Use of Kinetic Models to Predict Fuel Impact on Engine ...€¦ · 1) Kinetics model (ignition surrogate) selected from Westbrook, et al 2) Compared to surrogates and kinetics proposed

Med Colket and Steve Zeppieri

4th Annual Fuel Research Meeting

Multi-Agency Coordinating Council for Combustion Research

Argonne National Laboratories, Chicago, Ill

September 20-22, 2011

Use of Kinetic Models to Predict Fuel

Impact on Engine Performance

2

Background

Secure supply

Military needs guaranteed fuel supply in time of war

Global Warming:

Combustion of fossil-derived fuels

Combustion emissions (soots/particulates)

Physics-based combustor models

Existing models barely sufficient for design predictions

Physical vs. chemical effects

Physical effects presumed to dominate

Complexity of fuels

Large number of species

Unknown/uncertain reaction rate laws

DoD needs

AF – 50% utilization of alternative fuels by 2016 (in US)

Navy – Sailing the „Great Green Fleet‟ by 2016

SERDP – Understand impact on emissions

3

Objectives and Outline

Objectives:

Apply methods/tools that can incorporate both physical and chemical effects to predicting impact on engine performance

Outline:

Recommendations from 2010!!!!

Review general modeling approach

Applications to

Lean blow-out (review)

Altitude Relight

Emissions Unburned hydrocarbons

Soot

Summary

Recommendations

Challenge Problems

4

Recommendations/Needs – from 2010!!!!!

Validated mechanisms that predict ignition and extinction

particularly aromatics and cycloalkanes

Broader/validated tool for selecting surrogates that match properties of

petroleum fuels and synthetic fuels

Soot growth models incorporating effect of PAH surface addition

Models that describe HC emissions fingerprint and changes with fuels

Inclusion of ignition criteria for certification of alternative fuels

What should be the standard?

Incorporate physical effects?

New ignition data at low pressure needed

Fuel components

Jet Fuel

Alternative jet fuels

5

Tools for Assessing Fuel Impact on Performance

Alternate Fuel:

Phys & Chem Prop

Surrogate for

alternate fuel

Detailed Kinetics

for Surrogate

Selection of physics-

based tools sets

Simulations Probable impact

on combustor

Interpretations and

Engine Impact

Testing

Combustor Simulation

H. Pitsch

Emissions

Quenching Due to Fresh Reactant

Mixing with Reacting Mixture

0

500

1000

1500

2000

2500

3000

3500

-1.0E-06 9.0E-06 1.9E-05 2.9E-05 3.9E-05

Time (sec)

Te

mp

era

ture

(K

)

Ymix = 0

Ymix = 2,000

Ymix = 6,000

Ymix = 20,000

Ymix = 30,000

Ymix = 35000

Ymix = 40,000

Ymix = 60,000

Ymix = 100,000

Ymix = 200,000

Ignition (Ymix=0)

Ignition (Ymix=2,000)





tignition =

4.92 microsec

phi=1

all vapor

Ignition Modeling

6

Modeling Methods – Physical Models

Lean Blow Off – Extinction of a „Perfectly Stirred Reactor‟

Altitude Relight – Revised SENKIN with added mixing and vaporization

Initial energy provided by prescribed spark

Soot – (PSR) Network Reactor Code with coupled soot equations

Simplified, kinetics-based model of physical problem (CHEMKIN based)

1600

2100

2600

3100

3600

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

Tau (sec)

Tem

pera

ture

(F

)

n-Heptane/Air

Ti = 1857 F

P = 6 atm.

Phi = 2.0

PSR

Extinction Time

PSR Ignition Time

Extinction

Time

Ignition

Time

Te

mp

era

ture

Tau (sec)

Volume, V

Residence time, Temperature, T

Enthalpy, hk

Mass Fraction, Yk

inlet outletm

Y*k

h*k

m

Yk

hk

Heat loss

Q

. .Volume, V


Enthalpy, hk

Mass Fraction, Yk

inlet outletm

Y*k

h*k

m

Yk

hk

Heat loss

Q

Volume, V


Enthalpy, hk

Mass Fraction, Yk

inlet outletm

Y*k

h*k

m

Yk

hk

Heat loss

Q

. .

