Simulations of Viscous Detonations with

Detailed Kinetics Using Manifold and

Wavelet Techniques
by
Joseph M. Powers
Associate Professor
Department of Aerospace and Mechanical Engineering
University of Notre Dame
presented at
Los Alamos National Laboratory
15 July 1999
Support
National Science Foundation
Chemical and Transport Systems, CTS-9705150,
Air Force Oce of Scientic Research,
Directorate of Mathematics and Geosciences
AFOSR Grant No. F49620-98-1-0206,
and
Los Alamos National Laboratory
Acknowledgments
Prof. Samuel Paolucci, ND-AME
Dr. Christopher Bowman, Post-Doc, ND-AME,
Mr. Sandeep Singh, Ph.D. Candidate, ND-AME
Mr. Yevgenii Rastigejev, Ph.D. Candidate, ND-AME
Outline
Motivation
Goals
Description of ILDM technique
Summary of wavelet technique
Detailed results for H
2
O
2
detonation
Strategy for HMX combustion and preliminary results
Summary
Motivation
Detailed nite rate kinetics critical in reactive uid mechanics:
Candle ames,
Atmospheric chemistry,
Internal combustion engines,
Gas phase reactions in energetic solid combustion.
Common detailed kinetic models are computationally expensive.
150 hr supercomputer time for calculation of steady, laminar,
axisymmetric, methane-air diusion ame (Smooke)
Expense increases with
number of species and reactions modeled (linear eect),
stinessratio of slow to fast time scales, (geometric eect).
Fluid mechanics time scales: 10
5
s to 10
1
s.
Reaction time scales: 10
14
s to 10
2
s.
Reduced kinetics necessary given current computational resources.
Adaptive discretization necessary for ne spatial structures.
Inclusion of physical diusion necessary for numerical conver-
gence.
Why Diusion?
Diusion traditionally not modelled in detonation studies,
Argued that very thin shock structures, thickness = O(m), will
have minimal inuence on reaction events,
However, inviscid solutions to two-dimensional reactive Euler equa-
tions in mildly unstable regimes do not appear to converge, while
viscous counterparts do (Singh, Powers, Paolucci, AIAA-99-0966,
1999),
Hypothesis: inherent numerical diusion is selecting structures in
inviscid calculations; these evolve unphysically with grid size,
When physical diusion zones are resolved numerically, grid-
independent physical diusion dominates over numerical diu-
sion.
Prohibitively expensive to compute simultaneous viscous and re-
action zone structures with common numerical techniques and
actual physical parametric values.
SPP modelled systems with reaction length/diusion length 10
to achieve resolved results; much larger ratios necessary to model
real systems.
Goals
Implement robust new reduced kinetic method of
Maas, U., and Pope, S. B., 1992, Simplifying Chemical Ki-
netics: Intrinsic Low-Dimensional Manifolds in Composition
Space, Combust. Flame, 88: 239-264.
Lam, S. H., 1993, Using CSP to Understand Complex Chem-
ical Kinetics, Combust. Sci. Tech., 89: 375-404.
Extend method to systems with time and space dependency.
Extend method to systems in which uid and chemical phenom-
ena evolve over similar time scales.
Couple method with new wavelet collocation technique (Paolucci
& Vasilyev) for spatial discretization.
Applications:
ignition delay in shock tubes; detailed results,
unstable viscous detonations,
Bunsen burner ames,
rocket nozzle ows,
HMX gas phase reactions; preliminary manifolds.
Common Reduced Kinetics Strategies
Fully frozen limit: no reaction allowed, uninteresting
Fully equilibrated limit: commonly used in some problems
has value for events in which uid time scales are slow with
respect to reaction time scales,
misses events which happen on chemical time scales.
Simple one and two step models
require signicant intuition and curve tting,
can give good rst order results,
are often not robust.
Partial equilibrium and steady-state assumptions
again require intuition,
are not robust.
Sensitivity analysis
can remove need to include unimportant reactions,
not guaranteed to remove stiness.
Intrinsic Low-Dimensional Manifold Method (ILDM)
Uses a dynamical systems approach,
Does not require imposition of ad hoc partial equilibrium or
steady state assumptions,
Fast time scale phenomena are systematically equilibrated,
Slow time scale phenomena are resolved in time,
n-species gives rise to a n-dimensional phase space (same as com-
position space) for isochoric, isothermal combustion in well stirred
reactors,
Identies m-dimensional subspaces (manifolds), m < n, embed-
ded within the n-dimensional phase space on which slow time
scale events evolve,
Fast time scale events rapidly move to the manifold,
Slow time scale events move on the manifold.
Computation time reduced by factor of 10 for non-trivial com-
bustion problems; manifold gives much better roadmap to nd
solution relative to general implicit solution techniques (Norris,
1998)
Simplest Example
dx
dt
= 10x, x(0) = x
o
,
dy
dt
= y, y(0) = y
o
.
Stable equilibrium at (x, y) = (0,0); stiness ratio = 10.
ILDM is x = 0
-1 -0.5 0.5 1
x
-7.5
-5
-2.5
2.5
5
7.5
y
Parameterization of manifold: x(s) = 0; y(s) = s.
dy
dt
=
dy
ds
ds
dt
, chain rule
y(s) =
dy
ds
ds
dt
, substitute from ODE and manifold
s = (1)
ds
dt
, no longer sti!
s = s
o
e
t
,
x(t) = 0; y(t) = s
o
e
t
.
Projection onto manifold for s
o
, induces small phase error.
Formulation of General Manifolds
A well stirred chemically reactive system is modeled by a set of
non-linear ordinary dierential equations:
dx
dt
= F(x), x(0) = x
o
,
x : species concentration; x
n
Equilibrium points dened by
x = x
eq
such that F(x
eq
) = 0.
Consider a system near equilibrium (the argument can and must
be extended for systems away from equilibrium) with x = xx
eq
.
Linearization gives
d x
dt
= F
x
x,
where F
x
is a constant Jacobian matrix.
Schur decompose the Jacobian matrix:
F
x
= Q U Q
T
Q =
_
_
_
_
_
.
.
.
.
.
.
.
.
.
q
1
q
2
q
n
.
.
.
.
.
.
.
.
.
_
_
_
_
_
, U =
_
_
_
_
_
_
_
_

