understanding diesel spray combustion · 2011-09-12 · gurpreet singh 8th international symposium...

Understanding Diesel Spray CombustionUnderstanding Diesel Spray Combustion

Lyle M. Pickett*Sandia National Laboratories

* Sponsor: DOE Office of Vehicle Technologies

Program Manager: Gurpreet Singh

8th International Symposium Towards Clean Diesel Engines TCDE 2011

What DO we understand about diesel spray What DO we understand about diesel spray combustion?combustion?

• A short review of conceptual models introduced about the subject:

Dec 1997

2

Other conceptual models differ slightly.Other conceptual models differ slightly.

Kosaka, Aizawa, Kamimoto 2005

• Compared to Dec’s model:– Liquid penetration upstream of lift-off

– Diffusion flame not shown at head of

jet.

3

jet.

– No specific description of fuel-rich

premixed flame.

Research shows that the near liftResearch shows that the near lift--off soot off soot formation region differs with fuel type. formation region differs with fuel type.

Pickett and Siebers 2003, 2006

• Compared to Dec’s model:– Different sketch of the rich premixed

flame.

– Soot not spatially coincident with lift-off

4

– Soot not spatially coincident with lift-off

length or rich flame.

– First soot formation region depends

upon fuel type.

Ignition transients towards a quasiIgnition transients towards a quasi--steady flamesteady flame

5

Bruneaux 2008

(presented at last TCDE)

Our current understanding: A picture of diesel Our current understanding: A picture of diesel combustion emerges.combustion emerges.

• Liquid-phase penetrates to steady liquid

length.

• Continued vapor penetration.

• First-stage ignition and appearance of

formaldehyde.

• Second-stage ignition, OH appearance,

and establishment of lifted flame.

Dec 1997

and establishment of lifted flame.

• Fuel-rich partially-premixed flame.

• Soot precursor and soot formation.

• Outer non-premixed diffusion flame.

• High soot concentration within head of

vaporized jet.

• Does NOT address how soot may fail

oxidization and become PM.

• Model shown does not address multiple

spray interactions.

6

What DON’T we understand (even What DON’T we understand (even conceptually) about diesel spray combustion?conceptually) about diesel spray combustion?

• What difficulties show up when modeling?– If we can’t model it, we really don’t understand it!

– Knowledge is not retained until it is added to a model.

• How the conceptual model changes with operating conditions.

• Ignition location and timing.

• Lift-off stabilization.• Lift-off stabilization.

• Jet-jet, jet-wall interactions, wall films.

• Sources of UHC and CO.

• Dense spray region.

• Why spray plume spreading angle varies.

• Structure of fuel-rich, premixed flame.

• Soot precursors and soot.

7

How does the conceptual model change when How does the conceptual model change when operating conditions are varied?operating conditions are varied?

• When approaching soot-free operation.

• Structure of fuel-rich premixed flame.

• Conceptual model at low-temperature combustion conditions.

8

5

6

What does the conceptual model look like if we What does the conceptual model look like if we have soothave soot--free combustion?free combustion?

21% O2

[1.0]

15% O2

[10.0]

10% O2

21% O2

[1.0]

15% O2

[10.0]

10% O2

1.6

1.7

φ(H)

No

Soot

OH Chemiluminescence Study in a constant-volume chamber • 1000 K, 42 bar • 50 µm injector hole• Single, isolated spray

Adiabatic mixture flame temperature [K]

φ

Lift-off

Heat-release

21% O2

10% O2

600 1000 1400 1800 2200 2600 30000

1

2

3

4

5

6040200

10% O2

[160]

6040200 6040200

10% O2

[160]2.1

Lift-off increases with

increasing EGR.

LII shows no soot.

Low temperature combustion

using EGR and fast mixing

with mixing-controlled HR.

Pickett and Siebers SAE 2004-01-1399

5

6

5

6

LeanLean--burn mixingburn mixing--controlled diesel controlled diesel combustion with no combustion with no EGR:EGR:

Conditions: Ta: 850 KO2%: 21%

OH Chemiluminescence (time-average DURING injection)

[1.0]

100 µm φ(H) = 1.4

806040200

[15.0]

Distance from injector [mm]

50 µmφ(H) = 0.6

600 1000 1400 1800 2200 2600 30000

1

2

3

4

5

600 1000 1400 1800 2200 2600 30000

1

2

3

4

5

• Even though fuel injector is open,

mixture at lift-off is fuel-lean.

