Direct Numerical Simulation of Turbulent Hydrogen-Oxygen Reacting Flow

Yeong-Pei Tsai
a
and Chung-Chyi Chou
b
a
Department of Mechanical Engineering, Chung Hua University
a,b
Institute of Engineering Science, Chung Hua University
a
yptsai@chu.edu.tw,
b
dustin.c0410@msa.hinet.net
Abstract. The paper of this investigation is to develop to computing flow field properties generated
by supersonic mixing of two co-flowing fluid streams (H2/O2) with shock-induced combustion.
Direct numerical simulations (DNS) of this kind of turbulent, chemically reacting flow will provide
the detailed evolution of the turbulent velocities, temperature, chemical concentration of each
component of species, pressure distribution and location of shock wave. The fully coupled 2D
Navier-Stokes equations with finite rate chemistry are solved in conservation law form. There is no
subgrid turbulence model introduced. Since the time scales of the chemistry tend to be much
smaller than the time scale of fluid motion, the set of equations describing chemically reacting flow
is mathematically stiff. The calculations are done using the FLUENT version 6 commercial code.
The combustion used for the present study is that proposed by Moretti which consists of six
reacting species and eight reactions. A twostep combustion chemistry of oxygen and hydrogen is
also used for comparison purpose.
Keyword Navier-Stokes equations, Supersonic combustion, WENO scheme, Finite rate chemistry.
I. Introduction
The specific impulse of rocket engines is inherently
lower than that of air-breathing engines, there has been a
renewed interest in hypersonic air-breathing propulsion
and high speed aerothermodynamics during the last few
years. The interest is mainly associated with the
development of advanced space transportation systems.
To develop efficient supersonic combustion ramjet
(scramjet) engines, it is important to understand the
fundamental physics involved in the supersonic,
chemically reacting mixing layer. A shaped projectile
may be accelerated efficiently by means of shock
induced combustion modes. The exit velocity is up to 10
km/sec. However, this kind of experiment is very hard to
conduct. Since supersonic fuel / oxidizer mixing process
has many engineering applications as mentioned above,
numerical study in this topic becomes very important. In
addition to its engineering applications, physics of the
turbulent flow field of a supersonic mixing layer with
shock-induced combustion is also of fundamental
theoretical interest. Supersonic combustion involves
complex interactions between turbulence, shock waves
and combustion. Because of their capability to reproduce
unsteady effects, direct numerical simulations (DNS) is
attractive to model such supersonic reactive flows. A
computational study of shock-wave / boundary layer
interactions involving premixed combustible gases was
calculated by Shaye-Yungser. The analysis was carried
out using a fully-implicit, total variation diminishing
(TVD) code to solve the fully-coupled
Reynolds-Averaged Navier-Stokes equations. Rizzetla et
al. proposed a hybrid compact-Roe approach in order to
simulate a supersonic compression-ramp flow. In the
context of supersonic combustion, high temperature and
species gradients are expected in addition to shock
waves. In order to accurately predict the interactions
between these phenomena, the original methodology
proposed by Cook and Cabot is extended. An artifical
diffusivity based an high-order derivatives of the entropy
is introduced. The objective of this proposed research is
to develop and to validate a high-order, compact WENO
scheme adequate for resolving interactions between
shocks, turbulence and combustion.
The system of nonlinear equations describing reacting
flows are difficult to solve because they are
mathematically stiff. The stiffness is measured by the
Damkohler number, Da, where Da=tf/tch, and tf is the
characteristic time scale of fluid motion; tch is the
characteristic chemical reaction time. Generally, there is
a Damkohler number associated with each chemical
reaction. In the present research, consider the
shock-induced combustion of H2/O2 co-flow, the
Chung Hua Journal of Science and Engineering, Vol. 7, No. 2, pp. 9-14 (2009)
9
stiffness is mainly due to the chemistry. Thermally
perfect gas effects are incorporated into the algorithm in
the present study. The numerical algorithm and
combustion model are summarized as follows.
II. Numerical Formulation
2.1. Governing Equations
The Navier-Stokes equations for two-dimensional flow
are considered. For the case of chemically reacting flows,
the global continuity equation is replaced by all of the
species continuity equations. They can be expressed in
the following conservation form for a gas containing
n
species and in general curvilinear coordinates (,):
S
) G G ( ) F F (
t
Q
=
c
c
+
c
c
+
c
c
q
u u
where
Q =
(
(
(
(
(
(
(
(
(