1
u
12
u
1n
0
2
u
2n
0
.
.
.
.
.
.
0 0
n
_
_
_
_
_
_
_
_
, Q
T
=
_
_
_
_
_
_
_
_
q
T
1

q
T
2

.
.
.
q
T
n

_
_
_
_
_
_
_
_
Formulation of General Manifolds (cont.)
Qis an orthogonal matrix with real Schur vectors q
i
in its columns.
U is an upper triangular matrix with eigenvalues of F
x
on its
diagonal, sometimes placed in order of decreasing magnitude.
The Schur vectors q
i
form an orthonormal basis which spans the
phase space,
n
.
We then dene m slow time scales, m n.
Next dene a non-square matrix W which has in its rows the
Schur vectors associated with the fast time scales:
W =
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
q
T
m+1

q
T
m+2

.
.
.
q
T
n

_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
.
Letting the fast time scale events equilibrate denes the manifold:
W F(x) = 0.
If m = 0, no slow time scales, W = Q
T
, and W F(x) = 0
implies Q
T
F(x) = 0, implies F(x) = 0: the equilibrium point
is the low dimensional manifold!
A Simple Example
Consider
dx
dt
= 100x + y siny, x(0) = x
o
,
dy
dt
= x
3
y, y(0) = y
o
.
Equilibrium points:
F =
_
_
_
0
0
_
_
_ =
_
_
_
100x + y siny
x
3
y
_
_
_,
_
_
_
x
y
_
_
_ =
_
_
_
0
0
_
_
_.
Other equilibrium points exist!
Near the equilibrium point (0,0), linearization gives
_
_
_
dx
dt
dy
dt
_
_
_ =
_
_
_
100 0
0 1
_
_
_
_
_
_
x
y
_
_
_,
which is obviously stable.
Schur decomposition is trivial:
F
x
= Q U Q
T
_
_
_
100 0
0 1
_
_
_ =
_
_
_
1 0
0 1
_
_
_
_
_
_
100 0
0 1
_
_
_
_
_
_
1 0
0 1
_
_
_
Form the manifold:
W = ( 1 0 ) ,
W F(x) = ( 1 0 )
_
_
_
100x + y siny
x
3
y
_
_
_ = 0,
100x + y sin y = 0 The ILDM!
-0.40 -0.30 -0.20 -0.10 0.00 0.10 0.20 0.30 0.40
-20.0
-10.0
0.0
10.0
20.0
Phase Plot
solution
trajectories
equilibrium
point
x
y
ILDM : WF(x)=0
0.0 2.0 4.0 6.0
-0.20
0.00
0.20
0.0 2.0 4.0 6.0
0.0
5.0
10.0
15.0
20.0
t (s) t (s)
x
y
A Simple Example
dx
dt
= -100x + ysin(y) x(0) = x
dy
dt
= x - y y(0) = y
3
O
O
Time Variation of X and Y
Simple Example: Parameterization and Stiness Reduction
dx
dt
= 100x + y sin y,
dy
dt
= x
3
y.
Time scales near origin:
1
= 1.0,
2
= 0.01. Sti.
First approximation to manifold is x =
1
100
y siny.
Parameterize manifold as
x =
1
100
s sin s,
y = s.
Chain rule gives
dy
dt
=
dy
ds
ds
dt
.
Substitute from ODEs and parameterization:
x
3
(s) y(s) =
dy(s)
ds
ds
dt
,
1
10
6
s
3
sin
3
s s = (1)
ds
dt
,
ds
dt
=
1
10
6
s
3
sin
3
s s
Linearize near equilibrium at origin:
ds
dt
= s.
Time scale: = 1.0 No longer sti!
Solve ODE for s(t), substitute to get x(s(t)), y(s(t)):
x
1
100
s
o
e
t
sin
_
s
o
e
t
_
, y s
o
e
t
.
Example: Zeldovich Mechanism of NO Formation
Mechanism (two elements, ve species, two reactions):
1. O+N
2
NO+N, k
1
= 1.810
14 cm
3
mol s
exp
_
38370 K
T
_
,
2. NO +N O +N
2
, k
2
= 3.8 10
13 cm
3