• Global heat-release rate

proportional to mixing rate.

– Unlike HCCI or lean-burn SI.

Adiabatic mixture flame temperature [K]

φ

Lift-off21% O2


No diffusion flame !

Heat-release

SAE 2004-01-1399

Structure of lightlyStructure of lightly--sooting condition.sooting condition.10% O2, 1000 K

10

LII

HCHO/PAH

OH*

P

P

10

LII

HCHO/PAH

OH*

21% O2, 900 K

P

PPla

nar

355 L

IFP

lanar

• HCHO forms upstream of high-T reactions.

• Signal upstream of soot formation region: PAH fluorescence.

• Clear break visible between upstream and downstream signals

– Kinetic modeling shows HCHO not formed again, once destroyed.

– Similarly, PAH cannot be formed, destroyed and then formed again.

• HCHO consumption near jet center downstream of lift-off length.

0 10 20 30 40 50 60 70 80

10

5

0

-5

-10

OH*


0 10 20 30 40 50 60 70 80

10

5

0

-5

-10

OH*


Idicheria and Pickett SAE 2006-01-3434

LO

S

Structure at completely nonStructure at completely non--sooting condition.sooting condition.

LII

HCHO

LII

HCHO

10% O2, 900 K

P

P

LII

HCHO

LII

HCHO

21% O2, 850 K

P

P

• Consumption of HCHO marks location of high-T, fuel-rich,

premixed reaction.

• Fuel-rich reactions and diffusion flame, even for a no-soot

condition, but no PAH formation downstream.

0 10 20 30 40 50 60 70 80

10

5

0

-5

-10

OH*

0 10 20 30 40 50 60 70 80

10

5

0

-5

-10

OH*


0 10 20 30 40 50 60 70 80

10

5

0

-5

-10

OH*

0 10 20 30 40 50 60 70 80

10

5

0

-5

-10

OH*


ModerateModerate--soot soot condition imagingcondition imaging

10

LII

HCHO/PAH

10

LII

HCHO/PAH

10% O2, 1100 K

P

P

10

LII

HCHO/PAH

10

LII

HCHO/PAH

21% O2, 1000 K

• LII shows significant soot formation for both conditions

• HCHO completely enclosed within high-T zones

• No distinct break in signal; significant PAH formation

• HCHO formation at jet centerline for both conditions

0 10 20 30 40 50 60 70 80

10

5

0

-5

-10

OH*

0 10 20 30 40 50 60 70 80

10

5

0

-5

-10

OH*


0 10 20 30 40 50 60 70 80

10

5

0

-5

-10

OH*

0 10 20 30 40 50 60 70 80

10

5

0

-5

-10

OH*


Conceptual Conceptual model expanded to include firstmodel expanded to include first--stage chemistry.stage chemistry.

• No-soot condition

– HCHO upstream of lift-off

– No PAH formation

No-soot Condition

Low-soot Condition

10% O2

21% O2

Fuel-rich partially-premixed reaction

Fuel-lean premixed reaction

Diffusion flame

HCHO

HCHO

No-soot Condition

Low-soot Condition

10% O2

21% O2

Fuel-rich partially-premixed reaction

Fuel-lean premixed reaction

Diffusion flame

HCHO

HCHO

10 30 50 70 90 Distance from injector [mm]


Distance from injector [mm]0 10 20 30 40 50 60 70 80


10% O2

21% O2

Distance from injector [mm]Distance from injector [mm]




• Low-soot condition

– HCHO still upstream of lift-off

– Clear separation between HCHO and

PAH/soot

HCHO

Low-soot Condition

10% O2

21% O2

HCHO

PAH and sootPAH

PAH and sootPAH

HCHO

Low-soot Condition

10% O2

21% O2

HCHO

PAH and sootPAH

PAH and sootPAH



HCHO

Low-soot Condition

10% O2

21% O2

HCHO

PAH and sootPAH

PAH and sootPAH

HCHO

Low-soot Condition

10% O2

21% O2

HCHO

PAH and sootPAH

PAH and sootPAH




PAH/sootDistance from injector [mm]Distance from injector [mm]Distance from injector [mm]Distance from injector [mm]



