JID:AESCTE AID:4460 /FLA [m5G; v1.232; Prn:8/03/2018; 8:57] P.

1 (1-12)
Aerospace Science and Technology ••• (••••) •••–•••

1 67
Contents lists available at ScienceDirect
2 68
3 69
4 Aerospace Science and Technology 70
5 71
6 72
7 www.elsevier.com/locate/aescte 73
8 74
9 75
10 76
11 77
12
Combustion characteristics of hydrogen and cracked kerosene in a DLR 78
13 scramjet combustor using hybrid RANS/LES method 79
14 80
15 a b,∗ 81
Junsu Shin , Hong-Gye Sung
16 82
17 a 83
Department of Aerospace and Mechanical Engineering, Korea Aerospace University, Goyang-si, Gyeonggi-do, 10540, Republic of Korea
b
18 School of Aerospace and Mechanical Engineering, Korea Aerospace University, Goyang-si, Gyeonggi-do, 10540, Republic of Korea 84
19 85
20 86
21 a r t i c l e i n f o a b s t r a c t 87
22 88
Article history: This paper applies a zonal hybrid RANS/LES framework to analyze supersonic combustion in a model
23 89
Received 14 October 2017 scramjet combustor. The geometries and boundary conditions of model scramjet combustor are based
24 Received in revised form 1 March 2018 90
on an experiment conducted at DLR, German Aerospace Center. This model scramjet combustor was
25 Accepted 5 March 2018 91
designed to achieve free flight Mach number of 5.5 and total air temperature of 1500 K. Hydrogen
26 Available online xxxx 92
at subcritical conditions and thermal/catalytic cracked kerosene at supercritical conditions are injected
27 as fuel. A surrogate of thermal/catalytic cracked kerosene is composed of ethylene and methane in 93
Keywords:
28 supercritical conditions. To remain consistent with the hydrogen-fueled case, the total equivalence ratio is 94
Zonal hybrid RANS/LES
29 Hydrogen combustion set to 0.034 for both cases. The total equivalence ratio is quite small, so it is not induced flow separation 95
30 Cracked kerosene combustion in the combustor duct. The thermodynamic and transport properties of the supercritical thermal/catalytic 96
31 Combustion characteristics cracked kerosene are calculated using the Redlich–Kwong Peng–Robinson cubic equation of state and 97
Supercritical condition Chung’s model for viscosity and conductivity. This paper focuses on comparisons of the subcritical
32 98
33 hydrogen-fueled and supercritical cracked kerosene-fueled scramjet combustors in terms of intrinsic flow 99
34
and combustion features. The analysis is demonstrated via a reacting regime diagram in nonpremixed 100
turbulent combustion, flame index contours and scatter plots of the flamelet structure. It is found that
35 101
the cracked kerosene surrogate flame is more vulnerable to quenching than the hydrogen flame, and
36 102
flame quenching occurs in the immediate vicinity of the injector.
37 103
© 2018 Published by Elsevier Masson SAS.
38 104
39 105
40 106
41 107
42 1. Introduction the temperature increases above 750 K, the fuel is thermally and 108
43 catalytically cracked. An aircraft’s fuel pressurization system can 109
44 Scramjet engines have been studied for more than 50 years, generally pressurize up to 2–3 MPa, which exceeds kerosene’s crit- 110
45 but they have not yet reached a stage of development suitable for ical pressure value because of the need to maintain high pressure 111
46 practical applications. Researchers interested in air-breathing en- in the combustor and prevent the fuel from boiling in the cool- 112
47 gines are confronted with a number of fundamental difficulties, ing channel. The characteristics of the decomposition phenomenon 113
48 including the low degree of fuel–air mixing, complications with of kerosene under highly pressurized conditions are very compli- 114
49 respect to supersonic combustion mechanisms and the high heat cated, and this topic has also received attention in recent years. 115
50 loads on the combustor wall during scramjet operation. Within The thermal cracking of hydrocarbon aviation fuels was thoroughly 116
51 the scope of thermal structures, the active cooling system using reviewed by Edwards [1]. One of the challenging tasks is the choice 117
52 the endothermic decomposition of hydrocarbon fuels (particularly of surrogates for thermal/catalytic cracked kerosene. Various stud- 118
53 kerosene) onto the combustor wall is required to alleviate the ther- ies have looked at the thermal/catalytic cracking of kerosene under 119
54 mal loads and stress. For scramjet applications, the fuel may be supercritical conditions. Zhong et al. [2] experimentally investi- 120
55 used to regeneratively cool down internal engine walls. This indi- gated the thermal cracking and endothermicity of China No. 3 121
56 cates that fuel can be utilized as a heat sink source, and it can aviation kerosene and found that the main gaseous products were 122
57 then be injected into the combustor containing higher enthalpy. methane, hydrogen, ethane and ethylene under 3.0–4.5 MPa and 123
58 For hydrocarbon fuels, the temperature at the end of the cool- 780–1050 K conditions. More intensive research on the thermal 124
59 ing channel is typically between 640 to 670 degrees Kelvin, but as cracking phenomenon in regenerative cooling channels was con- 125
60 ducted by Jiang et al. [3]. They carried out experiments concerning 126
61 the detailed species concentration and heat transfer along the cool- 127
62 * Corresponding author. ing channel under supercritical conditions (5 MPa, 950–970 K). 128
63 E-mail address: hgsung@kau.ac.kr (H.-G. Sung). They also constructed an analytical model to study the thermal 129
64 130
https://doi.org/10.1016/j.ast.2018.03.006
65 131
1270-9638/© 2018 Published by Elsevier Masson SAS.
66 132
 JID:AESCTE AID:4460 /FLA [m5G; v1.232; Prn:8/03/2018; 8:57] P.2 (1-12)
2 J. Shin, H.-G. Sung / Aerospace Science and Technology ••• (••••) •••–•••

