Sources of CO and UHC Emissions in Low-Temperature Diesel Combustion Systems

Sandia National Laboratories CRF

Dae Choi, Will Colban, Isaac Ekoto

Duksang Kim, Sanghoon Kook

Paul Miles, Seungmook Oh

Michael Andrie, Dave Foster, Chad Koci

Roger Krieger, Rick Opat, Sung Wook Park

Youngchul Ra, Rolf Reitz

University of Wisconsin ERC

Russ Durrett, Manuel Gonzalez

John Pinson, Bob Siewer t General Motors R&D

Work supported by: US DOE Office of Vehicle Technologies

Gurpreet Singh, Program Manager

General Motors Corporation

GM-UW CRL & SNL Agreement FI083070326

DEER 2008, August 4-7, Dearborn, Michigan

We focus on early-injection, PCI-like combustion systems

Toyota

"smokeless”

combustion

(SAE 2001-01-0655)

AVL HCLI systems

(MTZ, Sept 2003)

16

14

12

10

8

6

4

2

0 1000 1500 2000 2500 3000 3500

Speed [RPM]

Load

[bar

]

Conventional Diesel Combustion

HPLI, Highly Premixed Late Injection

HCLI, Homogeneous Charge Late Injection

HPLI

HCLI

Injec ion Heat Release

Heat Release Injection

TDC

Low-temperature combustion

systems are attractive because:

t -PX /0x and PM emissions are

obtained simultaneously

t 5IFZ FNQMPZ Ucombustion sy

ZQJDBM EJFTFMstem FIE & Bo

wl

geometry

t $PNCVTUJPO UJNJOH JT DCZ UIF JOKFDUJPO FWFOU

POUSPMMFE

However:

t 5IFZIJHI $

PGUFO TVòFS GS PN B GVFM FDPOPNZ QFOBMUZ BTTPDJBUFE XJUI

0 BOE 6)$ FNJTTJPOT

t 5IFZ BSF BQQMJDBCMF P WFS POMZ B MJNJUFE TQFFE�MPBE S BOHF

CO and UHC emissions can stem from cool, fuel-lean regions as well as fuel rich regions

Constant φ & T, P = 60 bar, Δt=2 ms, 21% O 2 Soot/NOx contours from Kitamura, et al., JER 3, 2002

4

CCO 15% Soot 10%

5%

1%

500 ppm

NTC region

88 g/kg­fuel

NOx

5000 ppm

10002

3

Equ

ival

ence

Rat

io

CO [g / kg-fuel ]

1500

500

1

0

0600 1000 1400 1800 2200 2600

Temperature [K]

4

UHC 15%

10%

Soot

5%

1%

500 ppm

NTC regi 20on g/kg-fuel

NOx

5000 ppm

100

10

3

2

1Equ

ival

ence

Rat

io

0600 1000 1400 1800 2200 2600

Temperature [K]

UH

C [ g / kg-fuel ]

1000

1

t -PX 6)$ DBO S FTVMU FWFO GS PN P WFS�SJDI S FHJPOT

t 'PS MFBO NJYUVSFT U FNQFSBUVSFT BCPWF ���� , ���� , XJMM GVMMZ P YJEJ[F $ 0 6)$

t 6)$ BMTP TUFNT GS PN D PPM 5 � ��� , S FHJPOT B U BMM FRVJWBMFODF S BUJPT

How will EGR influence the kinetics of CO and UHC oxidation?

CO @ constant φ & T, P = 60 bar , Δt=2 ms 4

2200 2600

15%

10%

5%

1%

Soot

500 ppm

5000 ppm

NOx

10% O2

15% O2

21% O2

3

2

1

0 600 1000 1400 1800

Temperature [K]

EGR hardly impacts CO

yield from constant T, P

simulations

CO emissions impacted

only in low­

temperature crevice

regions

E

qu v

a en

ce R

at o

The isothermal constraint prevents heat

release from raising the temperature and

increasing the heat release rate

An adiabatic treatment (with pressure

matched to experiment) is more repre­

sentative of regions away from walls

2000

1800

21% O2

15% O2

10% O2

21% O2

15% O2

10% O2

88 g/kg-fuel

CO oxidation

threshold

UHC oxidation

threshold

20 g/kg-fuel

1600

1400

1200

UH

C_

EI

[g/k

g-f

ue

l]

