Effects of Exhaust Gas Recirculation (EGR) onTurbulent Combustion Emissions in Advanced Gas Turbine

Combustors with High Hydrogen Content (HHC) Fuels

Jay P. Gore and Robert P. Lucht, Purdue University

Maurice J. Zucrow LaboratoriesSchool of Mechanical Engineering

Purdue UniversityWest Lafayette, IN

DOE Award No. DE-FE0011822

National Energy Technology LaboratoryUniversity Turbine Systems Research Program

Project Review Meeting November 1-2, 20171

Collaborations

Yiguang Ju and Michael Mueller – PrincetonGaurav Kumar and Scott Drennan – Convergent Sci. Inc. NewBraunfels, TexasJeff Moder – NASA Glenn Research CenterRelated Sponsors – FAA, ONR, Rolls Royce, Siemens, GE

PhD students

Dong Han – CARS and PLIFHasti Veeraraghava Raju - CFD simulationsJupyoung Kim – PIV

Post Doctoral Associate : Aman Satija

DOE Program Manager: Mark Freeman

Acknowledgments

2

Content1. Piloted Axisymmetric Reactor Assisted Turbulent (PARAT) burner

development and testing under atmospheric and high-pressure conditions

2. Effects of CO2 addition on turbulent flame structure and burning velocity

3. Temperature and velocity measurements in CH4 /air/CO2 flames with different levels of CO2 addition using CARS and PIV

4. Development and validation of LES model for H2 piloted CH4/air/CO2 premixed turbulent flames

5. CH PLIF and IR imaging for turbulent premixed flames3

Experimental Apparatus: PARAT Burner

4

Lower plate

Upper plate

Flames with varying levels of CO2 addition

10% CO2, Φ=0.89 5% CO2, Φ=0.84 0% CO2, Φ=0.80

Re=10,000, Tad=2030 K, Le=1, P=1 bar

Flames designed to minimize thermal and transport effects on NOx

5

Large Eddy Reacting Flow Simulations

• Premixing tube simulated separately and the solutions patched

• Jet Reynolds number – 10000 • Domain (D= 18 mm): 36D x 64D x 36 D• Detailed chemistry solver with DRM19 mech.

Turbulence – 1 eq. dynamic structure model• Sensitivity study with base grid : 10 x 8 x 6 mm • 4 Level Adaptive Mesh Refinement based on

Velocity and Temperature, Max. 15 M cells• Mesh sensitivity studied with Max. 30 M cells

Burner surface

CH4+Air (Premixed)

Pilot H2Pilot H2

Y

X

Z

• DRM19 Mechanism: (http://combustion.berkeley.edu/drm/)• Elements : O, H, C, N, AR• Species: H2, H, O, O2, OH, H2O, HO2, CH2, CH2(S), CH3, CH4,

CO, CO2, HCO, CH2O, CH3O , C2H4, C2H5, C2H6, N2, AR• Number of Reactions: 84

Z=0 PlaneMesh

Chemistry

CFD Summary

7

Large Eddy Simulations

Instantaneous Temperature [K]

Mean Temperature [K]

Computational Grid

Region: 5 < y/d < 6

8

Inlet Boundary Conditions LES Comparison with Experiments

Non reacting flow simulation for jet velocity profiles at the burner exit

Flame 1 velocity contour

Boundary Condition & Turbulence Intensity at x/D=0.2

Mean & RMS velocity profiles

Integral length scale &turbulence intensity

9

Turbulence intensity. . rms

mean

uT Iu

=

Integral length scale

0( ) ( , *) *l r r r drρ

∞= ∫ 2

( ) ( )( , ) ; | * |( )

x x

x

u r u r rr r r r ru r

ρ′ ′ + ∆

∆ = ∆ = −′

10

LES Mean Temperature Contours on Z=0 Plane

Cent

erlin

e

MeanTemperature (K)

11

Axial Temperature Profiles with CO2 Addition

0% CO2 5% CO2 10% CO2

12

Cent

erlin

e

LES RMS Temperature on Z=0 plane

RMSTemperature (K)

13

0% CO2 5% CO2 10% CO2

T_RMS comparison between CARS and LESCenterline RMS temperature

14

Temperature comparison between CARS and LES

MeanTemperature (K)

CARS measurements

15

At y/d =5

At y/d =1.94

0% CO2 10% CO2

At y/d =5

At y/d =1.94

Temperature comparison between CARS and LESRadial mean and RMS temperature

Thin or Purely Wrinkled Flame Assumption is not adequate!

CO2levels

Characteristic flame

thickness, λ(μm)

Kolmogorov length scale

η, (μm)

0% 70 50

5% 80 52

10% 90 55

0u

L

DS

λ =1/43νη

ε

=

16

RMS max mean mean minT = (T -T )(T -T )

Effect of CO2 on the chemistry of turbulent flames with EGR are captured in the present computations!