Y*S YSSoot Sections, YS

Network Reactor Simulation

Fuel-spray shear layer

Recirculation zones

Quench

zones

Burn-out

zones

Volume, V

Residence time,

Temperature, T

Enthalpy, hk

Mass Fraction, Yk

inlet outletm

Y*k

h*k

m

Yk

hk

Heat loss

Q

. .Volume, V

Residence time,

Temperature, T

Enthalpy, hk

Mass Fraction, Yk

inlet outletm

Y*k

h*k

m

Yk

hk

Heat loss

Q

Volume, V

Residence time,

Temperature, T

Enthalpy, hk

Mass Fraction, Yk

inlet outletm

Y*k

h*k

m

Yk

hk

Heat loss

Q

. .

7

Surrogate Fuel Identification - Demonstration

1) Kinetics model (ignition surrogate)

selected from Westbrook, et al

2) Compared to surrogates and kinetics

proposed by others

3) Benchmarked model against Vasu et al

(C&F) expts

4) Kinetics models/surrogates predict

ignition times a factor of two too long

Liq vol

fraction

Mass

fraction

Molar

fraction

n-heptane 25 0.227 0.226

iso-octane (2,2,4-trimethylpentane) 20 0.184 0.161

methylcyclohexane 35 0.358 0.364toluene 20 0.231 0.249 Min Max Min Mean Max

grams/cc 0.752 0.775 0.84 0.796 0.817 0.837

MW 99.712

H/C 1.899 1.85 1.85 1.9 2.1

SP(mm) 21.14 19.00 19 21 26

HtComb (MJ/kg) 43.31 42.60 42.90 43.11 43.40

ONR JP-5 Spec PQIS DatabaseMil Spec PQIS Database

Surrogate – blend of reference

Fuels obtained by matching H/C,

Smoke point, Ht. of combustion

1

10

100

1000

10000

0.5 0.7 0.9 1.1 1.3 1.5

Ign

itio

n T

ime

(µs)

1000/T (1/K)

Models Correctly Predict Trends

(but a factor of two higher than Vasu et al, C&F 2008)

Vasu et al [1]

Vasu et al [1] current work

Dean et al [2]

Dean et al [2] current work

Ignition Surrogate

Dooley, et al (2010)

CSE - Jet-A

Model: P = 20 atm; φ = 1.0

Shock tube data at φ = 1.0 and scaled to 20 atm(time ~ P-1)

Based on HC classes

8

Observations from LBO Simulations

Small differences due to

physical effects JP-8 to JP-5

One fuel component had

longer PSR extinction times

– and controlled surrogate

extinction time. (Perhaps

due to uncoupled reaction

mechanism)

Most mechanisms are not (or

poorly) validated against

flame extinction

Better reaction models for

cycloalkanes are needed

Comparison of Extinction Behavior for Various

Fuels & Representative Fuel Blend

(10 atm, 750K Inlet Conditions)

0.000

0.001

0.002

0.003

1400 1800 2200 2600 3000

Adiabatic Flame Temperature (K)

Ex

tin

cti

on

Tim

e (

s)

HeptaneIso-octaneMCHToluene4 Component Fuel Blend

Mole Fraction

n-Heptane 22.63%

iso-Octane 16.06%

Toluene 24.95%

MCH 36.36%

Based on PSR Extinction Times

9

AutoIgnition: Strong pressure dependence

NTC shifts to higher temperatures with increased P

1.E-06

1.E-05

1.E-04

1.E-03

1.E-02

1.E-01

1.E+00

1.E+01

0.6 0.8 1 1.2 1.4

Ign

itio

n D

ela

y [

sec.]

