You are on page 1of 13

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 6 ( 2 0 2 1 ) 2 3 0 1 3 e2 3 0 2 5

Available online at www.sciencedirect.com

ScienceDirect

journal homepage: www.elsevier.com/locate/he

Effect of hydrogen jets in supersonic mixing using


strut injection schemes

S. Jeyakumar a, Jayaraman Kandasamy b,*, Mehmet Karaca c, K. Karthik a,


R. Sivakumar d
a
CFD Center, Aeronautical Engineering, Kalasalingam Academy of Research and Education, Krishnankoil, India
b
Mechanical Engineering, Middle East Technical University, Ankara, Turkey
c
Aerospace Engineering, Middle East Technical University, Ankara, Turkey
d
School of Mechanical Engineering, VIT University, Chennai, India

highlights

 Numerical investigation on a modified fuel injection strategy in scramjet combustor.


 Reynolds averaged Navier stokes equation adopted with SST k-u turbulence model.
 Dual fuel injection scheme in the flow reduces the impact of shock waves.
 Unsteady Detached Eddy Simulation results analyzed using Dynamic Mode Decomposition.
 Primary peaks corresponding to dominant coherent modes determined using DDES method.

article info abstract

Article history: The prevalence of complex phenomena associated with the fuel mixing of a supersonic
Received 26 February 2021 stream in scramjet combustor is inherently occurred due to the short residence time. An
Received in revised form efficient injection mechanism is required to enhance the mixing and improve combustion
18 April 2021 efficiency. This numerical simulation study aims to reveal the performance of modified
Accepted 20 April 2021 strut injection strategies within a Mach 2.0 flow field. Two-dimensional steady and tran-
Available online 22 May 2021 sient Navier-Stokes computations of the DLR scramjet experiment is performed for various
strut injection locations. The Reynolds Averaged Navier Stokes equation with the SST k-ε
Keywords: turbulence model is utilized to solve the flow field under steady conditions. The critical
Hydrogen parameters examined to investigate steady solutions are wall static pressure, flow Mach
Strut injection number, and total pressure loss across the combustor. The dual injection configuration in
Scramjet the flow considerably reduces the shock waves impact at the downstream of the strut and
Dynamic mode decomposition preserves the magnitude of internal forces acting on combustor walls and total pressure
Dominant modes loss. Unsteady Detached Eddy Simulation (DES) results for hydrogen concentration and
velocity field are analyzed by applying Dynamic Mode Decomposition (DMD). Multiple in-
jections are observed to alter the frequency and the number of dominant modes.
© 2021 Hydrogen Energy Publications LLC. Published by Elsevier Ltd. All rights reserved.

* Corresponding author.
E-mail address: jayaraman78@gmail.com (J. Kandasamy).
https://doi.org/10.1016/j.ijhydene.2021.04.123
0360-3199/© 2021 Hydrogen Energy Publications LLC. Published by Elsevier Ltd. All rights reserved.
23014 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 6 ( 2 0 2 1 ) 2 3 0 1 3 e2 3 0 2 5

Introduction

The air-breathing hypersonic propulsion system requires a


supersonic combustion ramjet engine (scramjet), which is
considered as the future for space transportation and high
altitude vehicles. The scramjet engine's challenge involves
fuel/air mixing and the subsequent combustion process at
supersonic speeds. Since the air stream's residence time in the
combustor is short, an efficient fuel injection strategy should
be adopted to enhance the scramjet engine's mixing and
combustion efficiency. Many researchers have investigated
various injection strategies to improve fuel/air mixing, com-
bustion, and low stagnation pressure loss [1e5]. The intrusive
Fig. 2 e Grid independence study: time-averaged static
strut injectors in supersonic combustion studies revealed that
pressure distribution along the bottom. Wall of the
the augmentation in the scramjet engine's performance using
combustor.
hydrogen injection is appreciably noticed [6,7].
The planar (two-dimensional) combustor configuration of
the Institute of Chemical Propulsion, DLR contains a strut to
inject hydrogen in supersonic air co-flow [8,9]. Many re-
searchers have extensively used these experimental data for the baseline strut configuration from DLR. Suneetha et al. [26]
validating numerical investigations [10e19]. Oevermann [20] investigated a diamond-shaped strut injector's performance
performed a computational study on the 2D supersonic in a Mach 2.5 flow field and reported specific inlet conditions
combustor model based on the DLR experiments. The results for optimum combustion efficiency. Donohue et al. [27]
showed that the flow properties are almost comparable with investigated the swept ramp injector performance in a su-
the experimental results. Wu et al. [21] showed good agree- personic flow. The authors noted that the interaction of
ment in 2D and 3D Large Eddy Simulation (LES) results of shock wave and the ramp injection vortices cause separation
shock train and turbulent mixing in DLR scramjet configura- in the shear layer.
tion. Based on the averaged flowfield results, it is concluded Various strut injection strategies and combustor wall
that 2D analysis is sufficient to capture the flow physics. Zha divergence angles by Nithish Reddy and Venkatasubbaiah [28]
et al. [22] computed the increase of turbulence intensity in the reported that the combustor wall divergence for varying inlet
mixing layer under supersonic conditions with the effect of conditions and multi-strut injection in the supersonic
shock waves. combustor plays a significant role in combustion perfor-
Fuel injection with two struts parallel to supersonic mance. Zhang et al. [29e33] performed a detailed numerical
airstream is computationally shown to enhance the mixing and experimental investigations using kerosene fuelled strut
and, therefore, combustion efficiency [23,24]. Also, oblique injector in dual model scramjet combustor. In a strut scramjet
shock waves that emerge from the strut demonstrated to combustor, Ambe Verma et al. [34] used computational
enhance mixing and efficiency in scramjet [25]. Strut rear modeling to investigate flush wall transverse injection at
wall modification studies by Aravind and Rajivkumar [10] various locations as well as parallel hydrogen jet injection. In
revealed that the complete mixing of air and fuel could be comparison to the other wall injection locations, their findings
established with a relatively shorter combustor length than showed that transverse injection at a 68 mm offers a