10% O2

21% O2




• Low-soot condition

– HCHO still upstream of lift-off

– Clear separation between HCHO and

PAH/soot

Moderate-soot Condition10% O2

21% O2

10 30 50 70 90

PAHHCHO PAH and soot

PAH PAH and soot


HCHO

Moderate-soot Condition10% O2

21% O2

10 30 50 70 90

PAHHCHO PAH and soot

PAH PAH and soot


HCHO


PAH/soot

• Moderate-soot condition

– HCHO enclosed within high-T reaction

zone

– Overlap between HCHO and

PAH/soot



10% O2

21% O2

Distance from injector [mm]Distance from injector [mm]

LowLow--temperature combustion operation with temperature combustion operation with highhigh--EGR, and positive ignition dwell.EGR, and positive ignition dwell.

• Vastly different structure between conventional and LTC.

– Thick diffusion flame

Conventional Low-temperature combustionMusculus, Miles, Pickett

(in preparation)12.7% O2

– Thick diffusion flame

– Soot formation in the head

vortex.

– Significant upstream first-

stage combustion products

remain, becoming a source of

UHC and CO emissions.

17

Luminosity visualization

Conceptual Model SummaryConceptual Model Summary

• A vast range of conceptual models are needed to describe diesel combustion in different modes and operating conditions.

18

Beyond a conceptual diesel model, what Beyond a conceptual diesel model, what QUANTITATIVE data do we lack?QUANTITATIVE data do we lack?

• Almost everything, at high-temperature engine conditions.

• Liquid volume fraction and droplet size in the dense spray region and near the liquid length.

• Mixture fraction (fuel/air ratio) distribution.

• Velocity and turbulence.

• Soot volume fraction and structure distribution, particularly during • Soot volume fraction and structure distribution, particularly during transients.

• Internal injector geometry for working injectors.

• Information about internal injector cavitation and flows.

• Can we build this type of dataset?

• Underlined to be addressed briefly in the rest of presentation.

19

The TNF workshop pursues quantitative The TNF workshop pursues quantitative datasets to understand turbulent flames.datasets to understand turbulent flames.

• Use a series of well-defined, canonical flames to promote model development applicable to turbulent combustion.

• This type of dataset and focused modeling effort does not yet exist for diesel engine conditions.

>90 participants at last workshop.

20

Building quantitative diesel spray datasets Building quantitative diesel spray datasets through the Engine Combustion Network.through the Engine Combustion Network.

• Collaborative modeling/experimental data archive.

• http://www.sandia.gov/ECN

21

ECN

activities

-Predictive

Goals of the Engine Combustion Network.Goals of the Engine Combustion Network.

Experim

enta

l

increasing sophistication

-Well-controlled B.C.s

-Engine conditions

-Quantitative

-Predictive

-Design optimization

-New hardware identifiedcurrent

increasing sophistication

Engine CFD Modeling

Experim

enta

l

Industrial and academic collaboration in the Industrial and academic collaboration in the Engine Combustion Network for “Spray A”Engine Combustion Network for “Spray A”

Michigan Tech. Univ.Vessel temperature composition

Argonne (x-ray source)Internal needle movementNear-nozzle liquid volume

IFP (France)Spray velocityCombustion

General MotorsMixingDroplet sizing

BoschDonates 10 “identical” common-rail injectors 90°C, 1500 bar

CMT (Spain)Rate of injectionSpray momentum

SandiaLiquid and vapor mixingCombustion diagnostics

Meiji (Japan)Soot formation

CaterpillarInt. nozzle geometryPenetration

• Operation at the same injector and ambient conditions.

• Voluntary participation.

900 K, 60 bar

Benefits of standardizing at an operating Benefits of standardizing at an operating condition for advanced diesel research.condition for advanced diesel research.

• Conceptual model changes with conditions—then keep conditions constant so that we can understand each other.

• Follows direction of successful research activities using more basic flames (TNF workshop).

• Leveraging of work by many experimental and modeling activities will accelerate research.