mol s
exp
_
425 K
T
_
,
3. N+O
2
NO+O, k
3
= 1.810
10 cm
3
mol s K
T exp
_
4680 K
T
_
,
4. NO+O N+O
2
, k
4
= 3.810
9 cm
3
mol s K
T exp
_
20820 K
T
_
,
Take T = 1400 K, then
1. k
1
= 2.252 10
2 cm
3
mol s
2. k
2
= 2.805 10
13 cm
3
mol s
3. k
3
= 8.905 10
11 cm
3
mol s
4. k
4
= 1.851 10
6 cm
3
mol s
Law of mass action for [N
2
], for example, gives
d[N
2
]
dt
= k
1
[N
2
][O] + k
2
[NO][N].
For all species, law of mass action yields ve non-linear ODEs:
d
dt
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
[N]
[NO]
[N
2
]
[O]
[O
2
]
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
=
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
1 1 1 1
1 1 1 1
1 1 0 0
1 1 1 1
0 0 1 1
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
k
1
[N
2
][O]
k
2
[N][NO]
k
3
[N][O
2
]
k
4
[NO][O]
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
Example: Zeldovich Mechanism of NO Formation, cont.
To elucidate naturally conserved variables, use elementary row
operations to cast system in non-unique row echelon form:
d
dt
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
[N]
[NO] [N]
2[N
2
] + [NO] + [N]
[O] + [N]
2[O
2
] + [NO] [N]
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
=
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
1 1 1 1
0 0 2 2
0 0 0 0
0 0 0 0
0 0 0 0
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
k
1
[N
2
][O]
k
2
[N][NO]
k
3
[N][O
2
]
k
4
[NO][O]
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
We are left with
two ODEs
three algebraic constraints: conservation of N atoms, O atoms,
and number of molecules
easily reduced to two ODEs in two unknowns: [N], [NO].
We will reduce the two ODEs to one ODE by imposing the man-
ifold equation W F(x) = 0, eectively equilibrating the fast
time scale.
Example: Zeldovich Mechanism of NO Formation, cont.
Consider rst the intrinsic algebraic constraints:
d
dt
_
_
_
_
_
_
_
2[N
2
] + [NO] + [N]
[O] + [N]
2[O
2
] + [NO] [N]
_
_
_
_
_
_
_
=
_
_
_
_
_
_
_
0
0
0
_
_
_
_
_
_
_
.
Integrate these equations:
2[N
2
] + [NO] + [N] = C
1
,
[O] + [N] = C
2
,
2[O
2
] + [NO] [N] = C
3
.
The constants C
1
, C
2
, C
3
come from initial conditions.
Solve equations for secondary variables in terms of [N], [NO]:
[N
2
] =
1
2
(C
1
[NO] [N])
[O] = C
2
[N]
[O
2
] =
1
2
(C
3
[NO] + [N])
Note that rearrangement of the algebraic constraints demonstrates ele-
ment and molecule conservation:
2[N
2
] + [N] + [NO] = C
1
,
2[O
2
] + [NO] + [O] = C
2
+ C
3
,
[N] + [NO] + [N
2
] + [O] + [O
2
] =
C
1
+ C
3
2
+ C
2
.
Example: Zeldovich Mechanism of NO Formation, cont.
Substitution of algebraic constraints into ODEs for [N] and [NO] gives
two autonomous ODEs well-suited for dynamic systems analysis:
d[N]
dt
=
k
1
2
(C
2
[N]) (C
1
[N] [NO])
k
2
[N][NO]