1/Temperature (x1000) [K-1]

Stoichiometric Heptane-Air MixturesLLNL n-Heptane MechanismConstant Volume Calculations

0.5 atm

2.5 atm

10 atm25 atm

50 atm

(tau ~ P-1)

in „hot‟ ignition regime

NTC

Review from

last year

10


NTC relatively region unimportant for present day engines

1.E-06

1.E-05

1.E-04

1.E-03

1.E-02

1.E-01

1.E+00

1.E+01

0.6 0.8 1 1.2 1.4

Ign

itio

n D

ela

y [

sec.]



0.5 atm

2.5 atm

10 atm25 atm

50 atmNTC

Altitude Relight and

Augmentor Ignition

11


NTC of increased interest with increased P

1.E-06

1.E-05

1.E-04

1.E-03

1.E-02

1.E-01

1.E+00

1.E+01

0.6 0.8 1 1.2 1.4

Ign

itio

n D

ela

y [

sec.]



0.5 atm

2.5 atm

10 atm25 atm

50 atm

Regime of interest in advanced,

high efficiency (next gen) engines

– and reciprocating engines!

12

Challenge Problem: Altitude Relight

Spark survival of >10 msec followed by ignition is inconsistent with

mixing/cooling rates

Spark energy

should

dissipate in 1-3

milliseconds !

Explanation may

be fuel-type

dependent

Spark Trajectories from Reid, Rogerson and Hochgreb, AIAA 2008-957

13

Igniter

Spark kernel

Flame kernel Flame propagation/

stabilization (not

modeled)

Fuel Droplets

Size Distribution,

fuel/air ratio

Air

Temperature,

Pressure

Droplet Evaporation

At Tkernal, and

Detailed kinetics

Spark

Energy added

Ignition? Ymix

„fresh‟

reactant

Reacting, vaporizing

mixture

t = 0

Coupled Kinetics/Mixing/Vaporization Model:

Flame Kernel Model

Full detailed kinetics (SERDP surrogate)

Specify SMD (drop size distribution)

Specify mixing rate Ymix (1/sec)

Added energy – after early, rapid relaxation

14

Altitude Relight/Ignition Modeling – Mixing Effect

Turbulent Mixing can quench chemical reactions and inhibit ignition

Quenching Due to Fresh Reactant

Mixing with Reacting Mixture

0

500

1000

1500

2000

2500

3000

3500

-1.0E-06 9.0E-06 1.9E-05 2.9E-05 3.9E-05

Time (sec)

Te

mp

era

ture

(K

)

Ymix = 0

Ymix = 2,000

Ymix = 6,000

Ymix = 20,000

Ymix = 30,000

Ymix = 35000

Ymix = 40,000

Ymix = 60,000

Ymix = 100,000

Ymix = 200,000

Ignition (Ymix=0)






tignition =

4.92 microsec

phi=1

all vapor

Example conditions:

Phi = 1

FracL=0 (fuel vaporized)

P=1.422 atm

T=440.9 K

E/M=943 J/gm

Gaseous

fuel: no

vaporization

15

Altitude Relight/Ignition Modeling – Mixing/Vaporization

Coupled Mixing with Vaporization complicates inhibition (thermal quench)

250

750

1250

1750

2250

2750

0 2 4 6 8 10

Tem

pe

ratu

re [K

]

Time [msec.]

SMD-20

SMD-40

SMD-60

SMD-80

SMD-100

SMD-120

SMD-140

SMD-160

SMD-180

SMD-200

SMD-220

All fuel initially liquid (variable Sauter Mean Diameter); Total phi = 3.5

Tinit = 265K; P = 0.36 atm; Ymix = 625 sec-1

20

220

16

Altitude Relight/Ignition Modeling – Mixing/Vaporization

Coupled Mixing with Vaporization complicates inhibition (vapor quench)

All fuel initially liquid (variable Sauter Mean Diameter); Total phi = 3.5

Tinit = 265K; P = 0.36 atm; Ymix = 625 sec-1

0.0

0.5

1.0

1.5

2.0

2.5

3.0

0 2 4 6 8 10

Vap

or

Ph

i

Time [msec.]