Spray A SpecificationsAmbient gas temperature 900 K

Ambient gas pressure 6 MPa

Ambient gas density 22.8 kg/m3

Ambient gas composition 15% O2, 0% O2

Common rail fuel injector Bosch solenoid-activated, generation 2.4

Fuel injector nozzle outlet diameter 0.090 mm

Nozzle K factor 1.5 { K = (dinlet – doutlet)/10 [use µµµµm] }

Nozzle hydro-erosion Discharge coefficient = 0.86 with 100 bar ∆∆∆∆P.

Spray full included angle 0°°°° (1 axial hole)

Fuel injection pressure 150 MPa

Fuel n-dodecane

Fuel temperature at nozzle 363 K (90°°°°C)

Common rail volume/length 22 cm3/ 28 cm (Use GM rail model 97303659)

Distance, injector inlet to common rail 24 cm

Fuel pressure measurement 7 cm from injector inlet / 24 cm from nozzle

Injection duration 1.5 ms

Approximate injector driver current 18 A for 0.45 ms ramp, 12 A for 0.345 ms hold

Spray A Specifications

Controlled highControlled high--T, highT, high--P experimental P experimental participation at Spray A.participation at Spray A.

Institution Facilities Personnel

Sandia Preburn CV Lyle Pickett, Julien Manin

IFPEN Preburn CV Gilles Bruneaux, Louis-Marie Malbec

CMT Cold CV, Flow PV Raul Payri, Michele Bardi

Chalmers Flow PV Mark Linne

GM Flow PV Scott ParrishGM Flow PV Scott Parrish

Mich. Tech. U. Preburn CV Jeff Naber, Jaclyn Nesbitt, Chris Morgan

Argonne Cold V, X-ray Sync. Chris Powell, Alan Kastengren

Caterpillar Flow PV Tim Bazyn, Glen Martin

Aachen Flow PV Heinz Pitsch, Joachim Beeckmann

Meiji U. Preburn CV Tetsuya Aizawa

Seoul Nat. U. Preburn CV Kyoungdoug Min

Eindhoven U. Preburn CV Maarten Meijor, Bart Somers

BLUE: In progress Red: Commencing soon.

First ECN workshop held, 13First ECN workshop held, 13--14 May 2011.14 May 2011.

• 60 participants met in Ventura, California before ILASS meeting.

• NOT a conference, but an opportunity for real experimental and modeling exchange.

– Coordinators gathered experimental and modeling results before conference to

compare side by side.

– Discussed standardization, quantification of uncertainties, and best practices for

model comparison.model comparison.

– Future spray targets identified.

• Results focused on Spray A and baseline n-heptane conditions.

• Results discussed here.– Geometry, needle movement, and rate of injection.

– Liquid-phase penetration (Mie scatter)

– Vapor-phase penetration (Schlieren) imaging

– Ignition delay

– Lift-off length

– Soot formation

26

Nozzle internal geometry measurementsNozzle internal geometry measurements

X-ray phase-contrast imaging

(Argonne)

Silicone molds (CMT)

27

X-ray tomography

(Caterpillar)

Injector #211201

SEM

(Sandia)

Optical

(Sandia)

PhasePhase--Contrast Imaging of Needle MotionContrast Imaging of Needle Motion

28

Mold DefectsInjector #210675

d/dt

Quantification of liquid penetration length.Quantification of liquid penetration length.

1. Side-illumination, high-res., f/4

2. Side-illumination, fast, f/4

3. Side-illumination, saturated, f/1.2

532 nm CW

532 nm CW

532 nm CW

1. Side-illumination, high-res., f/4

7. Diffuser back-illumination

3% of max. (Siebers):

Liq. Length = 10.5 mm

• Different illumination methods appear to yield

different liquid penetration.

• Beam steering affects quantification of

extinction with diffuser back illumination.

• Cross comparison with extinction shows

estimates for liquid volume is < 0.15% at

traditional 3% of max. scatter position.. 29

4. Sheet-illumination, saturated, f/1.2

5. Head-illumination, f/2

7. Diffuser back-illumination

8. Schlieren

I/I0

532 nm CW

Halogen lamp

Halogen lamp

450 nm LED

Vapor and liquidVapor and liquid--phase penetration shows phase penetration shows reasonable similarity at IFPEN and Sandia.reasonable similarity at IFPEN and Sandia.