Fig. 1 e (a) Schematic layout of 2D combustor model (b) details of strut injection locations for three cases.
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 6 ( 2 0 2 1 ) 2 3 0 1 3 e2 3 0 2 5 23015

Table 1 e Inflow conditions of air and fuel.


Variable Air H2
Ma 2.0 1.0
u (m/s) 706 1240
T (K) 340 250
P (bar) 1.0 1.0
r (kg/m3) 1.002 0.097
YO2 0.232 0
YH2 0 1
YH2O 0.032 0
YN2 0.736 0

combustion efficiency of 96%. The numerical investigation on Fig. 4 e Wall static pressure distribution for DLR strut
revolved wedge strut injector in a 2D scramjet combustor by injection of experimental and numerical simulation.
Kumitha et al. [35] reported that the mixing efficiency is
improved by 9% over the basic wedge strut. The effect of
multiport wall injection behind a strut base is numerically
studied by Jiang et al. [36] revealed that the mixing is improved patterns and named as the Koopman operator [39]. DMD
with increase in the injection port distances. The numerical analysis of the numerical results of DLR scramjet showed
investigation [37] on the increase in hydrogen jet pressures in unlikely coupling of pressure oscillations with the flame's
a reacting supersonic field revealed that the hydrogen jet oscillations, which was argued to relate to wake instability
penetration improves in the spatial direction with increase in [40].
total pressure loss. Dynamic analysis requires experimental or numerical ob-
The dynamics of variable density, compressible turbulent servations of well-resolved instantaneous snapshots of the
flows can be investigated using Dynamic Mode Decomposi- flow. LES approach is the one that is a well developed and
tion (DMD) for global mode analysis. Using DMD spatial and feasible method for the aforementioned numerically obser-
temporal dynamics: dominant structures, patterns, and their vation of turbulent flow problems. However, supersonic
corresponding frequencies can be extracted out of visual boundary layer flows still feature a wide range of spatial and
snapshots [38]. This method is equivalent to decomposing a temporal scales to be resolved for viscosity generation. Hence,
nonlinear dynamic system based on a linear operator's high resolution in time and space increases the computational
spectral analysis, which is regarded to span all transient cost while applying LES for the boundary layer domain. DES

Fig. 3 e Shadowgraph images of a single strut injection: (a) Experimental and (b) Numerical results.
23016 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 6 ( 2 0 2 1 ) 2 3 0 1 3 e2 3 0 2 5

Fig. 5 e Density line contours of shockwaves for considered strut injection schemes: (a) case 1 (b) case 2, (c) case 3, and (d)
case 4.

approach is a hybrid unsteady RANS/LES method proposed to injector. This stands as a motivation to take up the current
reduce computational cost [41]. Menter and Kuntz [42] pro- research, which is mainly aiming towards the mixing
posed a blending function for LES calculations without behavior of strut injection in a 2D DLR supersonic
compromising high resolution within the boundary layer to combustor model by shifting the fuel injection location in
prevent grid-related transition. The formulation of the shear the strut for constant hydrogen flow rate conditions. The
stress transport (SST) model was further extended by Spalart strut injection location effect on combustor performance
et al. [41]. This modification was also applied in Delayed De- has been examined, estimating the major critical parame-
tached Eddy Simulations [43]. ters like wall static pressure distribution and total pressure
Hence, most of the strut combustor scramjet studies loss. The results are compared with the baseline strut
were focused on using the central strut with a single configuration.
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 6 ( 2 0 2 1 ) 2 3 0 1 3 e2 3 0 2 5 23017

Fig. 6 e Hydrogen mass fraction contours with flow streamlines at the near jet-wake interaction field for different injection
locations.

URANS method is too dissipative to resolve the shear layer.


