Chapter 9: Flame wall interactions in canonical configurations
Transcript of Chapter 9: Flame wall interactions in canonical configurations
Andreas Dreizler | 1
Chapter 9: Flame wall interactions in
canonical configurations
A. Dreizler
Sponsored by Deutsche ForschungsgemeinschaftCenter of Smart InterfacesTechnische Universität Darmstadt
TU Darmstadt, GermanyDept. of Mechanical Engineering Institute for Reactive Flows and Diagnostics
Andreas Dreizler | 2
Literature
• Kosaka, H., Zentgraf, F., Scholtissek, A., Bischoff, L., Häber, T., Suntz, R., Albert, B., Hasse, C.,
Dreizler, A.: Wall heat fluxes and CO formation/oxidation during laminar and turbulent side-wall
quench-ing of methane and DME flames. International J. of Heat and Fluid Flow (2019) 70:181
– 192.
• Jainski, C., Rißmann, M., Jakirlic, S., Böhm, B., Dreizler, A.: Quenching of premixed flames at
cold walls: Effects on the local flow field. Flow Turbulence Combustion 100, 177 – 196 (2018).
• Jainski, C., Rißmann, M., Böhm, B., Janicka, J., Dreizler, A.: Sidewall quenching of atmospheric
laminar premixed flames studied by laser-based diagnostics. Combustion and Flame 183, 271
– 282 (2017).
• Rißmann, M., Jainski, C., Mann, M., Dreizler, A.: Flame-flow interaction in premixed turbulent
flames during transient head-on quenching. Flow Turbulence Combustion 98, 1025 – 1038
(2017).
• Jainski, C., Rißmann, M., Böhm, B., Dreizler, A.: Experimental investigation of flame surface
density and mean reaction rate during flame–wall interaction. Proceedings of the Combustion
Institute 36, 1827 – 1834 (2017). 1
• Bohlin, A., Jainski, C., Patterson, B.D., Dreizler, A., Kliewer, C.J.: Multiparameter spatio-
thermochemical probing of flame–wall interactions advanced with coherent Raman imaging.
Proceedings of the Combus-tion Institute 36, 4557 – 4564 (2017).
Andreas Dreizler | 3
Flame-wall interaction (FWI)
• Topical subject with relevance for
• Safety technology: flame arresters
• Catalytically assisted combustion
• Micro-combustion
• Gas turbine combustion (lean, low NOx)
• Internal combustion engines (cold start, downsizing)
10 %
Pilot,
90 %
Main
Chemilum. image; exposure
time 0.06s
fglobal = 0.7; p = 2 barTUD combustor
Davy’s mine lamp
British Museum (1817)
J. Hermann et al., TU Darmstadt
Pressurized combustor
Interaction of flame-flow-effusion cooling
Andreas Dreizler | 4
Flame-Wall Interaction
• Example: Spark ignition engines
Burned gas : 1500~2500 K
Wall surface: 350~700 K
Burned gas
Wall
Flame
• Large heat loss to wall
• Flame quenching = termination of
chemical reactions
• Sources of UHC and CO
→ Complex problem with high
practical relevance
Andreas Dreizler | 5
General properties of FWI
• When flame approaches closer than a few flame thicknesses to the wall
(for premixed combustion)
→ Intense coupling between flame and wall
• Large heat fluxes (exceeding 1 MW/m2 for HC-fuels at 1 bar)
• Flame quenching causing incomplete combustion
• Emission of unburned hydrocarbons and CO
• Relevant especially for engine combustion
• Contribution to heat loss ~30%
• Contribution to unburned hydrocarbons ~40%
Andreas Dreizler | 6
Dynamics of HOQ: 1D-simulationSource: T. Meier, G. Künne; A. Keteheun, J. Janicka (Darmstadt)
Sp
atia
l te
mp
.