Schlieren visualization.Liquid Mie-scatter border in blue.0% O2: non-reacting

IFP

70

80

90

100

Ma

xim

um

Pe

ne

tra

tion

[m

m]

Sandia

IFPReacting

Penetration

Sandia

30

0 500 1000 1500 2000 2500 3000 35000

10

20

30

40

50

60

70

Ma

xim

um

Pe

ne

tra

tion

[m

m]

Time ASI [µs]

Liquid

Vapor Non-reacting

80

Vapor boundary derived

Spray A vapor penetration shows reasonable Spray A vapor penetration shows reasonable agreement at 4 different institutions.agreement at 4 different institutions.

• Despite the significant challenge to control

high-T, high-P boundary conditions.

• Ongoing work to understand the sources

of discrepancy, including post-processing.

• Leveraging experimental work is possible!

IFP CMT

0 500 1000 1500 2000 2500 30000

10

20

30

40

50

60

70

80

Ma

xim

um

Pe

ne

tra

tion

[m

m]

Time ASI [µs]

Sandia

IFPEN

Caterpillar

CMT

Liquid

Vapor boundary derived from schlieren imaging(0% O2)Sandia Caterpillar

Questions remain when experimental Questions remain when experimental “boundaries” of spray are compared to models:“boundaries” of spray are compared to models:

• What does a spreading angle measured by schlieren really mean?– Low or intermediate fuel mass fraction or mixture fraction?

– How does schlieren measurement sensitivity change with conditions?

• Is the “liquid” spreading angle the same as that of the “vapor”?

• Should the model be tuned for spreading angle or spray penetration?

– Are spreading angle and spray penetration consistent?

• Guidance is needed because spray model constants are often tuned based on spreading angle.

• Quantitative mixing data is needed.

32

Quantitative mixing diagnostic in harsh highQuantitative mixing diagnostic in harsh high--temperature, hightemperature, high--pressure environment.pressure environment.

• Use Rayleigh scattering in vapor portion of fuel jet.

– Overcomes significant

temperature/composition uncertainties

compared to laser-induced

fluorescence.

• Measure both IR,a, IR,j

Raw image

Nf/Na

Raw image

Nf/Na

MIE

R,a R,j

– Allows in-situ calibration for IR,a variation

in laser sheet intensity.

– Beam-steering or divergence addressed

by using IR,a on bottom and top of jet.

• Measurement provides – Fuel mixture fraction (mass fraction)

– Mixture temperature

mix

a

fa

faaf

aR,

jR,

T

T

/NN1

/NN/σσ

I

I

+

+=

)/NN(T famix f=

20 30 40 50 60

-10

-5

0

5

10

Distance from Injector [mm]

700

800

900

1000

[K]

Tmix

20 30 40 50 60

-10

-5

0

5

10


700

800

900

1000

[K]20 30 40 50 60

-10

-5

0

5

10

20 30 40 50 60

-10

-5

0

5

10


700

800

900

1000

[K]

Tmix

When modeled* penetration matches When modeled* penetration matches experiment, modeled mixing is also accurate.experiment, modeled mixing is also accurate.

0.25

*Variable radial profile (Musculus and Kattke 2009)

100

Spreading-angle adjusted

to match experimental

(schlieren) penetration

rate.

Model predictions of mixture

fraction (same spreading angle)

agree with experimental

Rayleigh measurements.

(see SAE 2011-01-0686)

-10 -8 -6 -4 -2 0 2 4 6 8 100.00

0.05

0.10

0.15

0.20

0.25

x = 25 mm

x = 45 mm

Radial distance [mm]

mix

ture

fra

ctio

n

Experimental

Uncertainty

Model, θ=21.5°

0 1000 2000 3000 40000

20

40

60

80

100

Time ASI [µs]

Va

po

r p

en

etr

atio

n [m

m]

Experimental

Model, θ = 21.5°150 MPa

50 MPa

100 MPa

Mixing results confirmed at other injector and Mixing results confirmed at other injector and ambient operating conditions.ambient operating conditions.

• Mixture distribution shows self-similar, 1/x decay

– Such behavior is typical of gas jets.

– Documented now for the first time in

diesel sprays at engine conditions.