k
3
2
[N] (C
3
+ [N] [NO])
+k
4
[NO] (C
2
[N])
d[NO]
dt
=
k
1
2
(C
2
[N]) (C
1
[N] [NO])
k
2
[N][NO]
+
k
3
2
[N] (C
3
+ [N] [NO])
k
4
[NO] (C
2
[N])
Take as initial conditions
[N] = [NO] = [N
2
] = [O] = [O
2
] = 0.001
mole
cm
3
.
Equilibrium when right hand side zero
Three roots-one physical, two unphysical:
_
_
_
[N]
[NO]
_
_
_ =
_
_
_
1.16 10
11 mole
cm
3
2.78 10
6 mole
cm
3
_
_
_,
_
_
_
1.15 10
11 mole
cm
3
2.78 10
6 mole
cm
3
_
_
_,
_
_
_
2.00 10
3 mole
cm
3
0.00 10
0 mole
cm
3
_
_
_,
Example: Zeldovich Mechanism of NO Formation, cont.
Linearization of equations near physical equilibrium gives
d
dt
_
_
_
[N] 1.16 10
11
[NO] 2.78 10
6
_
_
_ =
_
_
_
9.67 10
8
3.38 10
3
8.11 10
8
4.03 10
3
_
_
_ F
x
_
_
_
[N] 1.16 10
11
[NO] 2.78 10
6
_
_
_
d
dt
_
_
_
[N] 1.16 10
11
[NO] 2.78 10
6
_
_
_ =
_
_
_
0.766 0.643
0.643 0.766
_
_
_ Q
_
_
_
9.67 10
8
3.38 10
3
0 1.19 10
3
_
_
_ U
_
_
_
0.766 0.643
0.643 0.766
_
_
_ Q
T
_
_
_
[N] 1.16 10
11
[NO] 2.78 10
6
_
_
_
Condition number (stiness ratio) =

9.6710
8
1.1910
3

= 8.1 10
5
.
Locally the ILDM is dened by
W F(x) = 0,
( 0.766 0.643)
_
_
_
F
1
([N], [NO])
F
2
([N], [NO])
_
_
_ = 0.
Use arc length continuation methods to dene complete ILDM
The physical equilibrium has negative eigenvalues: stable.
The non-physical equilibria have positive eigenvalues: unstable.
10
12
10
10
10
8
10
6
10
4
10
2
10
0
10
11
10
10
10
9
10
8
10
7
10
6
10
5
10
4
10
3
t (s)
[
N
]
,

[
N
O
]

m
o
l
e
/
c
m
3
10
12
10
10
10
8
10
6
10
4
10
2
10
6
10
5
10
4
10
3
[N] mole/cm
3
[
N
O
]

m
o
l
e
/
c
m
3
attracting manifold
equilibrium
point
solution trajectory
[NO]
[N]
.
Adaptive Multilevel Wavelet Collocation Technique
Summary of standard spatial discretization techniques
Finite dierence-good spatial localization, poor spectral local-
ization, and slow convergence,
Finite element- good spatial localization, poor spectral local-
ization, and slow convergence,
Spectralgood spectral localization, poor spatial localization,
but fast convergence.
Wavelet technique
See e.g. Vasilyev and Paolucci, A Fast Adaptive Wavelet Col-
location Algorithm for Multidimensional PDEs, J. Comp.
Phys., 1997,
Basis functions have compact support,
Good spatial localization, good spectral localization, and fast
convergence,
Easily formulated to adapt spatially to capture steep gradients
via adding collocation points,
Spatial adaptation is automatically and dynamically adaptive
to achieve prescribed error tolerance.
Ignition Delay in Premixed H
2
-O
2
Consider standard problem of Fedkiw, Merriman, and Osher, J.
Comp. Phys., 1996,
Shock tube with premixed H
2
, O
2
, and Ar in 2/1/7 molar ratio,
Initial inert shock propagating in tube,
Reaction commences shortly after reection o end wall,
Detonation soon develops,
Model assumptions
One-dimensional,
No diusion (one case); mass, momentum, and energy diu-
sion (another case),
Nine species, thirty-seven reactions,
Ideal gases with variable specic heats.
Compressible Reactive Navier-Stokes Equations for H
2
-O
2
Problem