17

OH PLIF video for flame with 0% CO2 addition

Hydrogen pilot flame

Φ=0.8, Re=10000

18

OH PLIF video for flame with 5% CO2 addition

Hydrogen pilot flame

Φ=0.84, Re=10000

19

OH PLIF video for flame with 10% CO2 addition

Hydrogen pilot flame

Φ=0.89, Re=10000

20

OH PLIF Images & Data Processing

0% CO2 5% CO2 10% CO2Products𝒄𝒄= 𝟏𝟏

Reactants𝒄𝒄= 𝟎𝟎 1

( )1( )

ni

i i

L cn A c=

Σ = ∑Flame surface density

21

Mean Reaction Progress

( )1/2

1/2,

'2 ' 1 1 exp'T rl uu l

u lτδ α τ

τ = − − −

Axial direction

Radial direction

Flame brush development & Taylor’s theory

1

1( , ) ( , )n

ii

c x r c x rn =

= ∑Mean progress variable

1, maxT r

dcdr

δ − = Data

Fit

22

Flame Surface Density

1

( )1( )

ni

i i

L cn A c=

Σ = ∑Mean flame surface density

0 < x/D < 3.6 3.6 <x/D<6.5Radial flame brush development Axial flame brush development

23

Global Consumption Speed

�𝒄𝒄 = 𝟎𝟎.𝟐𝟐𝒄𝒄𝒄𝒄𝒄𝒄𝒄𝒄𝒄𝒄𝒄𝒄𝒄𝒄

[ ]

2

, 21 2 ( 0.2) /T GC

USx c D

=+ =

24

Local Consumption Speed

w/o pockets

w/ pockets

• w/o pockets

• w/ pockets

• pockets

, 0 0T

T LC LL

AS S IA

=

, 0 0 maxT LC L TS S I δ= Σ

25

Fine-scale Unburned Pocket Consumption

Fine-scale pocket size Consumption speed

, 2up

T LCPe

AS

R tπ∆

=∆

upe

AR

π=

Fine-scale pocket: a pocket does not break up into smaller ones with flame-flame interaction

upA Unburned pocket area

t∆ Time step

26

CH PLIF: Wavelength & Signal Strength

Simulation using LIFBASE

CH

OH

27

CH PLIF video for flame with 0% CO2 addition

Hydrogen pilot flame

Φ=1, Re=10000

28

CH PLIF video for flame with 5% CO2 addition Φ=1, Re=10000

Hydrogen pilot flame

29

CH PLIF video for flame with 10% CO2 addition Φ=1, Re=10000

Hydrogen pilot flame

30

CH-OH PLIF video for flame with 0% CO2 addition

Hydrogen pilot flame

Φ=1, Re=10000

31

CH-OH PLIF video for flame with 5% CO2 addition Φ=1, Re=10000

Hydrogen pilot flame

32

CH-OH PLIF video for flame with 10% CO2 addition Φ=1, Re=10000

Hydrogen pilot flame

33

CH PLIF & Simultaneous CH and OH PLIF

0% CO2 5% CO2 10% CO2

(a) wrinkled flame front, (b) unburned reactant pocket.

Φ=1, Re=10000 Green: CH Layer Blue: OH Zone

Challenging for lean premixed flames with CO2 dilution due to low CH signal

34

IR imaging video for CH4/air flame

Φ=0.8, Re=10000

35

Instantaneous IR images

2.58±0.03 μm 2.77 ± 0.1 μm 4.38 ± 0.08 μmH2O H2O+CO2 CO2

0% CO2 10% CO2

Multiple bandpass filters with KC burner Varying CO2 with PARAT burner

CH4/air Φ=0.80 Re=10000 Re=10000

Φ=0.80 Φ=0.89

36

Time Averaged Radiation Model Validation

Turbulence radiation interaction (TRI) modeling:

Stochastic time and space series analysis (STASS)

2 2

1 1

( )

0(0) ( )bI I e d I e d dλ

λ λ λλ λ ττ τ τ

λ λ λ λ λ λλ λα λ α τ τ λ

∗− − −∗ ∗= +∫ ∫ ∫

37

Temperature Deconvolution

Computed temperature vs thin filament thermometry

Computed temperature vs CARS thermometry

38

Re=100,000, Φ=0.80

H2 Pilot, 6% HR

5 bar, Tinlet590 K

High Pressure PARAT Experiments and LES

39

Summary & Conclusions1. Developed a PARAT burner and demonstrated multiple diagnostic methods including PIV, CARS, OH/CH PLIF and IR imaging for turbulent premixed combustion applications.

2. Performed a comprehensive investigation of the non-thermal effects of CO2 addition on turbulent premixed combustion for the first time.

3. CO2 addition extends flame length, Modifies flame brush to be longer and thinner, alters local flame surface area, reduces burning velocities, and enhances pocket formation with negligible effects on pocket consumption speed

4. Developed LES simulation tool for CH4/air/CO2 flames and validated using temperature and velocity measurements

40

Flame # 1 2 3

Reynolds number (± 50) 10000

Adiabatic Temperature (± 50 K) 2030

Equivalence ratio (± 0.02) 0.80 0.84 0.89

CO2 % by total mass (± 0.1) 0.0 5.0 10.0

CH4 mass flow rate (± 2 mg/s) 111 110 109

Air mass flow rate (± 20 mg/s) 2440 2300 2150

CO2 mass flow rate (± 4 mg/s) 0.00 124 246

Pilot H2 mass flow rate (± 0.03 mg/s) 2.7

Pilot H2 heat release percent of total (%) 6

Lewis number 1

Laminar flame speed (cm/s) 34 30 25

Laminar flame thermal thickness (µm) 70 80 90

RMS turbulence fluctuation (m/s) 1.7

Integral length scale (mm) 1

Appendix Flame operating conditions

41

Cold flow Results with RANS based RNG k-Ԑ model

Good Agreement with Hot wire anemometer measurements

Appendix

42

Davidson et.al, IJHF, 27, 2006

Nozzle Exit: Specified mean velocity profile. Turbulent Fluctuations: Random Fourier Approach

Burner surface

CH4+Air (Premixed)

Pilot H2Pilot H2

Outflow Boundary

CFD Domain and BCs - Reacting

Premixing Tube Excluded to Reduce Computational Time

36D x 64D x 36D, D= 18 mm

Y

XZ

Appendix