1 67
2 Nomenclature 68
3 69
4 ai j Cholesky decomposition of Reynolds stress tensor Zc critical compressibility factor 70
5 ac , b c attraction and repulsive factor Z  2 transversal coordinate 71
6 cp specific heat at constant pressure of mixture z variance of the mixture fraction 72
7 cχ model constant Z st stoichiometric mixture fraction 73
8 d function of critical compressibility factor α function for cubic equation of state 74
9 E total energy Z diffusion thickness in mixture fraction space 75
10
fσ shape function δ1 , δ2 functions of critical compressibility factor 76
11
k turbulent kinetic energy εl intensity of lth SEM eddy 77
12
kc function of critical compressibility factor λ heat conduction coefficient 78
13
h specific enthalpy of mixture μ molecular viscosity 79
Mw molecular weight of mixture
14 μt eddy viscosity 80
N number of SEM eddies
15 ρ density 81
pc critical pressure
16 σ control parameter of turbulent structure size 82
p static pressure
17 σk coefficient for two-equation model 83
Rij Reynolds stress tensor
18
Ru universal gas constant
τ molecular stress tensor 84
19
T static temperature
τ Reynolds stress tensor 85
20 χ scalar dissipation rate 86
Tc critical temperature
21 χst stoichiometric scalar dissipation rate 87
Tr reference temperature
22 88
t time χq quenching scalar dissipation rate
23 89
u velocity ω specific dissipation rate
24 90
u velocity fluctuation ωc acentric factor
25 91
VB box of eddies Subscript
26 92
x streamwise coordinate
27 93
xl position of lth SEM eddy i, j spatial coordinate index
28 94
y spanwise coordinate Superscript
29 95
yc function of critical compressibility factor
30 96
Yk mass fraction of species k () Reynolds averaged
()
31 97
Z mixture fraction Favre averaged
32 98
33 99
34 100
cracking of hydrocarbon aviation fuels, and they compared their ditions using the Redlich–Kwong Peng–Robinson (RK–PR) equation
35 101
results with experimental values. They determined that the largest of state (EOS) [6] and Chung’s method, respectively [7].
36 102
portions of gaseous products were propylene, ethylene and ethane. In the present contribution, we tackle a fundamental question
37 103
In addition, liquid products such as alkenes, cycloalkenes and aro- – namely, how does a kerosene-fueled scramjet combustor differ
38 104
matics were observed. Vaden et al. [4] conducted experiments on from a hydrogen-fueled scramjet combustor in terms of the en-
39 105
fuel versus air systems for hydrocarbon fuels using an oscillatory- tire flowfield and the manner of their turbulent combustion? The
40 106
input opposed jet burner at atmospheric pressure. They identified DLR model scramjet combustor [8] is selected to explore these dif-
41 107
ferences. Accordingly, the thermal/catalytic cracked kerosene sur-
42 a simple surrogate with a composition of 64% ethylene and 36% 108
rogate under supercritical conditions is injected into the scramjet
43 methane by molar fraction. They determined that this was a vi- 109
combustor. To maintain consistency with respect to the hydrogen
44 able surrogate for cracked JP-7 kerosene for Hypersonic Interna- 110
combustion, the kerosene fueling calculation has an air–fuel equiv-
45 tional Flight Research Experimentation Program (HIFiRE) scramjets alence ratio identical to that of the hydrogen. The computational
111
46 in terms of flame strength, ignition characteristics and flameout domain is divided into RANS and LES regions. In the RANS re-
112
47 limit. Denman et al. [5] demonstrated a Mach 7.3 scramjet engine gion, the k–ω shear stress transport (SST) model and high-order
113
48 (rather than only a combustor) fueled by hydrogen and hydrocar- 114
upwind convective flux discretization method are applied. On the
49 bon fuels (methane and ethylene) in order to investigate ignition 115
contrary, the LES region employs a low-dissipative convective flux
50 116
and combustion characteristics with various fuels. By means of discretization method and improved delayed detached eddy simu-
51 117
their enthusiastic efforts, innovative methodologies and technolo- lation (IDDES). A synthetic eddy method (SEM) is used to connect
52 118
gies were made available for research on hypersonic propulsion the RANS and LES regions. The SEM provides synthesized unsteady
53 119
systems. velocity fluctuations to the LES region using statistically calculated
54 120
In the present study, the two-species surrogate proposed by turbulent kinetic energy based on the flow conditions of the RANS
55 121
Vaden et al. is applied to secure numerical simplicity and rea- region. The predictions of hydrogen-fueled scramjet combustor are
56 122
sonable flame dynamics. If fuel is placed under high pressure and compared with available experimental data, and then the com-
57 123
parisons of hydrogen-fueled and cracked kerosene-fueled scramjet
58 injected into a combustor just below the pressurization pressure, 124
combustors are carried out in terms of flow and combustion char-
59 it may be below or above the critical pressure. In addition, the 125
acteristics.
60 temperature will be high because the fuel will be exposed to the 126
61 heat transfer in the micro cooling channel. Thus, it will gener- 2. Methodology 127
62 ally rise above the critical temperature of the matter. Thus, in 128
63 the immediate vicinity of the injector, a small part of the fuel is 2.1. Governing equations 129
64 placed in a gray area of supercritical and subcritical conditions 130
65 even though the combustor is atmospheric. This paper considers Favre averaged compressible Navier–Stokes equations for mass, 131
66 thermodynamic and transport properties under supercritical con- momentum and total energy conservation are 132
 JID:AESCTE AID:4460 /FLA [m5G; v1.232; Prn:8/03/2018; 8:57] P.3 (1-12)
J. Shin, H.-G. Sung / Aerospace Science and Technology ••• (••••) •••–••• 3

1 67
2 68
3 69
4 70
5 71
6 72
7 73
8 74
9 75
10 76
11 77
12 78
13 79
14 80
15 81
Fig. 1. Schematic of physical and numerical domain.
16 82
  