Numerical modeling Transient simulations are performed using the Delayed DES
model based on k-ε model [42,48]. Steady RANS solutions are
ramjet model studies is accomplished by solving the two- taken as the initial field, and transient simulations are
dimensional mass, momentum, and energy conservative performed for a 6e-2 physical time before recording snap-
equations, as well known reported [26]. Appropriate turbulence shots of velocity, density, and hydrogen mass fraction. This
modeling is the most crucial task in the supersonic flow field to interval corresponds to more than ten flowthrough times of
observe the fundamental flow structure [44]. In the current inflowing air to ensure steady statistics. A thousand snap-
approach, the compressible Reynolds Averaged Navier-Stokes shots of instantaneous hydrogen mass fraction field are
equations (RANS) are solved by applying k-u SST turbulence recorded with a time interval of 1e-5 s for dynamic analysis.
model. Menter [45] substantiated that the k-u SST turbulence Grid adaptation is switched off when recording instanta-
model is most appropriate for scramjet applications. The k-u neous snapshots to prevent miscalculation of dynamic
SST turbulence model can predict the mixing layers, adverse modes.
pressure gradients, and separated flows, precisely as mentioned
[46]. The flow governing equations are discretized by the finite
volume method framework of ANSYSFluent [10]. The density- Numerical setup
based solver with the implicit formulation and advection up-
stream splitting methods are adopted in this work [47]. Computational domain
Two alternative methods for unsteady calculations are
the Unsteady Reynolds Averaged Navier-Stokes (URANS) The scramjet combustor model adopted for the present study
method and the LES method. Initially, we have tested both is tested at the German Aerospace Center (DLR) [8,9], as shown
of these methods for the one with identical configurations.
23018 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 6 ( 2 0 2 1 ) 2 3 0 1 3 e2 3 0 2 5

Fig. 7 e Mach number contours for different injection locations.

in Fig. 1. The combustor is 340 mm in length and 50 mm in and case 3. In case 4, the injection locations for both case 2 and
height. The upper wall diverges to an angle of 3 over the case 3 are simultaneously operational for identical operating
distance of 100 from the combustor inlet. The strut is located conditions.
at a distance of 77 mm from the combustor inlet. The length The compressible shear layer's growth rate with a density
and height of the strut are 32 mm and 6 mm, respectively. The gradient can be related to the convective Mach number,
incoming air at the combustor inlet is at Mach 2.0, and the following the work by Papamoschou and Roshko [49]. The
hydrogen fuel is injected parallel to the airstream at the sonic convective Mach number of the shear layer is 1.22 with a
condition. The non-reacting mixing flow is considered convective velocity of 274.7 m/s.
compressible and two-dimensional. Hydrogen is injected at
two various strut locations to analyze the mixing performance Boundary condition
under non-reacting conditions. Then results are compared
with the tested baseline cavity configurations. The details of The proper implementation of the boundary conditions is
the injection point locations are shown in Fig. 1b. Fig. 1(b) essential to define the computational fluid dynamics physics
demonstrates the distinct injection locations for computa- problems. The inlet conditions for the fuel injection and
tions which refer as case 1 (baseline strut injection), case 2, combustion process are listed in Table 1.
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 6 ( 2 0 2 1 ) 2 3 0 1 3 e2 3 0 2 5 23019

In both steady and transient simulations, Dirichlet inlet


boundary conditions are used for the inflow, whereas the
outlet is based on Neuman-type boundary conditions. No-slip
condition is adopted along the walls of the combustor. In the
steady-state calculations, the turbulence intensity is consid-
ered less than 10% at the inlet, and the hydraulic diameter is
selected accordingly for initialization. In transient simula-
tions, the air inlet and the hydrogen jets turbulence intensity
are taken as 0.5% and 5%, respectively, as reported in the ex-
periments [50].

Grid generation

In the steady-state analysis, the unstructured grid is


employed to resolve the strut injection combustor flow field.
Three different grid sizes are adopted to optimize the grid
resolution, so the numerical result accuracy is improved along
with reducing computational cost and time. The grid sizes,
namely coarse (146,146), medium (195,950), and fine (290,112)
meshes, are considered for grid convergence study. The
y þ value is smaller than 1.0 for the entire flow field, which
corresponds to the height for the first-row cell fixed at 1e-
3 mm. The wall static pressure variation is calculated based on
three meshes that denote the steady-state solution's grid in-
dependence. The convergence is observed since the static
pressure values provide less than 1% variation for all the mesh
sizes (see Fig. 2). Therefore, medium-sized meshes are
considered for this study to reduce the computational time.
Transient simulations are initiated from the results of
steady-state simulations. The unsteady flow field is computed in
1e-3 (physical time) seconds before it reaches quasi-steady-state
dynamic analysis. Meanwhile, adaptive mesh, which is refined
using up to 3e5 mesh nodes, provides a high resolution of flow
structures. The grid adaptation is accounted for both mixing
Fig. 8 e Wall static pressure distribution of various strut
layer and shock train, and therefore refinement parameter is
injection locations (a) bottom wall (b) centerline of the
chosen as a density gradient with a threshold of 1e5.
supersonic combustor.
The solution is progressed with time at a fixed time step of
1e-6 s while taking the snapshots of every 1e-5 s for dynamic
analysis. The unstructured mesh provides excellent flexibility
in handling boundary conditions with complex geometry and
performing grid adaptation in unsteady simulation using LES,
unsteady, unstructured with DES [16]. Adaptive refinement
results in a uniform mesh at locations where coherent struc-
tures exists in the flow. As a result, deviation of dynamics
because of the numerical errors which emanate from grid ir-
regularities are reduced in coherent flow structure zones. The
accumulating numerical errors due to the irregularities in
mesh size and rounding known to reduce the accuracy of
Large Eddy Simulations [51]. These errors known to cause
accumulation of energy in the refined side [52]. The flow field
zones with the dynamic coherent structures are resolved
using uniform grid to reduce the error in modal analysis
results.

