Aerospace Science and Technology 117 (2021) 106969

Contents lists available at ScienceDirect

Aerospace Science and Technology

www.elsevier.com/locate/aescte

Shock dynamics and expansion characteristics of an aerospike nozzle

and its interaction with the rotating detonation combustor
Yue Huang ∗ , Hanqing Xia, Xiaonan Chen, Zhenye Luan, Yancheng You
School of Aerospace Engineering, Xiamen University, Xiamen 361005, China

a r t i c l e i n f o a b s t r a c t

Article history: The flow of products through the aft-end of a detonation combustor is intrinsically unsteady, ultimately
Received 10 March 2021 resulting in periodic and nonuniform flow parameters at the exit plane of a rotating detonation
Received in revised form 18 June 2021 combustor. Therefore, the integrated design of the nozzle and combustion chamber is an important
Accepted 12 July 2021
challenge in rotating detonation engines (RDEs). This paper describes a thorough three-dimensional
Available online 19 July 2021
Communicated by Yu Lv
numerical simulation of a hydrogen–air rotating detonation combustor with/without an aerospike nozzle
and investigates the nozzle flow without a combustor. The simulation results allow us to analyze the
Keywords: coupling characteristics of the rotating detonation combustor and the aerospike nozzle. The new features
Rotating detonation of the shock wave dynamics and expansion characteristics in the nozzle with/without an integrated
Aerospike nozzle RDE combustor are compared. For the same rotating detonation combustor exit, the propagation mode
Spiral oblique shock wave of the detonation wave and backpressure in the RDE combustor exit are affected by the aerospike
Expansion characteristics nozzle. The intensity of the shock wave and the angle with respect to the direction of the air flow
Numerical simulation
are lower with the aerospike nozzle than in the nozzleless model, which is consistent with the effect
of backpressure increase on the shock dynamics. Regardless of whether the nozzle is connected to
the combustor, the nozzle affects the dynamic alternation of the design point state to over-expansion
and under-expansion along the circumferential direction. The shock wave intensity and total pressure
recovery inside the nozzle with the combustor are slightly stronger than those in the nozzle without
the combustor. Consequently, the thrust and total pressure recovery of the nozzle integrated with the
RDE combustor are enhanced by 7% and 5.6%, respectively. Thus, decoupling the nozzle from the rotating
detonation combustor is feasible for testing the design performance of the unsteady nozzle.
© 2021 Elsevier Masson SAS. All rights reserved.

1. Introduction from quasi-constant volume combustion with a significant pressure

gain [8]. The problem is how to effectively convert the substantial
Rotating detonation engines (RDEs) are viable pressure-gain pressure gain generated by detonation events into work output.
thrust-producing devices for a wide range of propulsion systems. A suitable nozzle is able to harness the additional work poten-
In an RDE, one or more rotating detonation waves propagate con- tial afforded by the pressure-gain detonation events [9,10], and
tinuously along the circumference of a combustor (annular or hol- so nozzle design is an important challenge in the development of
low) at a high speed of several kilometers per second [1–6]. One an RDE. Compared with the quasi-steady-state parameter distribu-
or more continuously traveling detonation waves cause the com- tion at the combustor exit plane of traditional constant-pressure
bustion of propellants in the kilohertz regime, which produces combustion propulsion systems, flow parameters such as the total
a stagnation pressure rise along the combustor [7]. Due to the temperature and pressure at the RDE exit plane vary as a func-
very nature of RDEs, the products flowing through the aft-end tion of time and space, spiking suddenly and then trailing off [11].
of the combustor are intrinsically unsteady, ultimately resulting The RDE nozzle design pressure ratio must thus be optimized ac-
in nonuniform and periodic flow parameters at the inlet plane of cording to the profile of the unsteady flow parameters, and the
the nozzle. Compared with traditional constant-pressure combus- nozzle contour should be designed to adjust over a wide range of
tion propulsion systems, detonation-based engines offer theoret- pressure ratios with an appropriate efficiency. In fact, the struc-
ically higher cycle thermal efficiency gains through heat release tures and propagation modes of rotating detonation have often
been investigated considerable in previous literatures [11–19], and
the emphasis about nozzle has always been placed on the different
* Corresponding author. nozzles geometry to improve the performance of the combustion
E-mail address: [email protected] (Y. Huang). chamber [20–24].