SMD-20SMD-40SMD-60SMD-80SMD-100SMD-120SMD-140SMD-160SMD-180SMD-200SMD-220

20

220

17

Prediction of Fuel Effects of Altitude Relight

SERDP Mechanism*

Three fuels

77% Dodecane/23% m-Xylene

Dodecane

m-Xylene

* UIC + USC + UTRC

SMD fracL phi E/M

10 0.9 0.5 1300

30 0.93 0.75 1400

50 0.96 1 1400

70 0.99 1.25 1750

90 1.5

110 1.75

130 2

150 2.25

170 2.5

190 2.75

3

3.25

3.5

3.75

4

4.25

4.5

4.75

5

5.25

5.5

Use fuel with range in ignition

characteristics – accentuate

effects

Identify conditions suitable for

sustained kernel

18

Final Kernel Temperatures for 1500J/gm

n-dodecane

Fuel: nC12

Boiling Point: 533 K

Spark Energy/Mass: 1500 J/gm

Fuel: 64.38% nC12 / 35.62 m-Xylene



Fuel: m-Xylene



n-dodecane/

m-xylene m-xylene

Tboiling = 463K

(lower limit for

Jet-A)

Fuel: nC12



Fuel: 64.38% nC12 / 35.62 m-Xylene



Fuel: m-Xylene



Tboiling = 533K

(upper limit for

Jet-A, e.g., JP-5)

19

Sustained Kernels at 1400J/gm

Blend, 1400 J/gm, 463 K boilnC12, 1400 J/gm, 463 K boil

Blend, 1400 J/gm, 533 K boil

nc12, 1400 J/gm, 533 K boil

n-Dodecane n-Dodecane/

m-xylene

463K

533K

20

„Statistical‟ Analysis (arbitrary and unvalidated)

Statistics suggest dodecane will light more easily than the JP-8

surrogate (blend) and that lower boiling point fuels will light

slightly easier

Fuel

E/M (J/gm) (463 K) (533 K) (463 K) (533 K) (463 K) (533 K)

1300 0.0% - 0.0% - - -

1400 55.4% 54.1% 39.6% 38.6% - -

1500 62.9% 61.3% 61.2% 59.6% 0.0% 0.0%

1750 63.8% 62.3% 63.1% 61.5% 59.9% 58.0%

(trial size: 840 cases)

n-Dodecane Blend Fuel m-Xylene

21

Challenge Problem: HC Emissions from Gas turbine Engines

Empirical evidence that emissions scale linearly – independent of engine or power

• What physical/chemical processes control this scaling?

• Can impact of alternative fuels on atmospheric pollution be anticipated?

Results from

APEX 2 and 3

22

Approach

Identify chemistry and physics controlling scalability of HC emissions

Methodology:

• Utilize reactor-network formulation to simulate flowfield of burner

• Fuel used: 77% dodecane, 23% m-xylene (vol %) – SERDP mechanism

• Initially assumed “bulk flow” conditions (e.g., total flow rates, total burner

volume, etc.)

• Modified to focus on conditions with measurable emission levels and

those exhibiting scaling

• Approximate “Idle”/low-power conditions:

P3 = 4.08 atm,

T3 = 478 K

overall ~ 0.25

23

Network Construction for Refocused Analysis

Based on fluid streaks

Front-End Residence time: .2-1.4 msec

Phi: 1.6-2.2

Quench Region Residence time: ~0.05-0.005 msec

Exit Phi: 0.1-0.35

Burn-out Residence time: na

Exit Phi: 0.1-0.35

Fuel

Primary Air

Secondary/Dilution Air (evenly added over 20 reactors)

Relaxed constraints relative to burner dimensions

(Emission levels expected to be high)