Validation

Fig. 9 e Total pressure loss across various axial length of The steady-state simulation results are qualitatively validated
the combustor. against the DLR experiments that were reported by Waidmann
et al. [8,9], as represented in Figs. 3 and 4. It is confirmed that
the shock generated from the strut and the reflected shock
23020 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 6 ( 2 0 2 1 ) 2 3 0 1 3 e2 3 0 2 5

Fig. 10 e Density contours for Case 1 using (a) URANS method and (b) DDES method.

waves from the wall and the wall static pressure distribution is fuel injection stream. Moreover, the shock train in the strut-
in good agreement with the experimental results. The marginal downstream also weakens, which leads to a reduction in the
variation in the static pressure distribution could be due to the flow field's static pressure.
unpredicted turbulence vortices near the wall. Flow streamlines and hydrogen mass fraction contours at
the strut's near wake region for various injection locations are
illustrated in Fig. 6. It is noticed that the hydrogen jet
Results and discussion streamlines are formed along with the jet injection location.
Low velocity (wake) regions are observed on both sides for
The mixing performance for different strut injection strategies case 1, whereas the wake regions are emerged out at the
in the DLR scramjet configuration is numerically investigated alternative sides of the hydrogen injection locations for case 2
using the CFD analysis. The focus of this work is the perfor- and case 3. In case 4, the wake regions are located at the
mance of strut-type scramjet for various injection locations. middle of the strut. Hence, the hydrogen injection induces a
The RANS equations are solved for steady-state flow at iden- wake region in the vicinity of the flow. Similar trends are
tical boundary conditions and operating variables. Unsteady substantiated in Mach number contours, as distinguished
computations are based on DDES approach. from Fig. 7 for the four cases.
Hydrogen injection in the shock train is investigated to Particularly in baseline case 1, the airflow accelerates,
reveal mixing features and total pressure loss across the reaching a maximum Mach number of 2, and subsequently
scramjet under non-reacting flow fields. Fig. 5 represents the becomes subsonic due to the formation of wake regions on
density contour images for the various strut injection both sides. In other cases, the airflow is limited to around
configurations. Mach 1.5. Hence, case 1 provides competent mixing effi-
It is observed from Fig. 5a that strong oblique shock waves ciency compared with the other three cases by observing the
are formed from the leading edges of the strut, which again gets mixing length of hydrogen mass fraction contours, as
reflected from the combustor wall. The reflected shock waves noticed from Fig. 6.
are deflected at various locations from the strut downstream.
The shock waves generated at the strut's trailing edges interfere Wall static pressure
with the shock reflected on the combustor wall and further
weaken towards the combustor's bottom wall vicinity. A similar The static pressure distribution along the bottom wall and the
trend exists in cases of 2 and 3 (Fig. 5b and c) with marginal combustor's centerline for different injection cases are pre-
variation in the injection locations vicinity. sented in Fig. 8. As shown in Fig. 8(a), the static pressure dis-
In the case of dual injection, Fig. 5d, i.e., case 4, the shock tribution is almost uniform at the inlet along the combustor
waves generated from the trailing edges are dissipated by the axis. Shock waves emanate from the strut leading/trailing
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 6 ( 2 0 2 1 ) 2 3 0 1 3 e2 3 0 2 5 23021

Fig. 11 e Global mode analysis results for density contours of the transient simulation of case 1.

edges and also the reflected shocks from the wall. Consecutive
ð
surge and plunge train in the static pressure creates respective
Po ru dA
compressive and expansion regions in the scramjet. Moreover,
the wall pressure distribution for all the strut injection stra- ht ¼ 1  ðA
tegies shows a marginal variation in the profile along the Poinl ru dA
combustor's axial length. A

From Fig. 8b, the static pressure is relatively more for the Fig. 9 shows the total pressure loss across the scramjet
distance, x ¼ 0.109 m at the middle plane of the combustor for the combustor for the various injection strategies. Case 4 shows
strut injection case 1. Higher static pressure is occurred due to an increment in the total pressure loss from the injection
the injection orifice: the center of the strut exit plane. The in- location downstream, mainly due to the strong shock wave
jection orifice is located away from the center planes for other emerging from the dual injection strategies. However, the
injection cases, which creates a vortex around the center plane injection strategies of case 2 and case 3 provide marginal
of the strut exit that essentially induces a flow static pressure variation in the total pressure loss as compared with baseline
region. strut injection.