17 ∂ ρ ∂ ρ uj 1 Ru Tc 83
18 + = 0, (1) bc = (6) 84
∂t ∂xj 3 yc + d − 1 pc
19 85
∂ ρ ui ∂ ρ ui uj ∂p ∂    kc
3
τi j + τij ,
20 86
+ + = (2) α= (7)
∂t ∂xj ∂ xi ∂xj 2 + Tr
21 87
22    88
23 ∂ ρ Ẽ ∂((ρ Ẽ + p )uj ) ∂ λ μt ∂ h̃ where T is the static temperature, T c the critical temperature, and 89
+ = +
24 ∂t ∂xj ∂xj cp P rt ∂ x j p c the critical pressure. Chung’s model [7] is used to calculate 90
25    transport properties such as viscosity and thermal conductivity in 91
∂ μt ∂ k  
26 + μ+ + ui τi j + τij , (3) high pressure as well as low-pressure conditions. The Champman– 92
27 ∂xj σk ∂ x j Enskog theory is used to consider the multi-component gas mix- 93
28 ture, and the properties at subcritical conditions are utilized to 94
29 where ρ is the density, p the static pressure, ui the velocity, Ẽ the predict the properties under supercritical conditions. 95
30 total energy, τi j the molecular stress tensor, and τij the Reynolds 96
31 stress tensor. The Reynolds stress tensor, which appears in Eqs. (2) 2.3. Numerical domain and hybrid numerical methodology 97
32 and (3), is closed by IDDES based on k–ω SST turbulence model [9] 98
33 with Sarkar’s compressibility correction [10]. The governing equa- This study considers the configuration and condition of the 99
34 tions presented are solved via structured finite volume method. model scramjet combustor experimented on at DLR. The schematic 100
35 of physical and numerical domain is shown in Fig. 1. The com- 101
36 2.2. Thermodynamic and transport properties bustor has a height of 0.05 m, width of 0.04 m and depth of 102
37 0.0024 m. The length of constant area is 0.109 m and remaining 103
38 In order to consider properties under supercritical conditions, area expand with an angle of 3◦ . This combustor utilizes strut- 104
39
the RK–PR EOS [6] and Chung’s model [7] are employed in the based injectors with 15 injector holes in the experimentation [8], 105
40
present study. Intermolecular forces and molecular body volume but in the present study, only one injector is employed as indicated 106
41 in Fig. 1. The strut has a length of 0.032 m and height of 0.006 m 107
are important factors when considering real fluid. To bring real
42 with an angle of 12◦ . Periodic boundary conditions are applied to 108
fluid into a numerical calculation, the van der Waals force, molec-
43 the transversal planes, taking account of the interactions between 109
ular dissociation and non-equilibrium thermodynamic quantities
44 adjacent injectors. The boundary conditions for all of the walls are 110
must be considered at high pressure and in a low-temperature
45 set to no-slip, adiabatic conditions. A 1.8 million grid mesh is ap- 111
46 regime. The supercritical modeling for kerosene is vulnerable to plied with minimum mesh spacing of 2 × 10−7 m for all of the 112
47 numerical instability because of the abrupt change in density dif- solid walls. 113
48 fusion [11–13]. The RK–PR EOS is known to be a more reasonable In this study, hybrid numerical methodology is developed to 114
49 cubic EOS for a wide range of molecular weights, including hydro- obtain spatially and temporally accurate solutions. The hybrid nu- 115
50 carbon species, than the Soave–Redlich–Kwong (SRK) EOS [14] and merical methodology involves low-dissipative convective flux dis- 116
51 Peng–Robinson (PR) EOS [15]. These methods are implemented in cretization method [17] and SEM [18]. The low-dissipative con- 117
52 our in-house code and carefully investigated. Kim et al. [16] com- vective flux discretization method uses fourth order central [19] 118
53 pared the thermodynamic properties obtained from these EOSes, 119
and fifth order upwind flux reconstruction schemes [20]. To com-
54 extended corresponding states (ECS) and experimental results. The 120
pute inviscid flux at the cell interface for upwind scheme, advec-
55 121
RK–PR EOS is cubic in that it defines δ1 , δ2 and kc in the func- tion upstream splitting method by pressure-based weight functions
56 122
tion of the critical compressibility factor, Z c , as well as an acentric (AUSMPW+) [21] is utilized. This upwind scheme shows great nu-
57 123
factor, ωc , which is also considered in the SRK and PR EOSes. The merical robustness in the scramjet engine flowpath [22]. The SEM
58 124
RK–PR EOS can be written as is applied to engage the RANS to the LES region and is placed
59 125
to the interface between two regions. It is used to achieve strong
60
ρ Ru T ac αρ 2 126
61 p= − (4) RANS–LES coupling. There are mainly two types of RANS–LES cou- 127
62
M w − bρ ( M w + δ1 bc ρ )( M w + δ2 bc ρ ) pling method. The first one is recycling/rescaling method. This 128
63 The constitutive equations of Eq. (4) are as follows. method requires recycling and rescaling plane which are consec- 129
64 utively situated at different streamwise locations in the numerical 130
  
65 3 y c2 + 3 y c d + d2 + d − 1 R 2u T c2 domain. Thus, it is difficult to apply this method to complex ge- 131
66 ac = (5) ometries. On the other hand, SEM requires one plane which should 132
(3 y c + d − 1)2 pc
 JID:AESCTE AID:4460 /FLA [m5G; v1.232; Prn:8/03/2018; 8:57] P.4 (1-12)
4 J. Shin, H.-G. Sung / Aerospace Science and Technology ••• (••••) •••–•••

1 67
2 68
3 69
4 70
5 71
6 72
7 73
8 74
9 75
10 76
11 77
12 78
13 79
Fig. 2. Schematic of zonal hybrid methodology. (For interpretation of the colors in the figure(s), the reader is referred to the web version of this article.)
14 80
15 81
16 be invoked. At the SEM plane, randomly distributed turbulent ed- Table 1 82
dies pass through. These turbulent eddies affect specific point on Boundary conditions for this study.
17 83
18 SEM plane. The theoretical formulation of SEM can be written as Vitiated air H2 jets Cracked kerosene jets 84
19 Ma 2.0 1.0 1.0 85
N
20 1   U , m/s 730 1200 500 86
21 u i = √ ai j εlj f σ (x) x − x , l
(8) T, K 340 250 700
87
N l =1 p, atm 1.0 1.0 40.0
22 Species (mass fraction) O2 : 0.232 H2 : 1.0 CH4 : 0.243 88
23 89
where u i is the velocity fluctuation. The number of SEM eddies, N,
N2 : 0.736 C2 H4 : 0.757
24 H2 O: 0.032 90
is defined as max( V B /σ 3 ) where V B is the volume of the box of Mass flow rate per – 0.1 0.2
25 91
26
eddies, and σ is the control parameter of turbulent structure size. injector, g/s
92
27 The Cholesky decomposition of Reynolds stress tensor, ai j , is used 93
28 to consider turbulent kinetic energy statistically obtained in RANS 94
2.4. Boundary condition
29 region and it reads 95
30 ⎛ √ ⎞ As described earlier, boundary conditions and geometries of
96
31 R 11 0 0 97
⎜ ⎟ model scramjet combustor are based on the experimentation con-
ai j = ⎜ R 22 − a221 ⎟
32 R /a 0 98
⎝ 21 11 ⎠ (9) ducted at DLR. This combustor was designed to achieve free flight
33 99
Mach number of 5.5. The total air–fuel equivalence ratio was set
34 R 31 /a11 ( R 32 − a21 a31 )/a22 R 33 − a231 − a232 100
to 0.034 which is relatively below typical conditions. Hence, the
35 101
flow separation induced by thermal choking is not expected in
36 where R i j is Reynolds stress tensor. The position, xl , and the inten- 102
this case. The inflow conditions of the vitiated air, hydrogen jets
37 sity, εlj , of lth SEM eddy are randomly generated. A shape function, and cracked kerosene surrogate are given in Table 1. The inflow
103
38 f σ , is utilized to compute the radius of acting of turbulent eddy. 104
conditions of the vitiated air and hydrogen jets are drawn from
39 For f σ it reads 105
Ref. [8]. However, for the cracked kerosene surrogate jets, sev-
40 106
41  l
 eral assumptions are made. Several investigations of the thermal
107
f σ (x) x − x cracking of kerosene under supercritical conditions have been car-
42 108
       ried out. Zhong et al. [2] conducted an experimental study in
43 x − xl y − yl z − zl 109
44 = V B σ −3 f f f (10) which China No. 3 aviation kerosene was thermally cracked in an
110
σ σ σ electrically heated tube at a pressure range of 3–4.5 MPa and a
45 111
temperature range of 780–1050 K. Jiang et al. [3] showed that if
46 where f is a simple tent function. Details of Eqs. (8)–(10) can be 112
the kerosene was pressurized at 5 MPa, an exit temperature of
47 found in Ref. [18]. 113
780–1050 K could be achieved for the regeneratively cooling mi-
48 Fig. 2 shows schematic of zonal hybrid methodology. It presents 114
crochannel. Sun et al. [25] measured the critical properties of China
49 the numerical domain is divided into RANS and LES regions. In 115
No. 3 kerosene and JP-7. They found a critical pressure of 2.39 MPa
50 the RANS region, k–ω SST turbulence model and upwind convec- 116
and critical temperature of 645 K for the former, and a critical
51 tive flux discretization method are employed. In the LES region, 117
pressure of 2.158 MPa and critical temperature of 670 K for the
52 IDDES turbulence model and low-dissipative convective flux dis- 118
latter. Based on previous studies, the boundary conditions of static
53 cretization method are utilized. The LES region is detached from 119
pressure and static temperature at the cracked kerosene surrogate
54 the wall with a thickness of 12 mm to prevent from generating 120
jet are assumed to 4 MPa and 700 K. To remain consistent with
55 spurious wall resolved eddies. The horizontal interfaces between 121
the hydrogen jets, the air–fuel equivalence ratio is matched.
56 the RANS wall regions and LES region are not operated as SEM in- 122
57 terface. The turbulent kinetic energy at this interface is relatively 123
2.5. Nonpremixed turbulent combustion model
58 low when comparing to vicinity of mixing layer or wall. Hence, 124
59 only vertical interfaces as indicated to green solid lines in Fig. 2 125
60 are operated as SEM interface in this study. The turbulent diffusion flame of the hydrogen and cracked 126
61 In order to alleviate numerical instabilities due to the high den- kerosene surrogate is modeled using the steady flamelet approach. 127
62 sity gradient in the compressible reacting flow, dual time stepping The flamelet/progress variable (FPV) approach is a state-of-the-art 128
63 with the preconditioning method [23] and lower–upper symmetric method in terms of the turbulent diffusion flame because it can 129
64 Gauss–Seidel (LU–SGS) method [24] are applied for time integra- predict an unstable branch in the S-curve, enabling the prediction 130
65 tion. An in-house code is paralleled using a message passing inter- of re-ignition and local quenching. However, the steady flamelet 131
66 face (MPI) library to speed up the calculation. approach still can provide meaningful information that identifies 132
 JID:AESCTE AID:4460 /FLA [m5G; v1.232; Prn:8/03/2018; 8:57] P.5 (1-12)
J. Shin, H.-G. Sung / Aerospace Science and Technology ••• (••••) •••–••• 5