https://doi.org/10.1016/j.ast.2021.106969
1270-9638/© 2021 Elsevier Masson SAS. All rights reserved.
Y. Huang, H. Xia, X. Chen et al. Aerospace Science and Technology 117 (2021) 106969

There have been several experiments or numerical studies on

the effect of nozzle configuration or comparisons with nozzleless
cases. Rankin et al. [25] took pressure measurements that show
the exhaust flow is periodic from the nozzle inlet to the throat
and nonperiodic near the nozzle exit, with the combination of a
conical center-body and a converging–diverging nozzle reducing
the pressure oscillations. Fotia et al. [26–28] experimentally found
that aerospike nozzles performed better than various bluff body
designs, and stated that a choked aerospike gave better propulsive
performance than an unchoked aerospike nozzle. There is no sig-
nificant difference between the conventional converging–diverging
nozzle and an aerospike plug-nozzle with the same outer-body Fig. 1. Schematic diagram of RDE combustor with aerospike nozzle.
length, although an aerospike nozzle with a shorter outer-body
clearly produces higher thrust than either of the longer outer- with the oblique shock wave. Due to the faster exhaust velocity
body configurations [27]. The production of thrust from a choked created by the nozzle throat, choked nozzles achieve higher per-
aerospike nozzle decreases as the nozzle internal expansion area formance than open nozzles [37]. A zero-dimensional analysis by
ratio increases, and the truncation of the aerospike expansion sur- Stechmann et al. [38] found that aerospike nozzles can naturally
face could lead to a reduction in the efficiency of thrust pro- adapt to different pressure ratios in the RDE cycle, leading to a spe-
duction [28]. Zhu et al. [29] experimentally found that the peak cific impulse increase of 2–6% over nozzleless cases. A computa-
pressures of rotating detonation combustor with aerospike nozzles tional study by Harroun et al. [39] found that RDEs with internal–
were higher compared to that of nozzleless case. A smaller throat external aerospike nozzles had a specific impulse some 16% higher
area will produce stronger reflected shock waves, which will affect than nozzleless cases, and stated that a multi-wave detonation
the propagation stability of the detonation wave. A rotating deto- head structure could make the exhaust plume more homogeneous.
nation rocket engine with various convergent nozzle designs under Subsequent experiments by Harroun et al. [40] agreed well with
different flow conditions was experimentally investigated by Ben- the numerical conclusions for the internal–external aerospike noz-
newitz et al. [30]. They found that the measured thrust does not zle and nozzleless designs.
increase with an increase in the contraction ratio of the nozzle, The above summary indicates that numerous analytic, experi-
mainly due to a greater presence of counter-propagating flow in mental, and numerical studies have been conducted to investigate
the case of high contraction ratios. Goto et al. [31] experimentally the performance and efficiency of RDEs. The structures of the ro-
found that the combustor pressure was almost proportional to the tating detonation can be determined by the geometric parameters,
nozzle throat mass flux. An experimental study by Wang et al. [22] reactive kinetics, reactant fill heights, and injection methods of
using a Laval nozzle with an RDE found that the detonation mode the RDE combustor. These efforts make it important to understand
was highly dependent on the equivalence ratio and throat width. how the unsteady pressure spikes of detonation products may ho-
In addition, hollow chamber RDEs with Laval nozzles have been mogenize and expand along the downstream nozzle, determine the
extensively studied [23,24]. effect of the nozzle type on the detonation initiation and traveling
A numerical study by Yi et al. [20] showed that an aerospike modes and RDE performance, and clarify how the nozzle design
nozzle achieves better performance than a straight extended noz- could be decoupled from the RDE combustor operating conditions.