Total pressure loss Transient solution

In general, the mixing enhancement due to shock waves in- Two alternative methods that are tested for the transient
duces total pressure loss. Pressure loss across the combustor simulations are Unsteady RANS and DES models. Both of
is calculated using the following expression [53]. these models are implemented for the Case 1 with a second-
23022 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 6 ( 2 0 2 1 ) 2 3 0 1 3 e2 3 0 2 5

Fig. 12 e Global mode analysis results for density contours of the transient flowfields computed.
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 6 ( 2 0 2 1 ) 2 3 0 1 3 e2 3 0 2 5 23023

order implicit bounded time-stepping method. DDES is Fig. 11. Eigenvalues of the companion matrix are arranged on
applied by using the blending function model constants of the unity circle or inside, which indicates that the system is
Cdes1 ¼ 0.78 and Cdes2 ¼ 0.61 [54]. Instantaneous density in a quasi-equilibrium (non-amplification) state. The Eigen
contours of transient simulations are shown in Fig. 10. field, which is the reconstructed eigenvectors, corresponding
The location and the patterns of the shock train and the to 12,835 Hz relates to the upper diverging wall boundary
jet's penetration are qualitatively observed as closer results for layer.
both transient simulations. However, the URANS result does The coherent flow structures are noticed along the upper
not resolve the unsteady turbulent structures, notably the wall that are prominent after 0.2 m. The flow patterns
transition on the upper divergence boundary layer. Conse- observed in the lower dominant frequency (8410 Hz) are
quently, the DDES method is chosen to generate unsteady identified to represent the mixing layer. The flow fluctuations
snapshots for dynamic analysis. arising from the jet interacting with the strut wake shear flow
and the boundary layer transition along the diverging upper
Dynamic analysis wall are decoupled.
To examine the dynamic characteristics of the hydrogen
The well-defined DMD method is used to investigate the mixing and to compare for four cases, the DMD analysis re-
flowfield dynamics resolved numerically and validated by sults are shown in Fig. 12. Almost all eigenvalues lay on a unit
comparing with the DLR experiment. circle for all four cases, which infer the mixing layer's quasi-
To adopt global mode analysis, we commence with a steadiness. There exist two dominant DMD modes for Case 2
sequence of (nþ1) solution fields unrolled in a vector uj with a and Case 3. The scales of the upper and lower flow structures,
constant time increment Dt. Two sets of the flowfield's tran- that arise from strut wake and jet interaction, differentiate
sient solutions are merged into matrices X1 and X2 using n with the span between strut corner and jet.
consecutive snapshots as DMD mode contours correspond to the dominant fre-
quency in the low range, which relate the strut Karman wake
X1 ¼ u1 ; u2 ; u3 ; …:; un and jet interaction. The injection location is observed that it
could alter the frequency of oscillations and thereby the
X2 ¼ u2 ; u3 ; u4 ; …:; uðnþ1Þ mixing characteristics. Hence, varying the injection port may
The basic hypotheses of modal analysis assume a linear provide dynamic control of mixing in the supersonic flow
operator A for a transient system such that uiþ1 ¼ A ui, named field.
the Koopman operator. This operator delineates the spatial and
dynamic characteristics by taking into account its eigenvalues
and eigenvectors. A companion matrix K, which is similar to the Conclusion
Koopman operator that can be obtained using X1 and X2, such as
A two-dimensional DLR scramjet model comprising various
X2 y X1 K strut injection locations is numerically investigated in a non-
reacting flow field using RANS equations. The injection loca-
The companion matrix K can be obtained first taking the
tions' effectiveness are analyzed based on the parameters like
singular value decomposition of X1
density contours, static pressure distribution, and total pres-
X1 ¼ USV sure loss across the combustor. It is confirmed that the adop-
ted numerical approach investigation is complimented for the
Square matrix S (n x n) yields the singular values of X1.
experimental DLR scramjet test results. It is noticed from the
After that, the eigenvalues and eigenvectors of the dynamic
density contour that the hydrogen injection from two orifices
system is calculated by decomposing the square matrix;
with identical mass flow rate weakens the oblique shock
~ ¼ UT X2 VS1
A waves formed downstream of the strut, which further reduces
the static pressure at the middle plane of the combustor than
The eigenvalues and eigenvectors characterize the linear- the single orifice injection. The simultaneous twin orifice
ized approximation and form DMD modes of the system. The hydrogen injection from the same plane provides about uni-
corresponding Koopman (DMD) modes which correspond to form wall static pressure profile and stagnation pressure loss
wave numbers li are given by argðli Þ=Dt. across the combustor than the single orifice injection config-
This analysis is susceptible due to the quasi-steady nature uration. Both URANS and DDES methods are applied for the
of the solution. Once the system is initiated for a steady-state base configuration (DLR experiment) by considering transient
solution, the snapshot capturing is delayed for more than ten simulations. The results of URANS method shown that it does
flowthrough periods. Later, 1000 instantaneous snapshots are not resolve the transition region of the upper boundary.
recorded. The convergence of the energy spectrum is achieved Therefore DDES method is chosen to analyze the dynamics of
for each case after ten flowthrough times to reach a stationary the four injection strategies. Results of Case 3 and Case 4 yield
solution. two dominant modes, and Case 4 has four modes. The domi-
The transient density fields are taken by subjecting into nant frequency in Case 3 and Case 4, which correspond to the
DMD analysis since one can evaluate coupled flow physics of Karman wake, is about 5000 Hz and the higher dominant fre-
jet mixing with density difference and flow pressure oscil- quency is about 12,500 Hz for both cases. Further investigation
lations. For Case 1, the transient density field eigenvalues, on the effectiveness of the injection strategies using reacting
spectrum, and two dominant DMD modes are shown in flows is considered in future work.
23024 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 6 ( 2 0 2 1 ) 2 3 0 1 3 e2 3 0 2 5