1 The scalar dissipation rate, χ̃ is modeled as χ̃ = 0.09 × c χ ω Z

 2 67
2 where c χ is model constant which is constant at 2.0. By using 68

these independent variables ( Z̃ , χ

3 69

st , Z ) species mass fractions can
4 70
5
be interpolated in the CFD code. 71
6
To clarify the combustion characteristics of the hydrogen and 72
7
cracked kerosene surrogate jets, the maximum temperature vari- 73
8
ation in the function of the stoichiometric scalar dissipation rate, 74
9
χst , is shown in Fig. 3. The temperature at the stoichiometric mix- 75
10
ture fraction, Z st , is chosen as the maximum temperature. The sto- 76
11
ichiometric mixture fraction corresponds to 0.028 for the hydrogen 77
12
flame and 0.062 for the cracked kerosene flame. This demonstrates 78
13
that the maximum temperature of the cracked kerosene surrogate 79
14
flame is higher than that of hydrogen flame until it is extinguished. 80
15
The extinction values, χq , are about 130 1/s for the hydrogen flame 81
16
and 124 1/s for the cracked kerosene flame. The temperature pro- 82
17
files in the mixture fraction space along with the stoichiometric 83
18
scalar dissipation rate, χst , are given in Fig. 4. Of note, the cracked 84
19
kerosene surrogate flame has broader and higher temperature pro- 85
Fig. 3. Comparison of maximum temperatures as a function of stoichiometric scalar files than the hydrogen flame because of the higher initial temper-
20 dissipation rate, χst , for hydrogen and cracked kerosene surrogate flame. 86
21
ature. 87
22 88
the combustion regime in the scramjet combustor. In the prepro- 3. Results and discussions
23 89
cessing stage, Flame-Master code [26] solves opposed jet diffusion
24 90
flame problem in fuel–air mixture fraction domain. The tempera- 3.1. Validation with experimental results
25 91
ture boundary conditions for Flame-Master code are set to 250 K
26 92
27
for the hydrogen injection and 700 K for the cracked kerosene sur- In order to validate numerical schemes presented, temperature 93
28
rogate injection. For the hydrogen oxidation process, the reaction and RMS velocity fluctuation are compared to available experimen- 94
29 chemistry developed by O’Connaire et al. [27] with nine species tal data for the case of hydrogen fueling. Fig. 5 shows temperature 95
30 and 19-step reactions is employed, and GRI 3.0 [28] is used for the profiles at different axial location. The overall trend is well cap- 96
31 methane and ethylene reaction mechanism. GRI 3.0 is comprised of tured at x = 0.12 and 0.275 m. However predicted flame thickness 97
32 53 species and 325 chemical reactions. Then, preprocessing code at x = 0.167 m is rather wider than experiment. The plots of RMS 98
33 integrates the flamelet data using presumed probability density velocity fluctuation are compared with experiment in Fig. 6. The 99
34 function (PDF) as follows predicted values at x = 0.12 and 0.199 m are in reasonable agree- 100
35
ment with experiment, and the profile at x = 0.167 m is relatively 101
36 ∞1 lower than experiment. Fig. 7 shows the turbulent kinetic energy 102
37 Yk = Y k ( Z , χst ) P ( Z , χst )d Zdχst . (11) spectrum at x = 0.131 m and y = 0.025 m. The plot follows the 103
38
standard −5/3 law based on the Kolmogorov theory. It supports 104
0 0
39
that the grid resolution used is sufficient to simulate innate turbu- 105
40 The shape of PDF is approximated using presumed beta-PDF. The lence cascading process. 106
41 output of preprocessing stage is chemistry library that includes 107
42 species mass fractions in the function of mixture fraction, Z̃ , sto- 3.2. Flow characteristics 108
43 ichiometric scalar dissipation rate, χ  st , and variance of mixture 109
44 fraction, 
Z  . In the calculation stage, CFD code solves conserva- In the present study, dissimilarities appear between the hydro- 110
45 tion equations of mixture fraction and variance of mixture fraction. gen- and kerosene-fueled scramjet combustors in terms of the 111
46 112
47 113
48 114
49 115
50 116
51 117
52 118
53 119
54 120
55 121
56 122
57 123
58 124
59 125
60 126
61 127
62 128
63 129
64 130
65 131
66 Fig. 4. Temperature profiles along the mixture fraction with varying stoichiometric scalar dissipation rate. 132
 JID:AESCTE AID:4460 /FLA [m5G; v1.232; Prn:8/03/2018; 8:57] P.6 (1-12)
6 J. Shin, H.-G. Sung / Aerospace Science and Technology ••• (••••) •••–•••