pro
file
CO
2-s
ou
rce
te
rmN
orm
. wa
ll
dis
tan
ce
as fc
t. of tim
e
Andreas Dreizler | 7
FWI for turbulent conditions
Source: T. Poinsot and D. Veynante, Theoretical and Numerical Combustion 2005
Andreas Dreizler | 8
FWI for turbulent conditions
Source: T. Poinsot and D. Veynante, Theoretical and Numerical Combustion 2005
Studied by DNS in simplified configurations and mostly simplified
chemical kinetics
Very limited number of comprehensive experimental studies going beyond
quenching distances and heat transfer
→ Requirement for more detailed insights into practically relevant
configurations
Andreas Dreizler | 9
Parameters of interest to better understand FWI
• Quenching distances, visualization of flames near walls
• Wall temperature and heat flux
• Flow fields near walls
• Velocity boundary layers u
• Thermo-chemical states during FWI
• Thermal boundary layers T
• Concentration boundary layers X
• Local heat release rates near walls
Q
wa
ll
Q
Quenching distance
Focus today
Andreas Dreizler | 10
Outline
Motivation
Multi-dimensional laser diagnostics for studying flame-wall interactions
Side-wall quenching of flames
Flame-wall interaction in IC engines
Summary
Andreas Dreizler | 11
Burner setup – Side-wall Quenching
40
mm
water/oil
• Premixed V-flame (fuel: methane, DME)
• Φ = 0.83, 1, 1.2
• Re = 5000
• Laminar and turbulent (by turb. grid)
• Temperature controlled wall
water/oilFuel+Air
SWQ-wall
Ceramic rod
Turbulence grid
Andreas Dreizler | 12
High-speed PIV/OH-PLIF: Experimental setup
• Two fields of view (18x18 mm2)
• 2C-PIV (AL2O3 particles)
• OH-PLIF
• High-speed dye laser system (35 µJ/pulse)
• Q1(6)-line
• Canny-edge filter for flame front detection
Gas velocity Flame front position
Particle Image Velocimetry (PIV)Planar LIF of OH radical
Flame front from OH gradient
2 D/2 C
Rep. Rate: 10 kHz and 10 Hz
2 D
Rep. Rate: 10 kHz and 10 Hz
Andreas Dreizler | 13
Visualization of flow field and flames near walls
Quenching
distance
Quenching
distance
Laminar Turbulent
• Extraction of flow features: Boundary layers, turbulence fields, … (not
shown)
Andreas Dreizler | 14
CARS/ CO-LIF/ OH-PLIF/ Phosphor Thermometry
Experimental setup
Gas Temperature CO Concentration Wall Temperature Flame front position
ro-vibrational
ns-CARS
Two Photon LIF of
CO molecule
Phosphor
thermometryPlanar OH-LIF
0 D
Rep. Rate: 10 Hz
0 D
Rep. Rate: 10 Hz
2 D
Rep. Rate: 10 Hz
2 D
Rep. Rate: 10 Hz
Front view Side view
Andreas Dreizler | 15
Gas phase and wall surface temperature (Re = 5000, Φ = 1.0, laminar)
Wa
ll
100 µm
Methane DME
Twall = 330 K
Twall = 450 KTwall = 540 K
Twall = 670 K
• DME flames: quench further upstream than methane flames due to increased laminar burning
velocity
• Higher Twall → increased Tgas within boundary layer
→ flame burns further upstream due to increased laminar burning velocity
Andreas Dreizler | 16
Wall-heat flux(Re = 5000, Φ = 1.0, laminar)
Methane DME
Twall = 330 K
Twall = 450 K
Twall = 540 K
Twall = 670 KWa
ll
100 µm
𝑞 = 𝜆𝑇𝑔𝑎𝑠 − 𝑇𝑤𝑎𝑙𝑙
Δ𝑦
Andreas Dreizler | 17
Quenching distance and wall-heat flux (Re = 5000, Φ = 1.0, laminar)
Wa
ll
Post - FWI
With higher Twall