e
u
J
n
2
1
1

; F=
(
(
(
(
(
(
(
(
(

+
+
+

) (
2
1
1
p e U
p uU
p uU
U
U
U
J
y
x
n

;
G=
(
(
(
(
(
(
(
(
(

+
+
+

) (
2
1
1
p e V
p uV
p uV
V
V
V
J
y
x
n
q
q

;
(
(
(
(
(
(
(
(
(

+
+
+

y y x x
yy y xy x
xy y xx x
d
n n
d
d
v
U
U
U
J
F
| |
t t
t t

2 2
1 1
1
Re
;
(
(
(
(
(
(
(
(
(

+
+
+

y y x x
yy y xy x
xy y xx x
d
n n
d
d
v
V
V
V
J
G
| q | q
t q t q
t q t q

2 2
1 1
1
Re
;
(
(
(
(
(
(
(
(
(

0
0
0
s
s
s
J S
n
2
1
1

and
x xy xx x
q u + = vt t |
,
y yy xy y
q u + = vt t |
,
The stress tensor components are given by
)] ( 3 / 2 ) ( 3 / 4 [
q q
v q v q t
y y x x xx
u u + =
)] ( 3 / 2 ) ( 3 / 4 [
q q
q v q v t u u
x x y y yy
+ + =
and the heat flux vector components are
_
=
+
c
c
=
n
i
i
d
i i x
h u
x
T
k q
1

,
_
=
+
c
c
=
n
i
i
d
i i y
h
y
T
k q
1
v
The variables are the velocity components
u
and
v
,
the pressure
p
,the energy per unit volume e, and the
density of the ith species
i

,with
i
n
i

_
=
=
1
,The
terms
i
s
represent the production of species from
chemical reactions. The grid Jacobian
J
and the
contravariant velocities
U
and
V
are defined as
follow
q q
y x y x J + =
1
v
y x
u U + =
,
v q q
y x
u V + =
q
Jy
x
=
,
q
Jx
y
=

q Jy
x
=
,

q Jx
y
=
The terms
q
x x ,
, etc, are the grid metric terms
c cx
,
q c cx
, etc. The contravariant diffusion
velocities
U
d
i
and
V
d
i
are defined as
d
i y
d
i x
d
i
u U v + =
,
d
i y
d
i x
d
i
u V v q + q =
The diffusion velocities are found by Fick's law:
] ) ( ) ( [
q
q
i x i x i
d
i i
c c D u c + =
] ) ( ) ( [
q
q v
i y i y i
d
i i
c c D c + =
where

i i
c =
. The mixture viscosity and thermal
conductivity, as well as the binary diffusivity of species
i
in the gas mixture, are expressed as:
turbulent ar la
+ =
min
turbulent ar la
k k k + =
min
turbulent
i
ar la
i i
D D D + =
min
The equation of state used is that for a mixture of
thermally perfect gases:
_
=
=
n
i i
i
RT
M
p
1

where M
i
is the molecular weight of the i
th
species. The
temperature T, is determined from the definition of the
total energy
_ _
)
= =
+ =
n
1 i
o
i i
2 2
n
1 i
T
vi i
h c ) v u (
2
1 e
dT c c

Chung Hua Journal of Science and Engineering, Vol. 7, No. 2, pp. 9-14 (2009)
10
where
vi
i
i
c , c