24

Results of Streak Analysis

0

20

40

60

80

100

120

140

160

180

200

0 20 40 60 80

HC

Exi

t EI

[gm

/kg/

fue

l]

CH2O Exit EI [gm/kg fuel]

CH4 EI

C2H4 EI

CH3CHO EI

C3H6 EI

C4H6 EI

C6H6 EI

m = 3.31

m = 0.58

m = 0.36

m = 0.12 m = 0.03

‘Correlations’ enable estimating HC sensitivity to CH2O EI values

25

Results of Streak Analysis (for two surrogates)

Reasonable agreement for methane, ethene, propene.

Butadiene, benzene, acetaldehyde not as well predicted

Linearity not always perfect

Uncertainties in scaling factors needs determination

• particularly large for propene

n-Dodecane/m-Xylene n-Dodecane/MCH Spicer (JP-5/CFM-56/idle) Aerodyne (AAFEX/JP-8)

CH4 0.36 0.29 0.43 -

C2H4 3.31 4.68 2.69 1.33

C3H6 0.58 0.55 0.79 0.51

C6H6 0.12 0.00 0.32 0.16

C4H6 0.03 0.10 0.30 -

CH3CHO 0.07 0.09 0.47 0.36

26

Summary of Simulation Conditions that Yielded Trend Data

One scenario: rapid quenching of rich pockets to very lean conditions

Critical conditions:

• Fuel rich phi: 1.6 to 2.2

• Quench residence time – 2.2 to 12 microseconds!

• Final phi - 0.1 to 0.2

Typical quench times:

In traditional RQL burners > 0.1 milliseconds!

Possible mechanisms discussed at ESS/CI – October (Storrs, CT)

27

Simulation of PM emissions from RQL combustor

Full gas-phase and soot kinetics with aerosol

dynamics model coupled into perfectly stirred

reactor (PSR) network model

0

0.2

0.4

0.6

0.8

1

1.2

1.4

0.38 0.4 0.42 0.44 0.46 0.48 0.5

Soo

t m

ass

((m

g/m

^3

)

Equivalence ratio

FT (100% n-dodecane)

JP8 (77%n-dodecane/23%m-xylene)

1SNETPSR model

Fuel-rich

front endQuench

ZoneLean, Burn-

Out Zone

Reduced-order model developed and

installed into LES codes for predicting soot

at combustor exit plane

Mass fraction

Size (cm)

Number density

(cc-1)

Reduced PM emissions with FT fuels Predictions of mass, size, and number

28

Summary

Coupling of chemistry amongst fuel molecules may be

needed for simulation of extinction/LBO

Physical effects are likely as/more important as

chemical effects

Kinetic models need to be accurate for phi variations

Fuel rich chemistry (especially for aromatics) may

need additional studies

Preliminary coupled ignition model coupling

kinetics/mixing/vaporization explains qualitative

observations

HC emissions fingerprint likely due to rapid quenching

of reacting gas element

28

29

Recommendations/Needs – 2011

Validated mechanisms that predict ignition and extinction

particularly aromatics and cycloalkanes

Inclusion of ignition criteria for certification of alternative fuels

What should be the standard?

Incorporate physical effects?

Soot models that predict fuel effects – (isoalkanes)

Statistical methods for simulation of relative (physical/chemical) effects

may be needed

Review rich side chemistry, especially for aromatics

30

Thank You!!!!!

Acknowledgements:

Office of Naval Research – Dr. Sharon Beermann-Curtin

Strategic Environmental Research and Development Program

Air Force Research Laboratories – Dr. Mel Roquemore and Dr.

Tim Edwards

United Technologies Research Center – Heidi Hollick

Pratt and Whitney – Dr. Jeff Lovett

Documents

Use of Kinetic Models to Predict Fuel Impact on Engine ...€¦ · 1) Kinetics model (ignition surrogate) selected from Westbrook, et al 2) Compared to surrogates and kinetics proposed