Experimental Conditions

1000 K gas temperature

14.8 kg/m3 gas density

0% O2-inert

1540 bar injection pressure

0.100 mm nozzle orifice

n-heptane

0.2

x=20 mm

x=40 mm

-10 -8 -6 -4 -2 0 2 4 6 8 100

0.05

0.1

0.15

0.2

Radial distance [mm]

mix

ture

fra

ctio

n

0 10 20 30 40 50 600

0.05

0.1

0.15

0.2

0.25

Axial distance [mm]

Ce

nte

rlin

e m

ixtu

re fra

ctio

n

Experimental

Uncertainty

Model, θ=24°

ModelinModelingg comparisons at ECN comparisons at ECN workshopworkshop

• Coordinators: Evatt Hawkes, UNSW and Sibendu Som, Argonne

• Input from 9 groups.

• Argonne National Laboratory: Sibendu Som, Douglas Longman

• Cambridge University: Giulio Borghesi, Epanimondas Mastorakos

• Universitat Politècnica de València CMT: Ricardo Novella, José Pastor, Francisco Payri, J.M. DesantesFrancisco Payri, J.M. Desantes

• TU Eindhoven: Bart Somers, Cemil Bekdemir, L.P.H. de Goey

• Penn. State: Dan Haworth, Hedan Zhang, Subhasish Bhattacharjee

• Politecnico di Milano: Gianluca D’Errico, Tommaso Lucchini, Daniele Ettore

• Purdue: John Abraham, Chetan Bajaj

• UNSW: Yuanjiang Pei, Sanghoon Kook, Evatt Hawkes

• U. Wisconsin ERC: Yue Wang, Gokul Viswanathan, Rolf Reitz, Chris Rutland

0.05

0.1

0.15

0.2

0.25

Mix

ture

Fra

ctio

n

x = 20 mm Experiment

Model 1

Model 2

Model 3

Model 4

Model 5

0.05

0.1

0.15

0.2

0.25

Mix

ture

Fra

ctio

n


Model 1

Model 2

Model 3

Model 4

Model 5

Comparison of experimental results to various Comparison of experimental results to various CFD predictionsCFD predictions..

37

-15 -10 -5 0 5 10 15

0

0.05

0.1

0.15

0.2

0.25

Radius [mm]

Mix

ture

Fra

ctio

n


Model 1

Model 2

Model 3

Model 4

Model 5

0

0

-15 -10 -5 0 5 10 15

0

0.05

0.1

0.15

0.2

0.25

Radius [mm]

Mix

ture

Fra

ctio

n


Model 1

Model 2

Model 3

Model 4

Model 5

HighHigh--speed luminosity to characterize ignition speed luminosity to characterize ignition and combustion. and combustion.

filter: 430 nm (10 nm BPF)

exposure: 99 µs

camera: Phantom (non intensified)

Dynamic background correct.

Caterpillar

• Shows upstream high-temperature chemiluminescence as well as downstream soot luminosity.

• High-temperature ignition is clearly marked.

38

Schlieren can also be used to indicate highSchlieren can also be used to indicate high--T T ignition, as well as cool flame beforehand.ignition, as well as cool flame beforehand.

Sandia

Liquid

50

Ma

xim

um

Pe

ne

tra

tion

[m

m] Vapor, non-react

Schlieren with luminosity boundary overlaid

0 200 400 600 800 10000

10

20

30

40

Ma

xim

um

Pe

ne

tra

tion

[m

m]

Time ASI [µs]

Liquid

Vapor, reactVapor, cool flame

Ignition

Cool flame

39

See SAE 2010-01-2106

SAE 2009-01-0658

Modeled liftModeled lift--off length versus %Ooff length versus %O22 at at baseline nbaseline n--heptane conditions.heptane conditions.

• What is responsible for the difference?– Mixing and turbulence model, chemical mechanism, or other?

• Lift-off length was not defined the same way for each model.

Measurements of soot distribution within Measurements of soot distribution within combusting diesel sprays.combusting diesel sprays.

D2

5

10

15

Soot volume

fraction [ppm]#2 Diesel

Flame Lift-offTotal soot in spray is 65 µg,

about 150 times Euro VI

PM emissions limit!

• Soot concentration (steady) measured by laser extinction and LII.

• Factor of five reduction in total soot formed within reacting spray for biodiesel.