=
is the specific heat at constant volume
of the i
th
species, and
o
i
h
is the heat of formation for
species i.
2.2. Combustion model
The combustion model used for the present study
consists of 6 reacting species H, O, H
2
O, OH, H
2
, O
2
.
Eight reactions are assumed to be significant
1. H+O
2
OH+O
2. H
2
+O H+OH
3. H
2
+OH H
2
O+H
4. 2OH O+ H
2
O
5. H
2
+X 2H+X
6. H
2
O+X OH+H+X
7. OH+X O+H+X
8. O
2
+X 2O+X
where X is any of the above six species. The forward and
backword reaction rate for the ith reaction, Kfi and Kbi
are given by the relation of the form
T / c bi
i i
i
e T A K

=
.
The coefficients of chemical reaction Ai, bi, and ci are
given by Evans and schexnayder. This combustion
model is due to Moretti. Denote Y
i
as the concentration
of species Y
i
i
i
i
M
y

=
, Mi=molecular weight of species i
The chemical equation for reaction j in a gas mixture
containing n species can be writlen as
s
n
1 s
j , s s
n
1 s
j , s
Y V Y V
_ _
= =
' ' '
where
j , s
V'
,
j , s
V' '
are the stoichiometric coefficients
appearing on the left and right in reaction j.
_
H H
=
' ' ' + ' ' ' =
'
=
'
=
j
j , i
i
V
s
n
1 s
j , b j , i j , i
V
s
n
1 s
j , f j , i j , i j , i
y
dt
dy
y K ) V V ( y K ) V V ( y
dt
d
j , i j , i