[15] Kummitha OR. Numerical analysis of passive techniques for


Declaration of competing interest optimizing the performance of scramjet combustor. Int J
Hydrogen Energy 2017;42:10455e65. https://doi.org/10.1016/
The authors declare that they have no known competing j.ijhydene.2017.01.148.
financial interests or personal relationships that could have [16] Kummitha OR, Pandey KM, Gupta R. Numerical investigation
of wavy wall strut fuel injector for hydrogen fueled scramjet
appeared to influence the work reported in this paper.
combustor. Int J Hydrogen Energy 2019;44:32240e53. https://
doi.org/10.1016/j.ijhydene.2019.10.147.
[17] Kummitha OR, Pandey KM, Gupta R. Numerical analysis of
references
hydrogen fueled scramjet combustor with innovative
designs of strut injector. Int J Hydrogen Energy
2020;45:13659e71. https://doi.org/10.1016/
[1] Ben-Yakar A, Hanson RK. Cavity flame-holders for ignition j.ijhydene.2018.04.067.
and flame stabilization in scramjets: an overview. J Propul [18] Ma S, Zhong F, Zhang X. Numerical study on supersonic
Power 2001;17:869e77. https://doi.org/10.2514/2.5818. combustion of hydrogen and its mixture with Ethylene and
[2] Biling FS. Research on supersonic combustion. J Propul Power methane with strut injection. Int J Hydrogen Energy
1993;9:499e514. https://doi.org/10.2514/6.1992-1. 2018;43:7591e9. https://doi.org/10.1016/
[3] Sun C, Barzegar Gerdroodbary M, Abazari AM, Hosseini S, j.ijhydene.2018.03.007.
Li Z. Mixing efficiency of hydrogen multijet through [19] Athithan AA, Jeyakumar S, Sczygiol N, Urbanski M. The
backward-facing steps at supersonic flow. Int J Hydrogen combustion characteristics of double ramps in a strut-based
Energy 2021. https://doi.org/10.1016/j.ijhydene.2021.02.030. scramjet combustor. Energies 2021;14:831. https://doi.org/
[4] Suneetha L, Randive P, Pandey KM. Implication of diamond 10.3390/en14040831.
shaped dual strut on combustion characteristics in a cavity- [20] Oevermann M. Numerical investigation of turbulent
based scramjet combustor. Int J Hydrogen Energy hydrogen combustion in a SCRAMJET using flamelet
2020;45:17562e74. https://doi.org/10.1016/ modeling. Aero Sci Technol 2000;4:463e80. https://doi.org/
j.ijhydene.2020.04.217. 10.1016/S1270-9638(00)01070-1.
[5] Liu X, Barzegar Gerdroodbary M, Sheikholeslami M, Moradi R, [21] Wu K, Zhang P, Yao W, Fan X. Numerical investigation on
Shafee A, Li Z. Effect of strut angle on performance of flame stabilization in DLR hydrogen supersonic combustor
hydrogen multi-jets inside the cavity at combustion with strut injection. Combust Sci Technol
chamber. Int J Hydrogen Energy 2020;45:31179e87. https:// 2017;189:2154e79. https://doi.org/10.1080/
doi.org/10.1016/j.ijhydene.2020.08.124. 00102202.2017.1365847.
[6] Gerlinger P, Br-uacute D, Ggemann. Numerical investigation [22] Zha Z, Ye Z, Hong Z, Ye K. Effects of unsteady oblique shock
of hydrogen strut injections into supersonic airflows. J Propul wave on mixing efficiency of two-dimensional supersonic
Power 2000;16:22e8. https://doi.org/10.2514/2.5559. mixing layer. Acta Astronaut 2021;178:60e71. https://doi.org/
[7] Masuya G, Tomoyuki T, Murakami A, Shinozaki N, Nakamura A, 10.1016/j.actaastro.2020.07.028.
