

0.02 0.06 0.1 0.14 0.18 0.22 0.26 0.3

Equivalence Ratio [φφφφ]

Fu

elC

arb

on

into

CO

[%]

1.0 bar, Expr. Tin = 157°C

1.2 bar, Expr. Tin = 140°C

1.0 bar, CHEMKIN, Tin = 121° C

1.2 bar, CHEMKIN, Tin = 117° C

0

10

20

30

40

50

60

70

80

90

100

0.02 0.06 0.1 0.14 0.18 0.22 0.26 0.3Equivalence Ratio [φφφφ]

Fu

elC

arb

on

into

Em

issi

on

s[%

]

CO2 100 kPa

CO2 120 kPa

CO 100 kPa

CO 120 kPa

HC 100 kPa

HC 120 kPa

OHC 100 kPa

OHC 120 kPa

0

10

20

30

40

50

60

70

0.02 0.06 0.1 0.14 0.18 0.22 0.26Equivalence Ratio [φφφφ]

Fu

elC

into

CO

&H

C[%

]

30

40

50

60

70

80

90

100

Co

mb

ust

ion

effi

cien

cy[%

]

CO 600 rpmCO 1200 rpmCO 1800 rpmHC 600 rpmHC 1200 rpmHC 1800 rpmCE 600 rpmCE 1200 rpmCE 1800 rpm

Effect of Engine Speed on Bulk-Gas Reactions

Tin adjusted to maintain combustion phasing at TDC for φ = 0.14.– Higher compression temperatures compensate for reduced time for reactions.

Engine speed has little effect on the fueling rate at which the onset ofincomplete bulk-gas reactions occurs – for iso-octane.– In agreement with CHEMKIN computations.

Results suggest that special combustion strategies will be required forlow-load operation.

GDI Fueling: Vary Injection Timing

Early InjectionProvides a fairly uniform mixture.

– Can lead to incomplete bulk-gasreactions at low loads, aspredicted by CHEMKIN.

Late InjectionCan provide partial chargestratification.

– Mixture locally richer for thesame fueling rate.

– Offers the potential to mitigateincomplete bulk-gas reactions atlight loads.

Also, could prevent fuel fromreaching ring-land crevice.

– Reduce baseline emissions.

Variation in Injection Timing: φφφφ = 0.1

Tin = 142° C; Pin = 120 kPa;1200 rpm; GDI fueling

Early injection (0-90° aTDCintake) provides a well-mixedcharge.

– High CO and low combustionefficiency for φ = 0.1.

Retarding injection improvescombustion and emissions forlow-load operation.

– Injection at 290° reduces COand HC emission substantiallywith only about 1g/kg-fuel NOX(4 ppm).

– Combustion efficiencyincreases from 59% to 82%.

Further improvements possiblewith optimized stratification.

0

10

20

30

40

50

60

70

80

90

100

0 45 90 135 180 225 270 315 360Injection Timing [deg. CA]

IME

Pg

[kP

a]&

Co

mb

ust

ion

effi

cien

cy[%

]

0

2

4

6

8

10

12

14

16

18

20

Std

.Dev

.of

IME

Pg

[%]

IMEPg [kPa]

Combustion efficiency [%]

Std. Dev. IMEPg [%]

0

200

400

600

800

1000

1200

1400

0 45 90 135 180 225 270 315 360Injection Timing [deg. CA]

CO

&H

C[g

/kg

fuel

]

0

5

10

15

20

25

30

35

NO

x,50

xS

oo

t[g

/kg

fuel

]

CO HC

Soot NOx

Summary and Conclusions - 1

Metal HCCI research engine appears to be functioning well.

– At fully combusting conditions: ηthermal ~ 46%, σIMEP < 1%, CO < 65g/kg (1000 ppm), HC < 35 g/kg (1200 ppm), NOX ~ 0.06 g/kg (1 ppm).

CHEMKIN results show that for fuel loads below φ ~ 0.16, bulk-gasreactions are incomplete, even for an idealized adiabatic engine.

– Significant combustion inefficiencies, very high CO, and increased HC.> Temperatures are too low to complete reactions, mainly CO → CO2.

– Indicate that significant OHC emissions should occur. (OHC is ~2x HC).

Experimental data show a very similar trend to the changes inemissions and combustion efficiency as fuel loading is reduced.

– CO levels match very closely with those of the model (~65% of fuel C).> Bulk-gas must be the source.

– Onset of incomplete bulk-gas reactions occurs at higher φ ~ 0.2 due toheat transfer cooling the charge.

Summary and Conclusions - 2

The “missing carbon” in the emissions measurements matches theexpected OHC trends.– HC detector (FID) has low sensitivity to OHC.

Combustion stability was good even at idle loads, σIMEP ≤ 2.6%.

Increasing Pin from 1.0 to 1.2 bars had little effect on the onset ormagnitude of incomplete bulk-gas reactions when ignition timingwas maintained.

Changing speed from 600 to 1800 rpm has almost no effect on theonset or magnitude of incomplete bulk-gas reactions for iso-octane.

– Increased compression temperatures required to maintain ignitiontiming compensate for reduced time to complete reactions.

Partial charge stratification by late-cycle fuel injection appears tohave a strong potential for mitigating the difficulties of low-loadoperation.