the components S
i
, of the chemical source term S, are
given by
i i i
M y s =
.
The coefficients of chemical reaction A
i
, b
i
, and C
i
are
listed in Table I and Table II for six species eight
reactions and seven species ten reactions respectively.
Table : Reaction rate coefficients for hydrogen
combustion (six species and eight reactions)
Forward rates Backrward rates Reaction
No.
A
i
b
i
C
i
A
i
b
i
C
i
1 2.2 10
14
0.0 8455 1.5 10
13
0.0 0
2 7.5 10
13
0.0 5586 3.0 10
13
0.0 4429
3 2.0 10
13
0.0 2600 8.4 10
13
0.0 10116
4 5.3 10
12
0.0 503 5.8 10
13
0.0 9059
5 5.5 10
18
-1.0 51987 1.8 10
18
-1.0 0
6 5.2 10
21
-1.5 59386 4.4 10
20
-1.5 0
7 8.5 10
18
-1.0 50830 7.1 10
18
-1.0 0
8 7.2 10
18
-1.0 59340 4.0 10
17
-1.0 0
Table II: Skeletal Reaction Mechanism for hydrogen
combustion (seven species and ten reactions)
Reaction No. Aj bj Cj
(1)H+O2
OH+O 2.00*10
14
0.0 70.3
(2)O+H2
OH+H 1.80*10
10
1.0 36.93
(3)OH+H2
H2O+O 1.17*10
9
1.3 15.17
(4)OH+OHH2O+O 6.04*10
8
1.3 0.23
(5)H+HO2
OH+OH 1.50*10
14
0.0 4.20
(6)H+OH+MH2O+M 2.20*10
22
-2.0 0.00
(7)H+H+MH2O+M 1.80*10
18
-1.0 0.00
(8)O+O+MO2+M 6.17*10
15
-0.5 0.00
(9)H+O+MOH+M 6.20*10
16
-0.6 0.00
(10)H+O2+MHO2+M 2.30*10
18
-0.8 0.00
units are cm, mol, KJ, and K.
III. Numerical procedure
The set of governing equations are integrated over the
control volumes, which produces a set of algebraic
equations. The SIMPLE algorithm is used to solve the
coupled system of governing equations. The set of
algebraic equations are solved sequentially. A
second-order upwind differencing scheme is used for the
formulation of the convection contribution to the
coefficients in the finite-volume equations. The
computation is terminated when the residuals for the
continuity and momentum equations get below 10
-5
and
Chung Hua Journal of Science and Engineering, Vol. 7, No. 2, pp. 9-14 (2009)
11
the residual for the energy equation gets below 10
-8
. The
more stringent convergence criterion for the energy
equation is used to ensure global heat balance. The
calculations are done using the FLUENT version 6
commercial code.
IV. Initial and boundary conditions
Table III: Initial and boundary conditions.
HYDROGEM JET FREE
STREAM
MACH NUMBER, M
TEMPERATURE, T, K
VELOCITY, u, m/s
PRESSURE, p, MPa
MASS FRACTION:
H2
O2
N2
H2O
2.00
251
2432
0.1
1.0
0
0
0
1.90
1495
1510
0.1
0
0.241
0.478
0.281
V. Result
The figture 1 and 2 shows the comparison the H
2
mass
fraction curve in the section of x=8.25D and 27.9D with
experiment result is filled square and present is line.
From to the there are phenomenon. We can see the the
mass fraction date of different vary large of the same
chemistry element could be different significantly.
Figture 3 and 4 comparison the H2O mass fraction curve
with experiment result is filled square and present is line.
The H
2
O produce are very small in the two step
reactions.
The figture 5 and 6 shows the comparison the N
2
mass
fraction curve in the different section of x=8.25D and
x=27.9D with experiment result is filled square and
present is line. The distribution place not enough in the
section of x=8.25D, comparison shows the large
difference with experiment result and simulations section
x=27.9D in the middle to x=0.75D. But comparison the
N
2
mass fraction with experiment result and simulations
very match.
VI. Conclusions
The foregoing discussion of results demonstrates that a
combined finite-rate chemistry/unmixedness model can
provide a analyzing supersonic mixing of two
co-flowing fluid streams (H2/O2) with shock-induced
combustion. All calculations were carried out with fixed
constants in the mixing and chemistry models, in
contrast to the Evans et al.(1978) using an eddy breakup
model in which large adjustments of a parameter were
necessary. Ignition characteristics of the flames were
found to be sensitive both to the chemical kinetics and to
the unmixedness correction. The principal conclusions
reached from the work reported here are:
1. Build a 2D numerical code for solving the
Navier-Stokes equations describing hypersonic flow
which involves shock-induced finite-rate combustion of
H
2
.
2. According to section of 8.25D and 27.9D, the mass
fraction date of different vary large of the same
chemistry element could be different significantly.
3. Form to H
2
O mass fraction date can shows H
2
O
produce found in the present study were lower than the