Murayamall M, et al. Ignition and combustion performance of [23] Choubey G, Pandey KM. Effect of variation of angle of attack
scramjet combustors with fuel injection struts. J Propul Power on the performance of two-strut scramjet combustor. Int J
1995;11:301e7. https://doi.org/10.2514/3.51425. Hydrogen Energy 2016;41:11455e70. https://doi.org/10.1016/
[8] Guerra R, Waidmann W, Laible C. An experimental j.ijhydene.2016.04.048.
investigation of the combustion of a hydrogen jet injected [24] Choubey G, Pandey KM. Effect of parametric variation of
parallel in a supersonic air stream. AIAA Papp 1991. https:// strut layout and position on the performance of a typical
doi.org/10.2514/6.1991-5102. 91e5102. two-strut based scramjet combustor. Int J Hydrogen Energy
[9] Waidmann W, Alff F, Brummund U, Bohm M, Clauss W, 2017;42:10485e500. https://doi.org/10.1016/
Oschwald M. Experimental investigation of the combustion j.ijhydene.2017.03.014.
process in a supersonic combustion ramjet (scramjet). [25] Xue R, Wei X, He G, Hu C, Tang X. Effect of parallel-jet
Hardthausen, Ger DLR Inst Chem 470 Propuls Eng addition on the shock train characteristics in a central-strut
1994:629e38. isolator by detached eddy simulation. Int J Heat Mass Tran
[10] Aravind S, Kumar R. Supersonic combustion of hydrogen 2017;114:1159e68. https://doi.org/10.1016/
using an improved strut injection scheme. Int J Hydrogen j.ijheatmasstransfer.2017.06.074.
Energy 2019;44:6257e70. https://doi.org/10.1016/ [26] Suneetha L, Randive P, Pandey KM. Numerical investigation
j.ijhydene.2019.01.064. on influence of diamond shaped strut on the performance of
[11] Ge nin F, Menon S. Simulation of turbulent mixing behind a a scramjet combustor. Int J Hydrogen Energy
strut injector in supersonic flow. AIAA J 2010;48:526e39. 2019;44:6949e64. https://doi.org/10.1016/
https://doi.org/10.2514/1.43647. j.ijhydene.2019.01.187.
[12] Wu K, Zhang P, Yao W, Fan X. Computational realization of [27] Donohue JM, McDaniel JC, Haj-Hariri H. Experimental and
multiple flame stabilization modes in DLR strut-injection numerical study of swept ramp injection into a supersonic
hydrogen supersonic combustor. Proc Combust Inst flowfield. AIAA J 1994;32:1860e7. https://doi.org/10.2514/
2019;37:3685e92. https://doi.org/10.1016/j.proci.2018.07.097. 3.12184.
[13] Huang Z wei, He G qiang, Wang S, Qin F, Wei X geng, Shi L. [28] Nithish Reddy P, Venkatasubbaiah K. Numerical
Simulations of combustion oscillation and flame dynamics investigations on development of scramjet combustor. J Aero
in a strut-based supersonic combustor. Int J Hydrogen Eng 2015;28:1e8. https://doi.org/10.1061/(ASCE)AS.1943-
Energy 2017;42:8278e87. https://doi.org/10.1016/ 5525.0000456.
j.ijhydene.2016.12.142. [29] Zhang J, Chang J, Kong C, Qiu H, Bao W. Flame oscillation
[14] Kumaran K, Behera PR, Babu V. Numerical investigation of characteristics in a kerosene fueled dual mode combustor
the supersonic combustion of kerosene in a strut-based equipped with thin strut flameholder. Acta Astronaut
combustor. J Propul Power 2010;26:1084e91. https://doi.org/ 2019;161:222e33. https://doi.org/10.1016/
10.2514/1.46965. j.actaastro.2019.05.037.
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 6 ( 2 0 2 1 ) 2 3 0 1 3 e2 3 0 2 5 23025