41

BD

Distance from Nozzle [mm]0 10 20 30 40 50 60 70 80

5

Soy Methyl Ester

Biodiesel 1000 K

67 bar

15% O2

TEM probe within the

combustion vessel

Spray direction

TEM grid

TEM grid held within the probe to

protect it from excessive heating

Soot sampling from sprays within highSoot sampling from sprays within high--temperature, hightemperature, high--pressure combustion vessel.pressure combustion vessel.

3 mm diameter

TEM grid:

copper mesh covered by an amorphous

carbon film

Restrictive passage quenches

reaction and protects the TEM grid.

0 100 200 300 400 500

0

100

200

300

400

500

siz

e [nm

]

Soot-laden gas

skims past the

probe, enters

through a 1-mm

diameter hole,

and deposits on

the TEM grid.

Jet TEM image of soot particles

0

20 30 40 50 60 70 80

5

10

15

20

30 mm collection position

0 100 200 300 400 500

100

200

300

400

500 43

siz

e [nm

]

0

20 30 40 50 60 70 80

5

10

15

20


440 100 200 300 400 500

100

200

300

400

500

siz

e [nm

]

20 30 40 50 60 70 80

5

10

15

20

0


450 100 200 300 400 500

100

200

300

400

500

siz

e [nm

]

0

20 30 40 50 60 70 80

5

10

15

20


0 100 200 300 400 500

100

200

300

400

500 46

siz

e [nm

]

A key input for soot models, the particle size A key input for soot models, the particle size and shape are now characterized.and shape are now characterized.

0

100

200

300

400

0

100

200

300

400

50 mm 86 mm

23% m-xylene77% n-dodecane

0 10 20 30 40 50 60 70 80

5

10

15

20

0 100 200 300 400 500

400

5000 100 200 300 400 500

400

500

Peak soot

location

Soot oxidation

region

Size [nm]

SummarySummary

• Our current understanding of diesel combustion is summarized in conceptual model form.

– But even the conceptual model depends upon specific operating conditions.

• New “Spray A” initiative started for the Engine Combustion Network.– Filling the need for an advanced (quantitative) experimental dataset.

– Provides a pathway towards more predictive spray combustion, more efficient

optimized engines.optimized engines.

– ECN workshop held. Data is available online.

• Results demonstrate reasonable similarity between institutions.– Opportunity to leverage experimental effort.

• CFD model improvement may proceed in a more quantitative way.– Mixture fraction comparison shows significant variation between models.

• Lift-off length trends follow.

– Liquid length defined more carefully in terms of liquid volume fraction.

– Soot volume fraction and soot distribution size and shape.

48

Acknowledgements

• Dennis Siebers, Mark Musculus, Sandia National Laboratories

• Caroline Genzale, Georgia Institute of Technology

• Gilles Bruneaux, Laurent Hermant, Louis-Marie Malbec, IFP Energies Nouvell.

• Chris Powell, Alan Kastengren, Sibendu Som, Argonne National Laboratories

• Tim Bazyn and Glen Martin, Caterpillar Inc.

• Cherian Idicheria, GM R&D

• Sanghoon Kook and Evatt Hawkes, UNSW Australia• Sanghoon Kook and Evatt Hawkes, UNSW Australia

• Raul Payri, Julien Manin, Michele Bardi, CMT Motores Térmicos

• Gurpreet Singh, Funding provided by US DOE Office of Vehicle Technologies

• French Ministry of Ecology, Energy, Sustainable Development and Sea

• David Cook and Godehard Nentwig, Robert Bosch LLC, donation of injectors.

Slides for questionsSlides for questions

50

Fuel temperature specification: 90Fuel temperature specification: 90°°°°°°°° CC

Stagnant fluid before fuel enters orifice

750 1.2

n-dodecane fuel properties (1 atm)

90° C

needle

(open)

viscosity

density

40 50 60 70 80 90 100 110 120650

670

690

710

730

n-dodecane temperature [C]

Fu

el d

en

sity

[ k

g/m

3 ]

0.2

0.4

0.6

0.8

1

Fu

el a

bs. vis

co

sity

[cP

]

Liquid length predictions

Spray A ambient

40 50 60 70 80 90 100 110 12010

10.5

11

11.5

12

12.5

13

n-dodecane temperature [C]

Liq

uid

len

gth

[m

m]

0-1-2-3-4-5

Distance [mm]

How does one control fuel temperature?How does one control fuel temperature?