CO

_E

I [

g/k

g-f

ue

l]

T

[K

] m

ax

1200

1000

800

600

400

200

0

250

200

150

100

50

0 0 0.1 0.2 0.3 0.4 0.5 0.6

Equivalence Ratio φ

t 5 IF QFBL U FNQFSBUVSF OFFEFE U P P YJEJ[FUHC & CO is independent of dilution

t %JMVUJPO TJHOJöDBOUMZ JODSFBTFT UIF FRVJWB­lence ratio needed to reach this temperature

Engine & experiment

Intake

Lasersheet

Exhaust

Line-imaging FOV

PLIF-imaging FOV

Removeable Mirror Laser

sheetICCD Filter

������ON ��ON�'8).�

PI MAX ICCD

t������CZ�����SFTPMVUJPO

t�����OT�HBUF

Princeton Instruments

ICCD-576E

t������CZ����

t�����OT�HBUF

Mirror

Collection lens

SPEX 270M Spectrometer

t������UP�����ON�#1

t����ON�SFTPMVUJPO

Bosch CRI2.2 Common Rail FIE

/P[[MF� ��IPMF �����¡ ������<DN�������T>

3BJM�1SFTTVSF� ����CBS

'VFM� 64����EJFTFM�GVFM

GM 1.9l cylinder head

#PSF� �����NN4USPLF� �����NN(FPNFUSJD�$3� ����

&òFDUJWF�$3� ����

Optical piston retains

prototype GM­

designed bowl

2-d PLIF Images:

1-d, spectrally-resolved: 30°CA

1 1.5 bar

0.8 CO

3.0 bar 0.6

0.4

0.2 4.5 bar

0

Radius [mm] 0 5 10 15 20 25 30 35 40

Nor

mal

ized

CO

[ - ]

Experimental PLIF images are corrected for distortion and laser sheet inhomogeneity

Clearance volume reference grid

Back-lit bowl reference grid

Before correction

Reference grid images in both the clearance volume

and the bowl obtained at each crank angle allow

separate distortion corrections for both images

A flat-field correction also accounts

for laser sheet intensity variation

After correction a near seamless

image is obtained

Numerical simulations – background

KIVA release 2 coupled with Chemkin

chemistry solver

Ignition/combustion model

Chemkin chemistry solver

Mechanism ERC-PRF mechanism (39 species, 131 reactions)

NOx mechanism Reduced GRI mechanism (4 species, 9 reactions)

Soot model 2-step phenomenological model

Turbulence model εRNG k- model

Atomization/breakupmodel

KH-RT model

ERC grid-size and time-step independent

models (ref. SAE 2008-01-0970)

Liquid/Gas phase momentum coupling Gas-jet model

Collision/Coalescence model

Radius-of-influence collision modelmodel

Time-step calculation Mean collision time step model

Parcel number control Re-group model

Computational grid at TDC:

Cut-plane

Cell # : 35000 (IVC) :15000 (TDC)

Operating conditions

We focus on a dilute operating

condition with rising CO and

UHC emissions, but reasonable

combustion efficiency (≈ 98%)

φglobal = 0.36

100 4.0

3.5

Pint=1.5 bar 3 bar load MBT timing

CO

UHC

NOx

8 10 12 14 16 1O2 Concentration [%]

Com

b. Eff. [%]

NO

x [g

/kg-

fuel

]C

O, U

HC

[% o

f fue

l ene

rgy]

98 3.0

2.5

2.0

1.5

1.0

0.5

0.0

96

Speed 1500 [rpm]

Load 1.5 – 3.0 – 4.5 [bar]

Intake [O�] 10.0 [%]

S OI (command) (-31.1) – (-26.6) – (-15.8) [° A5 DC ]

5 intake 95 [°C ]

P intake 1.5 [bar]

8

t 5 ISFF JOKFDUJPO UJNJOHT�

“Advanced” = -31.1°

i#BTFMJOFw .#5 � �����¡ “Retarded” = -15.8°

t 5 ISFF MPBET BSF BMTP considered

An optimal injection timing exists that minimizes CO, UHC, and fuel consumption

CO

Em

issi

ons

[Vol

. %]