• Maximum heat flux increases due to decreasing quenching distances
• Heat flux in post-flame region decreases as expected for non-reacting flows
Quenching position
Andreas Dreizler | 18
Temperature & wall-normal CO-profiles (Re = 5000, Φ = 1.0, laminar)
Wa
ll
pre-FWIFWI
post-FWI
Andreas Dreizler | 19
CO-temperature correlations: State space
Wa
ll
Spatial conditioning
(axial & wall-normal)
Flame tip fluctuates up
and down (± 150 µm)
→ Different thermo-kinetic
states at one
measurement location
± 150 µm
z = 48 mm, laminar
Upstream quenching position,
various y-positions
Andreas Dreizler | 20
Thermo-chemical states for z = 49.5 mm
@ quenching position (Re = 5000, Φ = 1.0, laminar)
Exp. Adiabatic Non-adiabatic
• CO formation branch: strongly influenced for y < 0.3 mm for methane flame
• CO consumption branch: shifted to lower temperatures for entire near-wall region with both
fuel types
Andreas Dreizler | 21
Time scale analysis(Re = 5000, Φ = 1.0, laminar)
Methane DME
y [mm] y [mm]
t H[m
s]
Time scales of
CO formation
Time scales of
CO oxidation
Time scales of
heat transfer
Time scale of heat transfer1
𝜏𝐻 =𝐿2
𝜆𝜌𝑐𝑝, 𝐿 represents the wall-normal
distance y
heat conductivity1 E. Marín, Characteristic dimensions for heat transfer, Latin-
American Journal of Physics Education 4 (2010) 56–60
Andreas Dreizler | 22
Time scale analysis(Re = 5000, Φ = 1.0, laminar)
Methane DME
y [mm] y [mm]
t H[m
s]
Time scales of
CO formation
Time scales of
CO oxidation
Time scales of
heat transfer
𝜏𝐻 =𝐿2
𝜆𝜌𝑐𝑝, 𝐿 represents the wall-normal
distance y
heat conductivity
• CO-formation: slower than heat transfer for y ≤ 0.2 mm (methane),
y < 0.1 mm (DME)
→ Only for methane at y ≤ 0.2 mm influence of wall heat loss
• CO-oxidation: both fuels are influenced by heat loss in near-wall region
Andreas Dreizler | 23
Thermo-chemical states for z = 42.5 mm(Re = 5000, Φ = 1.0, turbulent)
Exp. Adiabatic Non-adiabatic
• Both branch are influenced for entire near-wall region
• Intermediate states between both branches are observed
→ Increased wall heat transfer due to turbulence
Andreas Dreizler | 24
Heat release imaging – Scope
Measurement of relative heat release
rate (HRR) for premixed flames
interacting with cold walls
Investigation of flame structures during
flame-wall interactions
Use of correlation between HRR and
XOH x XCH2O
Andreas Dreizler | 25
Correlation between HRR and
Correlation between the product of and HRR derived from premixed
flame calculations for stoichiometric methane/air and DME/air flames
Curvature (horizontal axis), strain (area along the line) and the case of Le = 1
(no variation of strain) are considered
Andreas Dreizler | 26
Experimental setup for heat release imaging
Andreas Dreizler | 27
Results – laminar flame (1)
Ensemble average,
1000 samples
Andreas Dreizler | 28
Results – laminar flame (2)
Wall-normal profiles at quenching height
Andreas Dreizler | 29
Results – turbulent flame (1)
Single realization
Andreas Dreizler | 30
Results – turbulent flame (2)
Left: Wall-normal
profiles of curvature
(averaged) vs. wall-
normal distance
Right: Normalized
profiles of the HRR
against curvature.
HRR values are
normalized by the
value at curvature =
0.