t
+

x
(u) = 0, mass

t
(u) +

x
_
u
2
+ P
_
= 0, momentum

t
_
_

_
_
e +
u
2
2
_
_
_
_
+

x
_
_
u
_
_
e +
u
2
2
_
_
+ u (P ) + q
_
_
= 0, energy

t
(Y
i
)+

x
(uY
i
+ j
i
) =
M

j=1
a
j
T

j
exp
_
E
j
T
_

ij
M
i
N

k=1
_
Y
k
M
k
_

kj
, species
P = T
N

i=1
Y
i
M
i
, thermal equation of state
e =
N

i=1
Y
i
_
h
o
i
+
_
T
T
o
c
pi
(

T)d

T
_

, caloric equation of state

=
4
3

u
x
, Newtonian gas with Stokes assumption
j
i
=
N

j=1
D
ij
Y
j
x
, Ficks law
q = k
T
x
+
N

i=1
j
i
_
h
o
i
+
_
T
T
o
c
pi
(

T)d

T
_
augmented Fouriers law.
N = 9 species: H
2
, O
2
, H, O, OH, H
2
O
2
, H
2
O, HO
2
, Ar
M = 37 reactions
Operator Splitting Technique
Equations are of form

t
q(x, t) +

x
f(q(x, t)) = g(q(x, t)).
where
q =
_
_
, u,
_
_
e +
u
2
2
_
_
, Y
i
_
_
T
f models convection and diusion
g models reaction source terms
Splitting
1. Inert convection diusion step:

t
q(x, t) +

x
f(q(x, t)) = 0,
d
dt
q
i
(t) =
x
f(q
i
(t)).

x
is either Godunov or wavelet discretization operator.
2. Reaction source term step:

t
q(x, t) = g(q(x, t)),
d
dt
q
i
(t) = g(q
i
(t)).
Operator splitting with implicit sti source solution can induce non-
physical wave speeds! (LeVeque and Yee, JCP 1990)
ILDM Implementation in Operator Splitting
Form of equations in source term step:
d
dt
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_

_
e +
u
2
2
_
Y
i
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
=
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
0
0
0

_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
.
Equations reduce to
=
o
, u = u
o
, e = e
o
,
dY
i
dt
=

o
.
has dependency on , e, and Y
i
ODEs for Y
i
can be attacked with manifold methods when man-
ifold with , e, H and O parameterization is available.
In premixed problem, H and O element concentrations are re-
markably constant, reducing the dimension by two!
Full equations integrated until suciently close to manifold
Once on manifold, simple projection used to return to manifold
following convection-diusion step
Sample ILDM for H
2
O
2
Projection of ILDM in H
2
O, H
2
O
2
plane,
Adiabatic (e = 525 kJ/kg), isochoric ( = 0.25 kg/m
3
), element
concentrations of H and O constant,
Complete manifold tabulated in three dimensions: , e, Y
H
2
O
,
So we have e.g. P (, e, Y
H
2
O
) , T (, e, Y
H
2
O
) , Y
H
(, e, Y
H
2
O
) , . . .
Linear interpolation used for points not in table,
Captures 0.1 s reaction events.
0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1 0.11 0.12
0
1
2
x 10
5
H2O mass fraction
H
2
O
2

m
a
s
s

f
r
a
c
t
i
o
n
ILDM
Inviscid H
2
O
2
Ignition Delay with and without ILDM
No diusion,
Godunov spatial discretization, 400 uniform nite dierence cells,
Implicit (trapezoidal) convection step; Implicit (dlsode) or ILDM
reaction step,
Correction of Fedkiw adopted to suppress articial entropy layer
after shock reection (see Meniko, 1994).
0 0.02 0.04 0.06 0.08 0.1
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
x (m)
D
e
n
s
i
t
y

(
K
g
/
m
3
)
0 0.02 0.04 0.06 0.08 0.1
600
400
200
0
200
400
600
x (m)
V
e
l
o
c
i
t
y

(
m
/
s
)
0 0.02 0.04 0.06 0.08 0.1
0
1
2
3
4
5
x 10
5
x (m)
P
r
e
s
s
u
r
e

(
P
a
)
0 0.02 0.04 0.06 0.08 0.1
500
1000
1500
2000
2500
3000
x (m)
T
e
m
p
e
r
a
t
u
r
e

(
K
)
Inviscid H
2
O
2
Ignition Delay with and without ILDM
0 0.02 0.04 0.06 0.08 0.1
2
4
6
8
10
12
14
x 10
3
H2
0 0.02 0.04 0.06 0.08 0.1
0
0.002
0.004
0.006
0.008
0.01
0.012
0.014
O
0 0.02 0.04 0.06 0.08 0.1
0
0.02
0.04
0.06
0.08
H2O
0 0.02 0.04 0.06 0.08 0.1
0
0.02
0.04
0.06
0.08
0.1
0.12
O2
Viscous H
2
O
2
Ignition Delay with Wavelets
Mass, momentum, and energy diusion modelled,
Wavelet spatial discretization, explicit convection-diusion time step-
ping, implicit reaction time stepping,
300 collocation points, 15 wavelet levels,
Viscous shocks, induction zones, and entropy layers spatially resolved!
t = 180 s.
0 0.02 0.04 0.06 0.08 0.1
500
1000
1500
2000
2500
3000
x (m)
T
e
m
p
e
r
a
t
u
r
e