1 67
2 68
3 69
4 70
5 71
6 72
7 73
8 74
9 75
10 76
11 77
12 78
13 79
14 80
15 81
16 82
17 83
18 84
19 85
Fig. 7. Turbulent kinetic energy cascade at x = 0.131 m and y = 0.025 m.
20 86
21 Fig. 5. Temperature profiles at different axial locations. 87
22 88
23 over the shear layer upward and downward, as seen in Fig. 9b, 89
24 and a larger recirculation zone volume is observed compared to 90
25 that of Fig. 9a. 91
26 Fig. 10 shows the mean pressure profiles of the cases with 92
27 and without reaction on the combustor’s lower wall along the 93
28 streamwise direction. It can be seen that pressure profile of 94
29 kerosene-fueled case without reaction is slightly lower than that 95
30 of hydrogen-fueled case without reaction at x = 0.25 m. For the 96
31 reacting flows, the peak pressure at approximately x = 0.2 m is 97
32 predicted to be higher than that of the hydrogen-fueled case be- 98
33 cause of the stiff compression angle of the shear layer. 99
34 This tendency can also be found in Fig. 11, which corresponds 100
35 to the time averaged pressure contours. In Fig. 11a, the flowfield 101
36 constructed by the hydrogen flame clearly shows reflected shock 102
37 structures through the combustor. However, the reflected shock 103
38 structures in Fig. 11b disappear from x = 0.22 m. 104
39 This phenomenon originates from the pressure fluctuation, as 105
40 illustrated in Fig. 12. The RMS pressure fluctuation of the kerosene- 106
41 fueled case shows severe acoustic fluctuations downstream of the 107
42 combustor. Generally, a pressure fluctuation is dominated by large- 108
43
Fig. 6. RMS velocity fluctuation at different axial locations. scale vortices in a diffusion flame problem. The momentum of the 109
44 vortices at the shear layer dissipates into small-scale eddies, and at 110
45 a certain point, a shear layer breaks out and turns to begin shed- 111
46 boundary conditions at the injector and adiabatic flame temper- ding. 112
47 ature. The theoretical adiabatic temperature of a hydrogen flame In Fig. 13, the isosurface of the Q-criterion colored by the vor- 113
48 is 2483 K, but it climbs to 2513 K for an ethylene and methane ticity magnitude is represented. The turbulence dominates mix- 114
49 flame under atmospheric pressure and stoichiometric conditions. ing and combustion in recirculation zone at the backward face of 115
50 In addition, the hydrogen flame has a narrower flame thickness wedge. After some distance, shear layers start shedding out rep- 116
51 in the mixture fraction dimension. The instantaneous contours of resenting large-scale coherent vortices. For hydrogen flame case, it 117
52 the vorticity magnitude are shown in Fig. 8. The sliced contours of appears around x = 0.2 m and it is around x = 0.18 m for kerosene 118
53 x– y plane at z = 0.0012 m are depicted on the left-hand side. On flame case. It can be seen more vigorous eddies in the recircula- 119
54 the right-hand side, sliced contours of y–z plane at x = 0.115 m tion zone in the hydrogen flame case, so it takes more time to 120
55 and 0.125 m are illustrated. The eddies are generated in the recir- dissipate. Thus, the kerosene flame transfers to a fully turbulent 121
56 culation region, and they interact with the fuel core. The excited flame more quickly than the hydrogen flame. 122
57 vortices help the shear layer shed out. The shear layer generated Fig. 14 indicates the time averaged contours of the tempera- 123
58 at wedge corner as indicated in sliced contour of y–z plane at ture with the stoichiometric mixture fraction as a solid line, and 124
59 x = 0.115 m is split into small vortices as illustrated in y–z plane the corresponding instantaneous snapshots are shown in Fig. 15. 125
60 at x = 0.125 m for both cases. A higher magnitude of the vorticity The combustion of the cracked kerosene surrogate occurs over a 126
61 value is indicated in the hydrogen-fueled case because it is injected wide area, and the maximum temperature is close to the injec- 127
62 at a faster speed. Also, a smaller vortex structure is observed in tor. This can be explained by the higher temperature of the in- 128
63 the hydrogen-fueled case. The difference in injection conditions in- jectant. Generally, the maximum temperature value appears near 129
64 duces the entire flow structure. the stoichiometric mixture fraction line. In the hydrogen-fueled 130
65 Fig. 9 depicts the time averaged velocity magnitude contours case, high-temperature values are observed close to the stoichio- 131
66 for both injection cases. The highly pressurized injectant pushes metric mixture fraction line. However high-temperature values are 132
 JID:AESCTE AID:4460 /FLA [m5G; v1.232; Prn:8/03/2018; 8:57] P.7 (1-12)
J. Shin, H.-G. Sung / Aerospace Science and Technology ••• (••••) •••–••• 7

1 67
2 68
3 69
4 70
5 71
6 72
7 73
8 74
9 75
10 76
11 77
12 78
13 79
14 80
15 81
16 82
17 83
18 84
19 85
20 86
21 87
22 88
23 89
24 90
25 91
26 92
27 93
28 94
29 95
30 96
31 97
32 98
33 99
34 100
35 101
Fig. 8. Vorticity magnitude contours behind the injector (left-hand side: sliced contour of x– y plane at z = 0.0012 m, right-hand side: sliced contour of y–z plane at
36 102
x = 0.115 m and 0.125 m).
37 103
38 104
39 105
40 106
41 107
42 108
43 109
44 110
45 111
46 112
47 113
48 114
49 115
50 116
51 117
52 118
53 119
54 120
55 121
56 122
57 123
Fig. 9. Time averaged velocity magnitude contours behind the injector.
58 124
59 125
60 far from the stoichiometric mixture fraction line in the kerosene- case (Fig. 14b). These tendencies are also observed in Fig. 16. The 126
61 fueled case. As seen in the instantaneous temperature contour profiles of the mean temperature of the cracked kerosene sur- 127
62 (Fig. 15b), the cracked kerosene surrogate flame sheds and oscil- rogate flame suggest that strong combustion occurs close to the 128
63 lates with a larger width than that of the hydrogen flame (Fig. 15a) injector, and the flame width is wider. The temperature profiles 129
64 after x = 0.18 m. Thus, the maxima of the temperature values are downstream of the hydrogen flame indicate higher values, though 130
65 not located in the vicinity of stoichiometric mixture fraction line the hydrogen is injected at a lower temperature than the cracked 131
66 in the time-averaged temperature contour of the kerosene-fueled kerosene surrogate. 132
 JID:AESCTE AID:4460 /FLA [m5G; v1.232; Prn:8/03/2018; 8:57] P.8 (1-12)
8 J. Shin, H.-G. Sung / Aerospace Science and Technology ••• (••••) •••–•••