Evans et al.(1978) in the two reactions.
4. Application of this technique to simulate physical
problems such as scramjet, Ram accelerator, re-entry etc.,
is straightforward because all these kind of examples
involve mixing process with finite-rate chemical
reactions.
Acknowledgments
The authors are deeply appreciated to the financial
support of the Chung-Hua University, projects:
CHU-96-M-03
References
[1] Evans, J. S. and Schexnayder, C. J. and Jr. and
Beach, H.L., Application of a two- dimensional
parabolic computer program to prediction of
turbulent reacting flow,NASA TP 1169(1978).
[2] Evans, J. S. and Schexnayder, C. J. and Jr. and
Beach, H.L., Application of a two- dimensional
parabolic computer program to prediction of
turbulent reacting flow,NASA TP 1169(1978).
[3] S.B. Dworkin, B.A.V. Bennett, M.D. Smooke A
mass-conserving vorticity- velocity formulation
with application to nonreacting and reacting flows
Journal of Computational Physics 215, 430-447
(2006).
[4] B. Fiorina, S.K. Lele. An artificial nonlinear
diffus ivity method for supersonic reacting flows
with shocks Journal of Computational Physics
(2006).
[5] Pino SABIA, Eva SCHIE WOHL et al.
Numerical Analysis of Hydrogen Mild
Combustion Turkish J. Eng. Env. Sci. 30 (2006) ,
127 -134.
[6] Andrew W. Cook, William H. Cabot
Hyperviscosity for shock-turbulence interactions
Journal of Computational Physics 203, 379-385
(2005).
[7] V. Daru C. Tenaud High order one-step
monotonicity-preserving schemes for unsteady
compressible flow calculations Journal of
Chung Hua Journal of Science and Engineering, Vol. 7, No. 2, pp. 9-14 (2009)
12
Computational Physics 193 (2004) 563594
[8] Andrew W. Cook, William H. Cabot A
high-wavenumber viscosity for high-resolution
numerical methods Journal of Computational
Physics 195 (2004) 594601
[9] Cheng-Hung Huang, Chun-Ying Yeh, Helcio R. B.
Orlande A nonlinear inverse problem in
simultaneously estimating the heat and mass
production rates for a chemically reacting fluid
Chemical Engineering Science 58 (2003) 3741
3752
[10] Sergio Pirozzoli Conservative Hybrid
Compact-WENO Schemes for Shock-Turbulence
Interaction Journal of Computational Physics 178,
81-117 (2002).
[11] Levy, D., Puppo, G.,and Russo, G., A
Fourth-order Central WENO Schemes for
Multi-dimensional Hyperbolic System of
Conservation Laws, SAIM J. Sci Comput.
Vol.24,No.2,pp.480-506, 2002.
[12] Xiaogang Deng and Hanxin Zhang Developing
High-Order Weighted Compact Nonlinear
Schemes Journal of Computational Physics 165,
22-44 (2000).
[13] Levy, D., Puppo, G.,and Russo,G., A Third-order
Central WENO Schemes for 2D Conservation
Laws, Applied Numerical Mathematics, 33,
415-421, 2000.
[14] Beth Anne V. Bennett and Mitchell D. Smooke
Local Rectangular Refinement with Applocation
to Nonreacting and Reacting Fluid Flow
Problems, Journal of Computational Physics 151,
684-727 (1999).
[15] B. Fiorna, S.K. Lele An artificial nonlinear
diffusivity method for supersonic reacting flow
with shocks Journal of Computational Physics
(1999).
[16] Guang-Shan Jiang and Chi-Wang Shu Efficient
Implementation of Weighted ENO Schemes
Journal of Computational Physics 126, 202-228
(1996).
[17] Liu, Xu-Dong, Osher, S.,and Chan, T., Weighted
Essentially Non-oscillatory Schemes, J. of
Comput. Phys., 115,200-212, 1994.
[18] C.J. Montgomery, G. Kosaly, and J, J. Riley,
Direct Numerical Simulation of Turbulent
Reacting Flow Using a Reduced H2/O2
Mechanism Combustion and Flame
95:247-260(1993.)
Fig.1 Comparison the H
2
mass fraction curve with Evans
et al.(1978) experiment result; , and Present; , on the
x=8.25D.
Fig.2 Comparison the H
2
mass fraction curve with Evans
et al.(1978) experiment result; , and Present; , on the
x=27.9D.
Chung Hua Journal of Science and Engineering, Vol. 7, No. 2, pp. 9-14 (2009)
13
!
Fig.3 Comparison the H
2
O mass fraction curve with
Evans et al.(1978) experiment result; , and Present; ,
in the x=8.25D.
Fig.4 Comparison the H2O mass fraction curve with
Evans et al.(1978) experiment result; , and Present; ,
on the x=27.9D.
Fig.5 Comparison the N
2
mass fraction curve with Evans
et al.(1978) experiment result; , and Present; , on the
x=8.25D.
Fig.6 Comparison the N
2
mass fraction curve with Evans
et al.(1978) experiment result; , and Present; , on the
x=27.9D.
Chung Hua Journal of Science and Engineering, Vol. 7, No. 2, pp. 9-14 (2009)
14