(
K
)
0 0.02 0.04 0.06 0.08 0.1
600
400
200
0
200
400
600
x (m)
V
e
l
o
c
i
t
y

(
m
/
s
)
0 0.02 0.04 0.06 0.08 0.1
0
1
2
3
4
5
x 10
5
x (m)
P
r
e
s
s
u
r
e

(
P
a
)
0 0.02 0.04 0.06 0.08 0.1
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
x (m)
D
e
n
s
i
t
y

(
k
g
/
m
3
)
Viscous H
2
O
2
Ignition Delay with Wavelets
t = 190 s
0 0.02 0.04 0.06 0.08 0.1
500
1000
1500
2000
2500
3000
x (m)
T
e
m
p
e
r
a
t
u
r
e

(
K
)
0 0.02 0.04 0.06 0.08 0.1
600
400
200
0
200
400
600
x (m)
V
e
l
o
c
i
t
y

(
m
/
s
)
0 0.02 0.04 0.06 0.08 0.1
0
1
2
3
4
5
x 10
5
x (m)
P
r
e
s
s
u
r
e

(
P
a
)
0 0.02 0.04 0.06 0.08 0.1
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
x (m)
D
e
n
s
i
t
y

(
k
g
/
m
3
)
Viscous H
2
O
2
Ignition Delay with Wavelets
t = 200 s
0 0.02 0.04 0.06 0.08 0.1
500
1000
1500
2000
2500
3000
x (m)
T
e
m
p
e
r
a
t
u
r
e

(
K
)
0 0.02 0.04 0.06 0.08 0.1
800
600
400
200
0
200
400
600
x (m)
V
e
l
o
c
i
t
y

(
m
/
s
)
0 0.02 0.04 0.06 0.08 0.1
0
1
2
3
4
5
x 10
5
x (m)
P
r
e
s
s
u
r
e

(
P
a
)
0 0.02 0.04 0.06 0.08 0.1
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
x (m)
D
e
n
s
i
t
y

(
k
g
/
m
3
)
Viscous H
2
O
2
Ignition Delay with Wavelets
t = 230 s
0 0.02 0.04 0.06 0.08 0.1
500
1000
1500
2000
2500
3000
x (m)
T
e
m
p
e
r
a
t
u
r
e

(
K
)
0 0.02 0.04 0.06 0.08 0.1
600
400
200
0
200
400
600
x (m)
V
e
l
o
c
i
t
y

(
m
/
s
)
0 0.02 0.04 0.06 0.08 0.1
0
0.5
1
1.5
2
2.5
3
3.5
x 10
5
x (m)
P
r
e
s
s
u
r
e

(
P
a
)
0 0.02 0.04 0.06 0.08 0.1
0.1
0.2
0.3
0.4
0.5
0.6
0.7
x (m)
D
e
n
s
i
t
y

(
k
g
/
m
3
)
Comparison with Inviscid/ILDM Result at Same Time
t = 230 s
0 0.02 0.04 0.06 0.08 0.1
0.1
0.2
0.3
0.4
0.5
0.6
0.7
x (m)
D
e
n
s
i
t
y

(
K
g
/
m
3
)
0 0.02 0.04 0.06 0.08 0.1
600
400
200
0
200
400
600
x (m)
V
e
l
o
c
i
t
y

(
m
/
s
)
0 0.02 0.04 0.06 0.08 0.1
0
0.5
1
1.5
2
2.5
3
3.5
x 10
5
x (m)
P
r
e
s
s
u
r
e

(
P
a
)
0 0.02 0.04 0.06 0.08 0.1
500
1000
1500
2000
2500
3000
x (m)
T
e
m
p
e
r
a
t
u
r
e