1 values before x = 0.18 m, and the open symbol denotes the values 67
2 after 0.18 m. In the flame extinction zone, where χq /χ̃st < 1, fewer 68
3 dots are observed in the hydrogen-fueled case, indicating that the 69
4 quenched flame area in the hydrogen-fueled case is smaller than in 70
5 the kerosene-fueled case. This suggests that the cracked kerosene 71
6 72
surrogate flame is highly exposed to quenching comparing to the
7
hydrogen flame. Of note, no scatter plots after x = 0.18 m are 73
8 74
found in the flame extinction zone for both cases. Thus, the region
9 75
experiencing quenching is placed near the injector, specifically be-
10 76
fore x = 0.18 m. The other dots are located in separated flamelets
11 77
and connected flame zones. More scatter plots after x = 0.18 m are
12 78
discovered in the region of separated flamelets in the kerosene-
13 79
fueled case than in hydrogen-fueled case. This suggests that larger
14 80
15
fluctuations in the mixture fraction occur downstream. As illus- 81
16
trated in Fig. 15b, the mixing shear layer flutters severely along 82
17
time and space. When the fluctuation in the mixture fraction is 83
18 smaller, the flames are located in connected flame zones, and most 84
19 Fig. 10. Time averaged pressure profiles on the lower wall. of the connected flame zones are situated after x = 0.18 m in the 85
20 hydrogen-fueled case. The slopes in all of the scatter plots consti- 86
21 3.3. Combustion characteristics tute a −1/2 slope, which was the value theoretically obtained by 87
22 Peters and Williams [30]. 88
In order to characterize the combustion regime Fig. 18 indicates sliced contours of flame index. The flame in-
23
 in the non- 89
24 premixed turbulent combustion, a log–log plot of Z  2 / Z ver- dex is given by ∇ Y F · ∇ Y O where subscripts F and O denote the 90
25  fuel and oxidizer, respectively. The plus and minus signs in the in- 91
χst is illustrated in Fig. 17. The y-axis, Z 2 / Z , rep-
sus χq /
26 dex indicate premixed and nonpremixed flames, respectively. The 92
resents the fluctuation in the mixture fraction space where  Z
27 sliced x– y plane contours for hydrogen-fueled and kerosene fu- 93
is the diffusion thickness. The x-axis, χq /χ̃st , represents the time
28 eled case are illustrated in top and bottom images of Fig. 18. The 94
scale ratio. The quenching scalar dissipation rate, χq , and stoichio-
29 sliced y–z plane contours for both cases are drawn in middle im- 95
metric scalar dissipation rate, χ̃st , were discussed in Section 2.5.
30 age of Fig. 18. The premixed regions exist locally in small areas, 96
These values are extracted along the surface of the stoichiomet-
31 and most of the flame is nonpremixed. The lean nonpremixed re- 97
ric mixture fraction. Detailed information on the regime in the
32 gion surrounds the rich nonpremixed region because it adjoins the 98
nonpremixed turbulent combustion can be found in Ref. [29]. The
33 air oxidizer. Of note, there is a white region in the shear layer 99
scatter plots of the hydrogen- and kerosene-fueled cases are pre-
34 100
sented in Fig. 17. The scatter plots are extracted at the stoichiomet- of the kerosene-fueled case that is not observed in the hydrogen-
35 101
ric mixture fraction line, which is illustrated in Fig. 14. It should fueled case. For the kerosene-fueled case, it clearly can be seen
36 102
be notified that the scatter plots are divided at x = 0.18 m. This in sliced y–z contours at x = 0.13 m and 0.14 m that the void
37 103
location is selected to distinguish the upstream and downstream area exist in this region. This white region is where the flame in-
38 104
information. The fully turbulent combustion flame appears after dex is zero, meaning there is no gradient of fuel or oxidizer. This
39 105
x = 0.18 m for both cases. The closed circle symbol indicates the is caused by the oxygen deficiency inside the mixing shear layer.
40 106
41 107
42 108
43 109
44 110
45 111
46 112
47 113
48 114
49 115
50 116
51 117
52 118
53 Fig. 11. Time averaged pressure contours. 119
54 120
55 121
56 122
57 123
58 124
59 125
60 126
61 127
62 128
63 129
64 130
65 131
66 Fig. 12. RMS pressure contours. 132
 JID:AESCTE AID:4460 /FLA [m5G; v1.232; Prn:8/03/2018; 8:57] P.9 (1-12)
J. Shin, H.-G. Sung / Aerospace Science and Technology ••• (••••) •••–••• 9

1 67
2 68
3 69
4 70
5 71
6 72
7 73
8 74
9 75
10 76
11 77
12 78
13 79
14 80
15 81
16 82
17 83
18 84
19 Fig. 13. Isosurface of Q-criterion colored by vorticity magnitude. 85
20 86
21 87
22 88
23 89
24 90
25 91
26 92
27 93
28 94
29 95
30 96
31 97
32 98
Fig. 14. Time averaged temperature contours. The solid lines in a) and b) indicate Z st = 0.028 for hydrogen fueling and Z st = 0.062 for kerosene fueling.
33 99
34 100
35 101
36 102
37 103
38 104
39 105
40 106
41 107
42 108
43 109
44 110
45 111
46 Fig. 15. Instantaneous temperature contours. 112
47 113
48 114
49 The oxidizer is not able to diffuse into the shear layer because the the mixture fraction varies from 0 to approximately 0.6 for the 115
50 density of the injectant is higher than in the hydrogen-fueled case. kerosene-fueled case and 0.4 for the hydrogen-fueled case as the 116
51 Fig. 19 depicts two-dimensional scatter plots given by the tem- fuel diffuses downstream. In the fully turbulent zone (Fig. 19c), the 117
52 perature versus mixture fraction. The data points are reduced to maximum temperatures are similar to the transition zone values 118
53 1/50 for the clarity. The induction zone is defined as the region for both cases. The data points are recovered near the equilibrium 119
54 from x = 0.109 to x = 0.128, which represents the majority of the limits because the fluctuation in the mixture fraction downstream 120
55 fuel core diffusion for both cases. The transition zone is defined is insignificant, as seen in Fig. 17. 121
56 by different locations. For the hydrogen-fueled case, the region 122
57 from x = 0.128 to x = 0.2 m is set as the transition zone. It is 4. Conclusion 123
58 defined from x = 0.128 to x = 0.18 m for the kerosene-fueled case. 124
59 The end of the transition zone is defined as the point where the In the present study, the turbulent combustion of hydrogen and 125
60 shear layer starts shedding. The fully turbulent zone is set as the a thermal/catalytic cracked kerosene surrogate is simulated using a 126
61 rest of the area from the transition point. Table 2 summarizes the zonal hybrid RANS/LES method and steady flamelet approach. The 127
62 definition of regions for each case. In Fig. 19a, the scatter data cracked kerosene is assumed to be a surrogate of two species, and 128
63 calculated in the induction zone are distributed over the entire its thermal and transport properties under supercritical conditions 129
64 mixture fraction space. The temperature around the stoichiometric are calculated using the RK–PR EOS and Chung’s method. In order 130
65 mixture fraction varies dynamically because the flames experience to investigate dissimilarities between the hydrogen- and cracked 131
66 quenching in this area. In the transition zone depicted in Fig. 19b, kerosene surrogate-fueled scramjet combustors, a model scram- 132
 JID:AESCTE AID:4460 /FLA [m5G; v1.232; Prn:8/03/2018; 8:57] P.10 (1-12)
10 J. Shin, H.-G. Sung / Aerospace Science and Technology ••• (••••) •••–•••

