Title: Combustion mechanisms in detonation fronts Joe Shepherd, Caltech October 2004
1Combustion mechanisms in detonation frontsJoe
Shepherd, Caltech October 2004
- Progress report on work by
- Experiments Joanna Austin (UIUC), Florian
Pintgen (CIT) - Steady flame simulations Dan Lieberman (CIT),
Sandeep Singh (Detroit Diesel) - Transient flame simulations Marco Arienti
(UTRC), Ralf Deiterding (CIT), Patrick Hung (CIT) - Transient detonation simulations (Alexi Khokhlov,
U Chicago)
2Detonation Tube Experiments
- 8 m long, 280 mm diameter
- cookie-cutter attached to the 150 mm square
test section - Initiation with an exploding wire and an
acetylene-oxygen-driver
End plate soot foil
3Cellular Structure
R. Akbar 1997
4Themes
- Some experimental observations
- Speculations about combustion mechanisms
- The role of diffusive processes
- The classical theory of high-speed flames
- Transient processes at shear layers
- The turbulent flame
- Future studies
5Observations
- We observe a range of reaction front features
which are strongly correlated with the
sensitivity of the reaction rates to temperature
variations.
6OH fluorescence images - irregular structure
5 mm
8 mm
Shot 1597, 2H2O2 8 N2, 20kPa
Shot 1604, 2H2O2 8 N2, 20kPa
2H2O2 9 N2, 20kPa, Image height 50mm
All 150x150mm testsection
Reference F. Pintgen, J. M. Austin, and J. E.
Shepherd, Detonation front structure Variety and
characterization, Confined Detonations and Pulse
Detonation Engines, pages 105-116. Torus Press,
Moscow, 2003.
7Highly Irregular MixturesN2O H2 mixtures
N2-diluted
10 mm
10 mm
10 mm
Shot 1607
Shot 1608
Shot 1609
H2 N2O 3 N2, 20 kPa, Cellsize 70mm
All 150x150mm testsection
Reference F. Pintgen, J. M. Austin, and J. E.
Shepherd, Detonation front structure Variety and
characterization, Confined Detonations and Pulse
Detonation Engines, pages 105-116. Torus Press,
Moscow, 2003.
8Highly Irregular MixturesN2O H2 mixtures
N2-diluted
Shot 1643, H2 N2O 2 N2, 20kPa, Cell size 42mm
50 mm
50 mm
50 mm
20 mm
Shot 1609, H2 N2O 3 N2
Shot 1656, H2 N2O 2 N2
Shot 1643, close up
9Instability driven turbulence
highly unstable (high Ea/RTS)
f1
unstable
stable
weakly unstable (low Ea/RTS)
10What are we learning?
- Front structure not universal
- Strong function of chemical system
- Characterize with reaction zone structure
- Activation energy
- Shape of reaction zone ti/tr
- Correlated with longitudinal instability
threshold - Range of scales increases with increasing
activation energy - For fuel-air mixtures
- 102-103 range of length scales
- High frequency temporal fluctuations
- Decoupling at end of reaction zone
11Explosive events following decoupling
4mm
12Turbulent detonations?
- Reaction zone appears to be dominated by
fluctuations in leading shock velocity - Consequence of highly unstable structure
- Leads to large fluctuations in species and energy
release rate - Transverse waves
- Consequence or cause of spatial variation of
leading shock? - May lead to explosive events following decoupling
- Diffusive processes may operate at very small
scales but - Smallest scales not resolved in experiments
- Suggests that turbulent detonations may be
intrinsically different than turbulent
deflagrations
Is this a new or an old regime of turbulent
combustion?
13Why?
- 1. Why is the reaction zone turbulent?
- Conjecture The mean flow is highly unstable to
small perturbations. Numerous modes of transverse
and longitudinal instability are present
simultaneously. Nonlinear interactions of these
modes creates a highly fluctuating environment
within the reaction zone.
14What?
- 2. What is the nature of the turbulence?
- At least two possibilities
- Coherent quasi-periodic keystone structures
created by hierarchy of transverse waves - Incoherent random superposition of fluctuations
in front velocity creates random fluctuations
in OH thermicity. Transverse waves are acoustic
mode of turbulence.
15How?
- 3. How is this different from classical
turbulence? - Speculations
- Fluctuations in temperature much more significant
than fluctuations in velocity. - Three dimensional shock geometry more significant
than vorticity distribution.
16Possible Consequences
- 4. How does turbulence influence detonation
propagation? - Unknown but we speculate that
- Perturbations to reaction rate
- Increases mean reaction rate
- Decreases sensitivity to mean fluctuations
- Harder to quench detonation by diffraction or
energy absorption
17Numerical Simulations
- Two-dimensional inviscid simulations with
simplified reaction mechanism (1-step) but highly
refined spatial discretization (A. Khokhlovs FTT
method) - Ea/RT chosen to match regular and irregular
mixture thermochemical computations.
18Regular Instability, Ea/RT 6.5
Numerical
Experimental
19Irregular instability, Ea/RT 10
Experimental
Numerical
20Conclusions
- Two-dimensional simulations at CJ condition
feasible even with simplified reaction mechanism. - Cellular substructure, wrinkling of reaction
front, localized explosions consistent with
experimental observations. - Purely chemical-gasdynamic explanation for
inviscid turbulence
21However
- Computations are based on Euler model no
diffusive processes. Is this what happens in
nature? - What happens at even higher values of Ea/RT?
- Most probable situations for diffusive transport
- Thin layers of unreacted material surrounded by
products - Shear layers between warm (shocked) reactants and
hot products. - Could conventional diffusive turbulent combustion
be possible?