Andreas Dreizler | 31
Results – turbulent flame (3)
Correlation analysis
Curvature close to wall approaches to zero due to laminarization
Mean positive curvature (convex flame front towards unburnt gases) in the
FWI region
Mean negative curvature for flames approaching the wall (y = 1.5 – 4 mm)
Increase in HRR particularly with negative curvature observed for both fuels,
this is in accordance with one-dimensional flame calculations using a detailed
transport model
→ Indication that Lewis number effects are important for turbulent premixed
flames interacting with walls
Andreas Dreizler | 32
Outline
Motivation
Multi-dimensional laser diagnostics for studying flame-wall interactions
Side-wall quenching of flames
Flame-wall interaction in IC engines
Summary
Andreas Dreizler | 33
PTV
camera PLIF
Kamera
Flow field Flame position
Temperaturmessung
Thermographische
Phosphorthermometrie (TPT)
Acusto optical deflector
Planar laser-induced
fluorescence (PLIF) of SO2
Inert tracer
High temperature sensitivity
A B C D
Ignition ( °CA ) -14.2 -22.2 -22.2 -27.2
imep ( bar ) 5.7 1.9 6.2 2.1
200 cycles, λ = 1
Experimental setup
Andreas Dreizler | 34
Physical Processes in IC Engines
• Wall-resolvedParticleTracking
Velocimetry• Significant
deviations
from log-law
Renaud et int. Dreizler, Böhm.
International Journal of Heat and
Fluid Flow 71, 366-377
Andreas Dreizler | 35
Flame propagation
Flame frontProbability „burnt“
200 cycles
B
800 1/min
0.40 bar
A
800 1/min
0.95 bar
C
1500 1/min
0.95 bar
D
1500 1/min
0.40 bar
burnt
unburnt
Ding et int. Dreizler, Böhm. Proc. Combust Inst (2019) 37 (4):4973-4981
Andreas Dreizler | 36
Flame-flow-interaction
• 15°CA: similar flow structures for A-C
(confirmed by bulk flow characteristics)
• Case A: Strong acceleration of flow
• Case D: Deceleration of flow
A B C D
Different flame-flow
interactions dependent
on operation condition
t
i
m
e
Andreas Dreizler | 37
Boundary layer, cases A – D
× 𝛿50 = 𝑦piston(0.5 ⋅ 𝑈𝑥,𝑚𝑎𝑥)
Strong changes for A
Correlation between flow and
combustion?
Conditional
statistics
170 µm 180 µm 100 µm 140 µm
80 µm 110 µm 80 µm 130 µm
4.3 m/s
r = 0.61
Ux:
Closely above the piston
-9°CA
Strong / weak flow
Andreas Dreizler | 38
Conditional statistics, case A
Starke Strömung
Schwache
Strömung𝛿50 ≈ 170 µm 𝛿50 ≈ 110 µm𝛿50 ≈ 80 µm
× 𝛿50 = 𝑦Kolben(0.5 ⋅ 𝑈x,max)
Similar development over °CA
Differences in gradients
𝛿50 ≈ 140 µm
AS: Strong - Early acceleration
AW: Weak - Later acceleration
AM: Motored - Reduced velocity
magnitudes
Andreas Dreizler | 39
Conditional statistics, case A
𝑈𝑥50/𝛿50 ( m/s / mm )
𝜃 ( °KW ) A AW AS AM
-15 -10 -7.6 -13 -12
-9 -23 -15 -33 -11
-7.8 -35 -25 -46 -11
∙ 3,5
𝛿50 ≈ 170 µm 𝛿50 ≈ 110 µm𝛿50 ≈ 80 µm
𝛿50 ≈ 140 µmConclusions
Boundary layer flows strongly
influenced by turbulent flame (flow
magnitudes, gradients)
Boundary layer flows strongly
correlated with combustion
performance (PMI)
Variation with operating conditions
Open issue: Scaling law
Andreas Dreizler | 40
Outline
Motivation
Multi-dimensional laser diagnostics for studying flame-wall interactions
Side-wall quenching of flames
Flame-wall interaction in IC engines
Summary
Andreas Dreizler | 41
Summary
Chemically reacting flows such as combustion processes are multi-
dimensional in nature
Chemical and physical process acting at similar time-scales cause a
strong mutual interaction
Multi-parameter diagnostics are mandatory to disclose reaction-
transport coupling
Advanced laser diagnostics are unrivalled, for example
Spatial resolution of 20 µm in IC engine
Simultaneous measurement of wall & gas temperatures, heat flux, CO-
concentration and flame front position in generic FWI-burner
Andreas Dreizler | 42
Thank you for your kind attention