-30 -25 -20 -15 -10 -5 0 5 10

21 %

19 %17 %

15 %

14 % 12 %

11 %

10 O2%

1500 RPM 3 bar IMEP

Kook, et al. SAE 2005-01-3837

0.8

0.6

0.4

0.2

0

SOI [CAD ATDC]

0 UW Metal Engine

SNL Optical Engine 2000 RPM 6 bar IMEP

9% O2

$0

UHC

Opat, et al. SAE 2007-01-0193 Colban, et al. SAE 2008-01-1066

CO

Em

issi

on In

dex

[g/k

g-fu

el]

350 30

300 25

20

15

10

5

250

200

150

100

50

0-45 -40 -35 -30 -25 -20 -15

SOI [oCA ATDC]

UH

C Em

issions Index [g/kg-fuel]

t 5IF F YJTUFODF PG UIJT PQUJNVN EFNPOTUSBUFT UIF JNQPSUBODF BOEWBSJBCJMJUZ

PG NJYJOH QSPDFTTFT

t 5IJT CFIBWJPSW

IBT CFFO EVQMJDBUFE JO NVMUJQMF FOHJOF HFPNFUSJFT B UBSJPVT

MPBET t 0QUJDBM FOHJOF FNJTTJPOT NBUDI NFUBM FOHJOF FNJTTJPOT XFMM

0.4

0.8

1.2

600 bar

800 bar

1000 bar

1200 barCO

Em

issi

ons

[Vol

. %]

1500 RPM, 10% O2

SNL 0.42L Single

Kook, et al. SAE 2006-01-0197 1.6

-30 -25 -20 -15 -10 -5 SOI [CAD ATDC]

0

Considerable evidence suggests CO and UHC emissions are dominated by fuel-rich regions (mixing-limited)

For SOI retarded from MBT timing,

CO and UHC correlate inversely with

ignition delay

Improved emissions with additional

mixing time is inconsistent with CO

and UHC stemming from over-lean

mixtures

UH

C [g

/kg-

fuel

], τ i

g [°

CA

]

5

10

15

20

25

30

35

τig CO

UHC

2000 RPM, 6 bar IMEP

GM 1.9 L

Colban, et al. SAE 2008-01-1066 250

200

150

100

0 -45 -40 -35 -30 -25 -20 -15

SOI

CO

[g/kg-fuel]

t &OIBODFE NJYJOH BTTPDJBUFEinj reduces CO

XJUIincreased P

(Similar results reported by Opat, et al.

SAE 2007-01-0193)

t

-JUUMF JNQBDU PG 5 in on CO emissions

at the minimum (Opat, et al. SAE 2007-01-0193)

Numerical simulations and in-cylinder measurements clarify the spatial distributions of UHC

40

5

-10 0 10 20Crank Angle

0

10

15

20

25

30

35

App

aren

t Hea

t Rel

ease

Rat

e [J

/°CA

]

Simulation

-15°

Experiment

Bulk fuel distribution

shortly after EOI

Simulation

Experiment

-10°

-5°

Single-cycle images

are typically bi-modal

Simulation

Experiment

Simulation

TDC

Simulation indicates UHC

reaches ring-land

Experiment

30

Simulation results show that UHC (and CO) become separated into distinct ‘rich’ and ‘lean’ regions

15

10

5

0 -10 0 10 20

Crank Angle

20

25

30

35

40

App

aren

t Hea

t Rel

ease

Rat

e [J

/°CA

]

5°

Equivalence ratio UHC

3.0

0.0

2.3 1.5

0.8

30

15°

Equivalence ratio

UHC

CO

3.0

0.0

1.5 2.3

0.8 30°Equivalence

ratio

UHC

CO

These separate regions

persist throughout the

remainder of the cycle (see also Cook, et al. SAE 2008-01-1666)

Three dominant

regions form:

- Centerline and squish area (lean)

- Deep bowl (rich)

Experiments provide support for the simulated late-cycle UHC and CO distributions

Experiment

25°

30°

Simulation 50-cycle average 230.18 nm excitation PLIF Image (no imaging in bowl)

UHC is observed in the ring-land

crevice and centerline regions:

25°

Single-cycle 355 nm excitation PLIF Image

and can be strongest in a plane

bisecting the sprays

LIF-signal spectral characteristics near

the crevice and injector are consistent

with largely undecomposed fuel

CO is observed distributed through-

out the (fuel-lean) squish volume:

Experiment

25°

30°

Simulation 50-cycle average 230.09 nm excitation less offline 230.18 image (no imaging in bowl)

25°

30°

UHC is seen in the lower central

region of the bowl...