22Regimes for Conventional Diffusive Turbulent
Combustion
from N. Peters (2000) Turbulent Combustion
23Can we put some points on this plot?
- Are there regimes where laminar flames can exist
in detonation fronts? - What is the characteristic values of laminar
flame speed in post-shock conditions? - What are the range of length scales and
fluctuation velocities in turbulent detonations?
24Classical analysis of diffusive transport in
high-speed combustion
- Numerical study of Clarkes one-dimensional fast
flames - Re-examining conclusions reached with asymptotics
for realistic detailed chemistry. - Examine key issue not addressed by Clarke When
do diffusion flames cease to be well defined?
25One-Dimensional Fast Flame
26Postshock Conditions
Stoichiometric CH4-air mixtures at various shock
speeds
27Governing Equations
28(No Transcript)
29(No Transcript)
30Two Extremes
CJ-ZND
Adiabatic flame
31Magnitude of the terms in the energy balance
equation
v2m/s
v5.81m/s
tconv
tcond
tdiff
treac
tpress
tsum
v20m/s
v100m/s
32Fast flame study conclusions
33Existence of flames behind shocks
34(No Transcript)
35Results of flame computations
36Summary of fast flame study
37Characterization of fronts
Coastlines
Distribution of lengths
38Scale-dependent fractal?
39Wrinkled laminar flamelet analysis
M r u A
M r SL AT
AT/A U/SL
40Conclusions
- Identified potential relevance of
diffusion-controlled mechanisms in irregular
detonation cellular structure - Highly unstable reactive mixtures provide
sufficient time-scale separation between thotcv
and twarmcv in the decaying part of the
detonation cell cycle. - Molecular transport and thermal conduction can
accelerate chemical kinetics in the time-scale of
interest
41Transient ignition of flames at shear layers
- Motivated by OH PLIF observations and numerical
simulations. - Analog of classic Marble-Adamson problem for
ignition behind flame holder
Warm reactants
Hot products
42The Marble-Adamson problem
- Chemical changes occur by a single reaction step
- Boundary layer scaling ?
- Diffusive term is of the same order as convective
terms? ?
?
- Schvab-Zeldovich formulation
- Steady, low speed (isobaric) flow
- Binary diffusion coefficients of all species
equal - Unit Lewis number
43The Marble-Adamson problem
propagating laminar flame
cold combustibles ? u1
hot inerts ? u2
boundary layer
- But
- Warm not cold combustible stream close to
reaction. - Hot stream is not inert (role of radicals
diffusion vs. thermal diffusion).
44Time scale competition
- Two streams
- hot, with temperature Th and adiabatic (c.v.)
induction time thotcv - warm, with temperature Tw and adiabatic (c.v.)
induction time twarmcv - Time to ignite a homogeneous mixture
- Time for a diffusive flame to developwhere
- Diffusive flame at shear layer if
- ? Look for large separation of thotcv and twarmcv
45Local triple point analysis
- Polar analysis for non-reactive gas.
- Flow is steady in the reference frame of the
triple point. - Constant track angle f from incident leading
shock. - Known shock speed D at the cell centerline.
- All waves are straight.
46Comparison of adiabatic ignition times
2H2-O2-7Ar P16.67kPa Miller Bowman
C2H4-3O2-8N2 P120kPa GRI-Mech 3.0
47Homogeneous mixing ignition at shear layer
idealizedcell
48Dependence from chemical mechanism
- 0-dimensional calculations at constant pressure
usingthe CHEMKIN library Kee et al., 1989.
0.21 O2-0.79 N2
0.25 H2-0.75 N2
1100K
300K
Z kg
(1-Z) kg
1 atm
Mixture fraction Z
tHMI(Z)
49Time scales competition
C2H4-3O2-8N2 P120kPa GRI-Mech 3.0
50Flame development 1-D unsteady code
- AMROC provides access to adaptive data management
as well as to detailed reaction mechanisms (via
CHEMKIN). - Operator splitting approach added subroutine to
compute diffusive fluxes
- Species molecular diffusion, mixture thermal
conductivity and thermal diffusion coefficients
from TRANSPORT software package. - Mixture averages using CHEMKIN gas-phase
utilities.
51Test 1 H2-O2 premixed laminar flame
- Comparison with steady solution from PREMIX
program.
2H2-O2-7Ar mixture at P11 atm
H2-O2 mechanism by Miller Bowman (1989)
52Test 2 H2-O2 non-premixed laminar flame
0.25 H2-0.75 N2 (300K) / 0.21 O2-0.79 N2 (1100K)
at P11 atm H2-O2 mechanism by
Kreutz Law (1996)
53Weakly and strongly unstable reactive mixtures
2H2-O2-8N2
2H2-O2-7Ar
Composition at 10 peak temperature in hot-stream
c.v. explosion. H2-O2 mechanism by Miller
Bowman (1989)
54Example I Diffusion flame at shear layer
2H2-O2-8N2
Composition hot stream at equilibrium ? or inert
? at 2213 K warm stream
unreacted at 1020 K ? twarmcv 1470 ms
? Estimated tHMI(0.95) 17. ms
55Example II Convective explosion at shear layer
2H2-O2-8N2
T at 23.6 ms
XOH at 23.6 ms
Composition hot stream at equilibrium ? or
inert ? at 2213 K warm
stream reactants at 1200 K ? twarmcv 19.8 ms
? Estimated tHMI(0.95) 8. ms
56Example III Energy vs. radical diffusion effects
2H2-O2-11N2
Composition hot stream at equilibrium ? or inert
? at 1427 K warm stream
reactants at 1000 K ? twarmcv 327 ms
? Estimated tHMI(0.95) 162 ms