Temperature-controlled water inlet

Copper gasket seal

Radial Seal

Cooling JacketWater

65°C, steady-stateRadial Seal

Ceramic insulator

Teflon insulationMetal Housing

CRIP2 injector

90°C, steady-state

150°C, steady-state

65°C, steady-state

LaserLaser--induced phosphorescence measurements induced phosphorescence measurements show that preburn heats the injector.show that preburn heats the injector.

80

90

100

110

No

zzle

su

rface t

em

pera

ture

[°° °°C

]

IFP

53

-1 0 1 2 3 4 5 6 7 8 940

50

60

70

80

Time after spark ignition [s]

No

zzle

su

rface t

em

pera

ture

[

Without ceramic cover

With ceramic cover

Internal injector temperature distributionInternal injector temperature distribution

0 mm

1 mm1 mm1.5 mm2 mm

3 mm

4 mmSteady state T°

decreases with

distance to

nozzle tip

T°is affected by

preburn. Distance to

nozzle acts as a filter.

Liquid length increases slightly with increasing Liquid length increases slightly with increasing time after start of injection.time after start of injection.

11.5

12

12.5

13

Ma

xim

um

Pe

ne

tra

tion

[m

m] Sandia

55

0 1000 2000 3000 4000 5000 60009

9.5

10

10.5

11

Ma

xim

um

Pe

ne

tra

tion

[m

m]

Time ASI [µs]

• Suggests that fuel cools the

injector tip with increasing

time ASI.

• 0.5 mm → approx. 20°C.

10-1

100

Re

lativ

e li

gh

t in

ten

sity

[a

.u.]

Light scattering intensity drop off with respect Light scattering intensity drop off with respect to axial distance.to axial distance.

3% of maximum

1 2 3 4 5 6 7 8 9 10 11 12 13 14 1510

-4

10-3

10-2

Distance [mm]

Re

lativ

e li

gh

t in

ten

sity

[a

.u.]

Diag. 1

Diag. 2

On-chip mean

Inst. image mean

±2σ inst. max. liq. len.

56

3% of maximum

LineLine--ofof--sight extinction may offer a more sight extinction may offer a more consistent liquid length measurement since it consistent liquid length measurement since it is inherently based on relative intensity.is inherently based on relative intensity.

1.2

1.4

1.6

1.8

2

Ce

nte

rlin

e o

ptic

al t

hic

kn

ess τ

[-] HeNe, Diagnostic 9

Back illum.,

Diagnostic 7

57

0 2 4 6 8 10 12 14 16 18 200

0.2

0.4

0.6

0.8

1

1.2

Axial distance [mm]

Ce

nte

rlin

e o

ptic

al t

hic

kn

ess

Beam steering

LineLine--ofof--sight extinction measurements can sight extinction measurements can also help provide an estimate for LVF at the LL.also help provide an estimate for LVF at the LL.

0.05

0.1

0.15

0.2

pa

th-le

ng

th a

ve

rag

ed

LV

F [%

] Axial distance = 10.4 mm (ττττ = 0.374)liquid width = 1.4 mm (5% max thickness)

6090

120

Angular scattering intensity for a

5 µm n-dodecane droplet in air

58

10-1

100

101

0

0.05

Assumed droplet diameter [µm]

pa

th-le

ng

th a

ve

rag

ed

LV

F [%

]

30

210

240270

300

150

330

180 0

103

105

220 mrad

The role of spray combustion research for The role of spray combustion research for highhigh--efficiency engines.efficiency engines.

• Future high-efficiency engines use direct injection.

– Diesel, gasoline direct injection, partially-

premixed compression ignition.

• Complex interactions between sprays, mixing, and chemistry.

– Multiple injections

BLUE: liquid boundary

Schlieren: vapor boundary

– Multiple injections

– Mixing driven by spray

– Two-phase system

– Complicated internal flows within injectors

• Optimum engine designs discovered only when spray combustion modeling becomes predictive.

understanding diesel spray combustion · 2011-09-12 · gurpreet singh 8th international symposium...

Documents