(
K
)
Viscous H
2
O
2
Ignition Delay with Wavelets
t = 180 s
species mass fractions plotted vs. distance
0 0.05 0.1
0
0.5
1
1.5
2
2.5
x 10
3
H
0 0.05 0.1
0
0.005
0.01
0.015
O
0 0.05 0.1
0
0.005
0.01
0.015
H2
0 0.05 0.1
0
0.05
0.1
0.15
O2
0 0.05 0.1
0
0.005
0.01
0.015
OH
0 0.05 0.1
0
0.05
0.1
H2O
0 0.05 0.1
0
2
4
6
8
x 10
5
HO2
m
0 0.05 0.1
0
1
2
3
x 10
5
H2O2
m
0 0.05 0.1
0
0.5
1
Ar
m
Viscous H
2
O
2
Ignition Delay with Wavelets
t = 190 s
species mass fractions plotted vs. distance
0 0.05 0.1
0
0.5
1
1.5
2
2.5
x 10
3
H
0 0.05 0.1
0
0.005
0.01
0.015
O
0 0.05 0.1
0
0.005
0.01
0.015
H2
0 0.05 0.1
0
0.05
0.1
0.15
O2
0 0.05 0.1
0
0.005
0.01
0.015
OH
0 0.05 0.1
0
0.05
0.1
H2O
0 0.05 0.1
0
2
4
6
8
x 10
5
HO2
m
0 0.05 0.1
0
1
2
3
x 10
5
H2O2
m
0 0.05 0.1
0
0.5
1
Ar
m
Viscous H
2
O
2
Ignition Delay with Wavelets
t = 200 s
species mass fractions plotted vs. distance
0 0.05 0.1
0
0.5
1
1.5
2
2.5
x 10
3
H
0 0.05 0.1
0
0.005
0.01
0.015
O
0 0.05 0.1
0
0.005
0.01
0.015
H2
0 0.05 0.1
0
0.05
0.1
0.15
O2
0 0.05 0.1
0
0.005
0.01
0.015
OH
0 0.05 0.1
0
0.05
0.1
H2O
0 0.05 0.1
0
2
4
6
8
x 10
5
HO2
m
0 0.05 0.1
0
1
2
x 10
5
H2O2
m
0 0.05 0.1
0
0.5
1
Ar
m
Viscous H
2
O
2
Ignition Delay with Wavelets
t = 230 s
species mass fractions plotted vs. distance
0 0.05 0.1
0
0.5
1
1.5
2
2.5
x 10
3
H
0 0.05 0.1
0
0.005
0.01
0.015
O
0 0.05 0.1
0
0.005
0.01
0.015
H2
0 0.05 0.1
0
0.05
0.1
0.15
O2
0 0.05 0.1
0
0.005
0.01
0.015
OH
0 0.05 0.1
0
0.05
0.1
H2O
0 0.05 0.1
0
2
4
6
8
x 10
5
HO2
m
0 0.05 0.1
0
1
2
x 10
5
H2O2
m
0 0.05 0.1
0
0.5
1
Ar
m
Post Reection Entropy Layer?: Viscous Wavelet Results
No signicant entropy layer evident on macroscale after shock
reection when resolved viscous terms considered,
Inviscid codes with coarse gridding introduce a larger entropy
layer due to numerical diusion,
Unless suppressed, unphysically accelerates reaction rate.
0.115 0.116 0.117 0.118 0.119 0.12
500
1000
1500
x (m)
T
e
m
p
e
r
a
t
u
r
e

(
K
)
0.115 0.116 0.117 0.118 0.119 0.12
0
100
200
300
400
500
600
x (m)
V
e
l
o
c
i
t
y

(
m
/
s
)
0.115 0.116 0.117 0.118 0.119 0.12
0
0.5
1
1.5
2
x 10
5
x (m)
P
r
e
s
s
u
r
e

(
P
a
)
0.115 0.116 0.117 0.118 0.119 0.12
0.1
0.2
0.3
0.4
0.5
0.6
x (m)
D
e
n
s
i
t
y

(
k
g
/
m
3
)
Post Reection Entropy Layer: Viscous Wavelet Results
small entropy layer evident on ner scale,
temperature rise 5 K; dissipates quickly,
inviscid calculations before adjustment give persistent tempera-
ture rise of 20 K; reaction acceleration small.
0.115 0.116 0.117 0.118 0.119 0.12
1185
1190
1195
1200
x (m)
T
e
m
p
e
r
a
t
u
r
e

(
K
)
0.115 0.116 0.117 0.118 0.119 0.12
6
4
2
0
2
x (m)
V
e
l
o
c
i
t
y

(
m
/
s
)
0.115 0.116 0.117 0.118 0.119 0.12
1.18
1.185
1.19
1.195
1.2
x 10
5
x (m)
P
r
e
s
s
u
r
e

(
P
a
)
0.115 0.116 0.117 0.118 0.119 0.12
0.372
0.374
0.376
0.378
0.38
x (m)
D
e
n
s
i
t
y