1 67
2 68
3 69
4 70
5 71
6 72
7 73
8 74
9 75
10 76
11 77
12 78
13 79
14 80
15 81
16 82
17 83
18 84
19 85
20 86
Fig. 16. Profiles of mean temperature at different streamwise locations.
21 87
22 88
23 89
24 90
25 91
26 92
27 93
28 94
29 95
30 96
31 97
32 98
Fig. 18. Instantaneous contours of flame index (top image: sliced x– y plane contour
33 99
in hydrogen-fueled case, middle image: sliced y–z plane contour for both cases,
34 bottom image: sliced x– y plane contour in kerosene-fueled case). 100
35 101
36 102
37 103
38 fueled case, the shedding mode of the mixing shear layer ap- 104
39 pears close to the injector when compared to the hydrogen-fueled 105
40 case, and this leads to a faster transition to a fully turbulent 106
41 flame. 107
42 To investigate the combustion characteristics, the nonpremixed 108
43 turbulent combustion regime, flame index and flamelet structure 109
44 are presented. The analysis of nonpremixed turbulent combustion 110
Fig. 17. Regime in nonpremixed turbulent combustion. regime reveals that nonpremixed turbulent flame encompasses
45 111
46 flame extinction, separated flamelets and connected flame zones. 112
47 The entire scatter plots follow the slope of −1/2 which is the- 113
jet combustor experimented on at DLR is selected. This model
48 oretically obtained value. Also, distinctive observations are found 114
scramjet combustor is designed to achieve free flight Mach num-
49 ber of 5.5 and inflow Mach number at combustor inlet is 2.0. with respect to quenching characteristics. The scatter plots in- 115
50 The desired total air–fuel equivalence ratio is 0.034 which is quite dicate that more scatter dots are observed in flame extinction 116
51 small value. Thus, flow separation induced by thermal choking is zone of cracked kerosene surrogate flame. It suggests that cracked 117
52 not expected. In order to validate numerical methodologies used, kerosene surrogate flame is more vulnerable to quenching com- 118
53 the numerical results are compared with available experimental pared to the hydrogen flame, and the quenching region is sit- 119
54 data. uated in the vicinity of the injector for both cases. It is also 120
55 The flow characteristics of the supercritical kerosene-fueled supported by the fact that quenching scalar dissipation rate of 121
56 scramjet combustor and subcritical hydrogen are investigated. In- cracked kerosene surrogate flame is less than that of hydrogen 122
57 jection at a higher pressure and density causes differences in the flame. As shown in the flame index analysis, most of the flame 123
58 flowfield and combustion regarding the location at which the fully is nonpremixed. However, there is an oxygen free pocket within 124
59 turbulent flame occurs and the combustion process in the com- the shear layer in the kerosene-fueled case, which occurs be- 125
60 bustor. With respect to the flow structure, the higher pressure cause the oxygen is not able to diffuse into the shear layer due 126
61 and density of the injectant generate a larger recirculation zone, to the higher injectant density. The scatter plots of the tempera- 127
62 and this causes a rise in the pressure profile at the lower wall ture versus mixture fraction suggest that the temperature varies 128
63 for the kerosene-fueled case. Also, the time averaged pressure dynamically in the region where the flame experiences quench- 129
64 contours indicate that the reflected shock structure is no longer ing. However, the scatter plots converge into the equilibrium limit 130
65 present because of the pressure fluctuation downstream accord- due to the narrow mixture fraction variation in the fully turbulent 131
66 ing to the RMS pressure fluctuation contours. In the kerosene- zone. 132
 JID:AESCTE AID:4460 /FLA [m5G; v1.232; Prn:8/03/2018; 8:57] P.11 (1-12)
J. Shin, H.-G. Sung / Aerospace Science and Technology ••• (••••) •••–••• 11

1 Table 2 67
2 Definition of regions for induction, transition and fully turbulent zones. 68
3 Cases Induction zone [m] Transition zone [m] Fully turbulent zone [m] 69
4 70
Start End Start End Start End
5 71
Hydrogen 0.109 0.128 0.128 0.200 0.200 0.340
6 72
Cracked kerosene surrogate 0.109 0.128 0.128 0.180 0.180 0.340
7 73
8 74
9 Conflict of interest statement 75
10 76
11 None declared. 77
12 78
13
Acknowledgements 79
14 80
This work was supported by the Ministry of Trade, Industry and
15 81
Energy and the Korea Aerospace Technology Research Association
16 82
under project number 10050539.
17 83
18 84
References
19 85
20 [1] T. Edwards, Cracking and deposition behavior of supercritical hydrocarbon avia- 86
21 tion fuels, Combust. Sci. Technol. 178 (2006) 307–334, https://doi.org/10.1080/ 87
22 00102200500294346. 88
[2] F. Zhong, X. Fan, G. Yu, J. Li, Thermal cracking and heat sink capacity of avia-
23 89
tion kerosene under supercritical conditions, J. Thermophys. Heat Transf. 25 (3)
24 (2011) 450–455, https://doi.org/10.2514/1.51399. 90
25 [3] R. Jiang, G. Liu, X. Zhang, Thermal cracking of hydrocarbon aviation fuels in 91
26 regenerative cooling microchannels, Energy Fuels 27 (5) (2013) 2563–2577, 92
https://doi.org/10.1021/ef400367n.
27 93
[4] S.N. Vaden, R.L. Debes, E.L. Lash, R.S. Burk, C.M. Boyd, L.G. Wilson, G.L. Pel-
28 let, Unsteady extinction of opposed jet ethylene/methane HIFiRE surrogate fuel 94
29 mixtures vs air, in: 45th AIAA/ASME/SAE/ASEE Joint Propulsion Conference & 95
30 Exhibit, Denver, Colorado, USA, 2009, AIAA-2009-4883. 96
[5] Z.J. Denman, W.Y.K. Chan, S. Brieschenk, A. Veeraragavan, V. Wheatley, M.K.
31 97
Smart, Ignition experiments of hydrocarbons in a Mach 8 shape-transitioning
32 98
scramjet engine, J. Propuls. Power 32 (6) (2016) 1462–1471, https://doi.org/10.
33 2514/1.B36099. 99
34 [6] M. Cismondi, J. Mollerup, Development and application of a three-parameter 100
35 RK–PR equation of state, Fluid Phase Equilib. 232 (1–2) (2005) 74–89, https:// 101
doi.org/10.1016/j.fluid.2005.03.020.
36 102
[7] T.H. Chung, M. Ajlan, L.L. Lee, K.E. Starling, Generalized multiparameter corre-
37 lation for nonpolar and polar fluid transport properties, Ind. Eng. Chem. Res. 103
38 27 (4) (1988) 671–679, https://doi.org/10.1021/ie00076a024. 104
39 [8] W. Waidmann, F. Alff, M. Böhm, W. Clauß, M. Oschwald, Supersonic combus- 105
tion of hydrogen/air in a scramjet combustion chamber, Space Technol. 15 (6)
40 106
(1995) 421–429.
41 [9] M.L. Shur, P.R. Spalart, M.K. Strelets, A.K. Travin, A hybrid RANS–LES approach 107
42 with delayed-DES and wall-modeled LES capabilities, Int. J. Heat Fluid Flow 108
43 29 (6) (2008) 1638–1649, https://doi.org/10.1016/j.ijheatfluidflow.2008.07.001. 109
[10] H.-G. Sung, S.-J. Kim, H.-W. Yeom, J.-Y. Heo, On the assessment of compress-
44 110
ibility effects of two-equation turbulence models for supersonic transition flow
45 with flow separation, Int. J. Aeronaut. Space Sci. 14 (4) (2013) 387–397, https:// 111
46 doi.org/10.5139/IJASS.2013.14.4.387. 112
47 [11] H. Huo, V. Yang, Large-eddy simulation of supercritical combustion: model 113
validation against gaseous H2 –O2 injector, J. Propuls. Power 33 (5) (2017)
48 114
1272–1284, https://doi.org/10.2514/1.B36368.
49 [12] J.S. Kang, J.Y. Heo, H.G. Sung, Y.B. Yoon, Dynamic characteristics of a cryogenic 115
50 nitrogen swirl injector under supercritical conditions, Aerosp. Sci. Technol. 67 116
51 (2017) 398–411, https://doi.org/10.1016/j.ast.2017.04.010. 117
[13] J.S. Kang, H.G. Sung, Kerosene/GOx dynamic combustion characteristics in a
52 118
mixing layer under supercritical conditions using the LES-FPV approach, Fuel
53 203 (2017) 579–590, https://doi.org/10.1016/j.fuel.2017.04.088. 119
54 [14] G. Soave, Equilibrium constants from a modified Redlich–Kwong equation of 120
55 state, Chem. Eng. Sci. 27 (6) (1972) 1197–1203, https://doi.org/10.1016/0009- 121
2509(72)80096-4.
56 122
[15] D.Y. Peng, D.B. Robinson, A new two-constant equation of state, Ind. Eng. Chem.
57 Fundam. 15 (1) (1976) 59–64, https://doi.org/10.1021/i160057a011. 123
58 [16] K.J. Kim, J.Y. Heo, H.G. Sung, Investigation of thermophysical properties of the 124
59 kerosene using the surrogate model fuel at supercritical conditions, J. Korean 125
Soc. Aeronaut. Space Sci. 38 (8) (2010) 823–833, https://doi.org/10.5139/JKSAS.
60 126
2010.38.8.823.
61 [17] F. Génin, S. Menon, Studies of shock/turbulent shear layer interaction using 127
62 Large-Eddy Simulation, Comput. Fluids 39 (2010) 800–819, https://doi.org/10. 128
63 1016/j.compfluid.2009.12.008. 129
64 [18] N. Jarrin, R. Prosser, J.-C. Uribe, S. Benhamadouche, D. Laurence, Reconstruction 130
of turbulent fluctuations for hybrid RANS/LES simulations using a Synthetic-
65 131
Eddy Method, Int. J. Heat Fluid Flow 30 (2009) 435–442, https://doi.org/10.
66 Fig. 19. Scatter plots of temperature versus mixture fraction. 1016/j.ijheatfluidflow.2009.02.016. 132
 JID:AESCTE AID:4460 /FLA [m5G; v1.232; Prn:8/03/2018; 8:57] P.12 (1-12)
12 J. Shin, H.-G. Sung / Aerospace Science and Technology ••• (••••) •••–•••