...but not on all cycles

Indicating that rich UHC sources may

be less important than the simula­

tions suggest

With advanced SOI, the squish volume becomes fuel-rich, and the bowl lean throughout

Simulations indicate that

UHC and CO emissions

stemming from the squish

volume dominate

Equivalence ratio

3.0

0.0

1.5 2.3

0.8 CO

30° UHC

Advanced SOI MBT Timing

UHC

CO

UHC 30°

CO

50-cycle average near 230 nm excitation (no imaging in bowl)

Increased squish volume CO and UHC seen with

advanced SOI is only consistent with rich mixture

25° 30°

50-cycle average, 355 nm excitation

Only diffuse, low UHC fluorescence is

observed within the bowl

Conversely, with retarded SOI the squish volume is over-lean and regions of the bowl over-rich

Simulations indicate that

increased UHC and CO

emissions stem from both

the squish volume and the

bowl

Equivalence ratio

3.0

0.0

1.5 2.3

0.8

30° UHC

CO

Retarded SOI MBT Timing 30°

UHC

COCO

UHC

50-cycle average near 230 nm excitation(no imaging in bowl)

Increased squish volume UHC with decreased CO

provides evidence supporting over-lean mixtures

30°

50-cycle average (no imaging in bowl)

Single-cycle image

CO and UHC PLIF images both sup­

port the simulated distributions

As load decreases, simulations show that the import­ance of lean emissions from the squish region increases

30° Equivalence

ratio

UHC

CO

3.0 bar IMEP 1.5 bar IMEP

3.0

0.0

1.5 2.3

0.8

Simulation results

Measured radial profiles of

UHC and CO fluorescence

signal support this finding:

Nor

mal

ized

CO

[ - ]

N

orm

aliz

ed U

HC

[ - ]

1

0.8

0.6

0.4

0.2

0

1

0.8

0.6

0.4

0.2

00 5 10 15 20 25 30 35 40

Radius [mm]

30°CA

UHC 1.5 bar

3.0 bar

4.5 bar

30°CA

CO

1.5 bar

3.0 bar

4.5 bar

Recap (and redress)

UHC and CO emissions from

PCI-like combustion systems

are found to stem from three

main regions of the cylinder.

The cylinder centerline region (including dribble)

The squish region, including the crevice

The central bowl region

t #PUI TJNVMBUJPOT BOE F YQFSJNFOUT JOEJDBUF UIBUbo

FNJTTJPOT GS PN UIF TRVJTI BOEwl regions will dominate

t 5IF TRVJTIPQFSBUJOH D

PS CPXM S FHJPOT DBO CF FJUIFS GVFM�SJDI PS GVFM�MFBO EFQFOEJOH PO UIF

POEJUJPO JOKFDUJPO

UJNJOH

MPBE

FUD�

t " U .#5 UJNJOH SJDI CPXM SFHJPO

BO PQUJNBM GVFM EJTUSJCVUJPO JT BDIJFWFE UIBU SFTVMUT JO B OPU�UPP­

BOE B OPU�UPP

�MFBO

U

TRVJTI S FHJPO� 4JNVMBUJPOT JOEJDB F

FNJT­

TJPOT TUFN BMNPTU FRVBMMZ GS PN UIFTF S FHJPOT

t 8JUI MJHIUFS MPBE PS SFUBSEFE 40* FNJTTJPOT GSPN UIF MFBO TRVJTI BSFB CFDPNF

NPSF QSPOPVODFE (increased Tin, high pressure EGR, VVA, decreased EGR)

t "U IJHIFS MPBE FNJTTJPOT GSPN UIF SJDI CPXM DPNF UP EPNJOBUF (increase mixing)