(
k
g
/
m
3
)
Viscous H
2
O
2
Ignition Delay with Wavelets
Close-up: Viscous Shock Stucture and Induction Zone
t = 230 s,
Induction zone length: 470 m,
No signicant reaction in viscous shock zone.
0.028 0.0282 0.0284 0.0286 0.0288
500
1000
1500
2000
2500
3000
x (m)
T
e
m
p
e
r
a
t
u
r
e

(
K
)
0.028 0.0282 0.0284 0.0286 0.0288
600
400
200
0
200
400
600
x (m)
V
e
l
o
c
i
t
y

(
m
/
s
)
0.028 0.0282 0.0284 0.0286 0.0288
0
0.5
1
1.5
2
2.5
3
3.5
x 10
5
x (m)
P
r
e
s
s
u
r
e

(
P
a
)
0.028 0.0282 0.0284 0.0286 0.0288
0.1
0.2
0.3
0.4
0.5
0.6
0.7
x (m)
D
e
n
s
i
t
y

(
k
g
/
m
3
)
Viscous H
2
O
2
Ignition Delay with Wavelets
Closer-up: Viscous Shock Stucture Only
t = 230 s
predicted shock thickness: 50 m.
2.788 2.79 2.792 2.794
500
1000
1500
2000
2500
3000
x (cm)
T
e
m
p
e
r
a
t
u
r
e

(
K
)
2.788 2.79 2.792 2.794
600
400
200
0
200
400
600
x (cm)
V
e
l
o
c
i
t
y

(
m
/
s
)
2.788 2.79 2.792 2.794
0
0.5
1
1.5
2
2.5
3
3.5
x 10
5
x (cm)
P
r
e
s
s
u
r
e

(
P
a
)
2.788 2.79 2.792 2.794
0.1
0.2
0.3
0.4
0.5
0.6
0.7
x (cm)
D
e
n
s
i
t
y

(
k
g
/
m
3
)
Viscous H
2
O
2
Ignition Delay with Wavelets,
Instantaneous Distributions of Collocation Points
t = 180 s, two-shock structure with consequent collocation
point distribution,
t = 230 s, one-shock structure with evolved collocation point
distribution.
0 0.02 0.04 0.06 0.08 0.1 0.12
0
5
10
15
x (m)
L
e
v
e
l
0 0.02 0.04 0.06 0.08 0.1 0.12
0
5
10
15
x (m)
L
e
v
e
l
Application to Gas Phase HMX System
Simulating isobaric HMX combustion computationally intensive,
Most eort in solving gas phase convection, reaction, diusion,
Based on 45 species, 232 step mechanism of Yetter, et al.,
Fastest time scales predicted 10
16
s (non-physical?),
Stiness ratio 10
11
(vs. 10
9
for H
2
O
2
),
Equations for gas phase combustion of HMX are of form

t
q(x, t) +

x
f(q(x, t)) = g(q(x, t)),
Adiabatic, isobaric,
Operator splitting appropriate,
For non-premixed problem, higher dimension ( 8?!) manifolds
necessary!
Will need to parameterize by (h, , H, O, N, C, Ar, one free parameter)
10
7
< h < 10
11
erg/g; 10
5
< < 10
3
g/cm
3
; 10
32
<
Ar
< 10
2
;
0 <
C
< 10
1
; 0 <
H
< 10
1
; 0 <
N
< 10
1
; 0 <
O
< 10
1
.
(Liau, 1999)
Three-dimensional manifold for preliminary premixed problem?
ILDM for Gas Phase HMX System
Based on 45 species, 232 step mechanism of Yetter, et al.,
Adiabatic (h = 62 10
9
erg/g) , isobaric (P = 32 bar),
projection in Y
N
2
, Y
CO
2
plane.
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5
0
0.01
0.02
0.03
0.04
0.05
0.06
0.07
0.08
0.09
N
2
(mass fraction)
C
O
2

(
m
a
s
s

f
r
a
c
t
i
o
n
)
Summary
Robust method in place to compute manifolds with arbitrary
variables held constant (e.g. P, , h),
Eort still needed on improving technique of projecting onto man-
ifold initially,
Fast linear interpolation scheme in place for table lookup,
Robust method in place to solve less sti dierential equations
on or near manifold,
Operator splitting allows implementation of manifold in solving
PDEs,
Adaptive multilevel wavelet collocation method gives dramatic
spatial resolution,
Full coupling of ILDM and wavelet methods soon forthcoming,
Detailed studies of eciency improvement necessary,
More general manifold techniques need developed to allow strong
uid-chemistry coupling and relaxation of eigenmodes to steady
state solutions.