1 [19] M. Rai, S. Charrvarthy, Conservative high-order-accurate finite-difference meth- [25] Q.M. Sun, Z.T. Mi, X.W. Zhang, Determination of critical properties (T_c, P_c) 67
2 ods for curvilinear grids, in: 11th Computational Fluid Dynamics Conference, of endothermic hydrocarbon fuels RP-3 and simulated JP-7, J. Fuel Chem. Tech- 68
3 Fluid Dynamics and Colocated Conferences, Orlando, Florida, USA, 1993, AIAA- nol. 34 (4) (2006) 466–470 [in Chinese]. (Note: Article published in another 69
93-3380. language).
4 70
[20] K.K. Kim, C.A. Kim, Accurate, efficient and monotonic numerical methods for [26] H. Pitsch, https://www.itv.rwth-aachen.de/index.php?id=flamemaster. (Ac-
5 multi-dimensional compressible flows, Part II: Multi-dimensional limiting pro- 71
cessed 26 September 2017).
6 cess, J. Comput. Phys. 208 (2005) 570–615, https://doi.org/10.1016/j.jcp.2005. 72
[27] M. O’Connaire, H.J. Curran, M. Simmie, J. Pitz, C.K. Westbrook, A comprehensive
7 02.022. 73
modeling study of hydrogen oxidation, Int. J. Chem. Kinet. 36 (2004) 603–622,
[21] K.H. Kim, C. Kim, O.-H. Rho, Methods for the accurate computations of hyper-
8 https://doi.org/10.1002/kin.20036. 74
sonic flows I. AUSMPW+ scheme, J. Comput. Phys. 174 (2001) 38–80, https://
9 doi.org/10.1006/jcph.2001.6873. [28] P.G. Smith, D.M. Golden, M. Frenklach, N.W. Moriarty, B. Eiteneer, M. Gold- 75
10 [22] H.-W. Yeom, B.-G. Seo, H.-G. Sung, Numerical analysis of a scramjet engine enberg, C.T. Bowman, R.K. Hanson, S. Song, W.C. Gardiner, V. Lissianski, Z. 76
with intake sidewalls and cavity flameholder, AIAA J. 51 (7) (2013) 1566–1575, Qin, 2007, http://www.me.berkeley.edu/gri-mech/releases.html. (Accessed 26
11 77
https://doi.org/10.2514/1.J051677. September 2017).
12 78
[23] H. Meng, V. Yang, A unified thermodynamics treatment of general fluid mix- [29] N. Peters, Turbulent Combustion, 1st edn., Cambridge University Press, Cam-
13 tures and its application to a preconditioning scheme, J. Comput. Phys. 189 (1) bridge, UK, 2000, 324 pp. 79
14 (2003) 277–304, https://doi.org/10.1016/S0021-9991(03)00211-0. [30] N. Peters, F.A. Williams, Liftoff characteristics of turbulent jet diffusion flames, 80
15 [24] S. Yoon, A. Jameson, Lower–upper symmetric Gauss–Seidel method for the Eu- AIAA J. 21 (3) (1983) 423–429, https://doi.org/10.2514/3.8089. 81
ler and Navier–Stokes equations, AIAA J. 26 (9) (1988) 1025–1026, https://
16 82
doi.org/10.2514/3.10007.
17 83
18 84
19 85
20 86
21 87
22 88
23 89
24 90
25 91
26 92
27 93
28 94
29 95
30 96
31 97
32 98
33 99
34 100
35 101
36 102
37 103
38 104
39 105
40 106
41 107
42 108
43 109
44 110
45 111
46 112
47 113
48 114
49 115
50 116
51 117
52 118
53 119
54 120
55 121
56 122
57 123
58 124
59 125
60 126
61 127
62 128
63 129
64 130
65 131
66 132