[30] Zhang J, Chang J, Wang Z, Gao L, Bao W. Flame propagation [42] Menter FR, Kuntz M. Adaptation of eddy-viscosity turbulence
and flashback characteristics in a kerosene fueled models to unsteady separated flow behind vehicles. In:
supersonic combustor equipped with strut/wall combined McCallen R, Browand F, Ross J, editors. Symp “the aerodyn
fuel injectors. Aero Sci Technol 2019;93:105303. https:// heavy veh truck buses trains” monterey, USA, 2e6 Dec 2002.
doi.org/10.1016/j.ast.2019.105303. Berlin Heidelb New York: Springer; 2004. p. 339e52. https://
[31] Zhang J, Chang J, Quan F, Bian L, Bao W. Ignition doi.org/10.1007/978-3-540-44419-0_30.
characteristics in a thin strut-equipped dual mode [43] Spalart PR, Deck S, Shur ML, Squires KD, Strelets MK,
combustor fueled with liquid kerosene. Acta Astronaut Travin A. A new version of detached-eddy simulation,
2019;161:125e38. https://doi.org/10.1016/ resistant to ambiguous grid densities. Theor Comput Fluid
j.actaastro.2019.05.013. Dynam 2006;20:181e95. https://doi.org/10.1007/s00162-006-
[32] Zhang J, Chang J, Ma J, Wang Y, Bao W. Investigations on 0015-0.
flame liftoff characteristics in liquid-kerosene fueled [44] Zhao M, Ye T. URANS study of pulsed hydrogen jet
supersonic combustor equipped with thin strut. Aero Sci characteristics and mixing enhancement in supersonic
Technol 2019;84:686e97. https://doi.org/10.1016/ crossflow. Int J Hydrogen Energy 2019;44:20493e503. https://
j.ast.2018.11.017. doi.org/10.1016/j.ijhydene.2019.06.027.
[33] Chang J, Zhang J, Bao W, Yu D. Research progress on strut- [45] Menter FR. Two-equation eddy-viscosity turbulence models
equipped supersonic combustors for scramjet application. for engineering applications. AIAA J 1994;32:1598e605.
Prog Aero Sci 2018;103:1e30. https://doi.org/10.1016/ https://doi.org/10.2514/3.12149.
j.paerosci.2018.10.002. [46] Li C, Chen X, Li Y, Musa O, Zhu L, Li W. Role of the backward-
[34] Ambe Verma K, Murari Pandey K, Ray M, Kumar Sharma K. facing steps at two struts on mixing and combustion
Effect of transverse fuel injection system on combustion characteristics in a typical strut-based scramjet with
efficiency in scramjet combustor. Energy 2021;218. https:// hydrogen fuel. Int J Hydrogen Energy 2019;44:28371e87.
doi.org/10.1016/j.energy.2020.119511. https://doi.org/10.1016/j.ijhydene.2019.09.023.
[35] Kummitha OR, Pandey KM, Padidam AKR. Effect of a [47] Choubey G, Pandey KM. Effect of variation of inlet boundary
revolved wedge strut induced mixing enhancement for a conditions on the combustion flow-field of a typical double
hydrogen fueled scramjet combustor. Int J Hydrogen Energy cavity scramjet combustor. Int J Hydrogen Energy
2021;46:13340e52. https://doi.org/10.1016/ 2018;43:8139e51. https://doi.org/10.1016/
j.ijhydene.2021.01.089. j.ijhydene.2018.03.062.
[36] Jiang Y, Barzegar Gerdroodbary M, Sheikholeslami M, [48] Shur M, Spalart PR, Strelets M, Travin A. Detached-eddy
Babazadeh H, Shafee A, Moradi R, et al. Influence of simulation of an airfoil at high angle of attack. 1999. https://
upstream strut on hydrogen fuel distribution inside the doi.org/10.1016/b978-008043328-8/50064-3.
supersonic combustion chamber. Int J Hydrogen Energy [49] Papamoschou D, Roshko A. The compressible turbulent
2020;45:22032e40. https://doi.org/10.1016/ shear layer: an experimental study. J Fluid Mech
j.ijhydene.2020.06.026. 1988;197:453e77. https://doi.org/10.1017/S0022112088003325.
[37] Jeyakumar S, Jayaraman K, Sivakumar R, Karaca M, K K. [50] Ingenito a. Theoretical investigation of air vitiation effects on
Numerical investigations on the hydrogen jet pressure hydrogen fuelled scramjet performance. Int J Hydrogen
variations in a strut based scramjet combustor. Aircraft Eng Energy 2015;40:2862e70. https://doi.org/10.1016/
Aero Technol 2021. https://doi.org/10.1108/AEAT-08-2020- j.ijhydene.2014.12.014.
0162. ahead-of-p. [51] Poinsot T, Garcia M, Senoner JM, Gicquel L, Staffelbach G,
[38] Schmid PJ. Dynamic mode decomposition of numerical and Vermorel O. Numerical and physical instabilities
experimental data. J Fluid Mech 2010;656:5e28. https:// in massively parallel les of reacting flows. J Sci Comput
doi.org/10.1017/S0022112010001217. 2011;49:78e93. https://doi.org/10.1007/s10915-010-9432-8.
[39] Rowley CW, Mezi I, Bagheri S, Schlatter P, Henningson DS. [52] Goodfriend E, Katopodes Chow F, Vanella M, Balaras E. Large-
Spectral analysis of nonlinear flows. J Fluid Mech Eddy simulation of flow through an array of cubes with local
2009;641:115e27. https://doi.org/10.1017/ grid refinement. Boundary-Layer Meteorol 2016;159:285e303.
S0022112009992059. https://doi.org/10.1007/s10546-016-0128-y.
[40] Li Q, Wang Z. Dynamic mode decomposition of turbulent [53] Gerlinger P, Stoll P, Kindler M, Schneider F, Aigner M.
combustion process in DLR scramjet combustor. J Aero Eng Numerical investigation of mixing and combustion
2017;30:1e10. https://doi.org/10.1061/(ASCE)AS.1943- enhancement in supersonic combustors by strut induced
5525.0000747. streamwise vorticity. Aero Sci Technol 2008;12:159e68.
[41] Spalart PR, Jou WH, Strelets M, Allmaras S. Comments on the https://doi.org/10.1016/j.ast.2007.04.003.
feas1ibility of LES for wings, and on a hybrid RANS/LES [54] Strelets M. Detached eddy simulation of massively separated
approach. Proc first AFOSR int conf DNS/LES, Ruston. flows. 39th Aerosp Sci Meet Exhib 2001. https://doi.org/
Louisiana Greyden Press; 1997. p. 137e47. 10.2514/6.2001-879.

You might also like