zle and variable channel with divergent nozzles. The aerospike In the present study, a thorough numerical investigation of RDE
nozzle angle and length have a negligible effect on the propulsive models with/without an aerospike plug nozzle is carried out to
performance, as the nozzle is only subjected to a spiral expansion analyze the coupling characteristics between the RDE combustion
flow. Shao et al. [21] numerical displayed that the Laval nozzle chamber and the nozzle. Based on the time-averaged parameters
gives better performance than a constant-area nozzle, diverging of the nozzleless RDE combustion chamber outlet, a plug noz-
nozzle, or converging nozzle. Braun et al. [32,33] numerical found zle design method is proposed and several new features of the
that the divergent and convergent–divergent nozzles both damp shock wave dynamics and expansion characteristics in the nozzle
out flow fluctuations at the RDE combustor exit while maintaining with/without an RDE combustor are compared. The changes in det-
a reasonable pressure loss, and the smooth divergent profile was onation modes caused by the RDE nozzle and the expansion state
assessed to be the best design [32]. The nozzles produce a signifi- change in the aerospike nozzle are discussed in detail.
cant total pressure loss compared with nozzleless RDE combustors,
which is essentially made up of the viscous loss and spiral oblique 2. Methodology
shock [33]. Most studies have shown that aerospike nozzle is a po-
tentially better choice for a wide operation range of RDEs. 2.1. Physical model
Schwer et al. [34] and Schnabel et al. [35] focused on the
dynamic flow characteristics and pressure distribution of an RDE The RDE model shown in Fig. 1 has three different parts,
aerospike nozzle through three-dimensional unsteady simulations. namely an annular combustion chamber, aerospike nozzle, and far
The oblique shock wave from the RDE combustor exit is directed field. The detonation combustion chamber is considered to be a
into the aerospike nozzle, forming a complex shock wave structure traditional annular combustion chamber with an inside radius (d1 )
that bounces around the expansion region of the nozzle [34]. Based of 40 mm, outside radius (d2 ) of 50 mm, and axial length (l1 ) of
on a steady-state design methodology, the nozzle flow overexpands 80 mm. The detonation nozzle adopts the aerospike nozzle struc-
throughout most of the RDE cycle, but the altitude-compensating ture, with an axial length (l2 ) of 100 mm, contraction section
characteristics of the aerospike nozzle preserve most of the ideal length (lc ) of 20 mm, outer wall length (lwall ) of 80 mm, and outer
performance under different pressure ratios [35]. Computations radius (d3 ) of 50 mm. The far field of the model is a cylindrical
have identified several performance advantages of coupling an RDE region with an axial length (l3 ) of 200 mm and a radius (d4 ) of
combustor with an aerospike nozzle. Jourdaine et al. [36] and Ku- 150 mm.
rita et al. [37] found that aerospike nozzles could improve the There are two predominant design methods for the contours
working performance of RDEs, and observed a reflected shock wave of aerospike nozzles. One is the method of characteristics (MOC)
propagating back upstream into the combustor before interacting [41], and the other employs the use of an isentropic area ratio

2
Y. Huang, H. Xia, X. Chen et al. Aerospace Science and Technology 117 (2021) 106969

2.2. Computational domain and conditions

The computational domain and grid structure of the RDE model

are shown in Fig. 3. The inlet boundary of the RDE combustion
chamber is the mass flow inlet. The mass flow rate is 0.53 kg/s and
specifying a constant mass flow rate boundary condition permits
the total pressure to vary in response to the interior solution. The
injection pressure and temperature are 1 Mpa and 300 K. The out-
let boundary is the pressure far field, set to 10000 Pa to correspond
to the air pressure at an altitude of 24 km. The internal and exter-
nal walls of the RDE model are adiabatic and non-slip boundaries.
There is a high-temperature (1500 K) and high-pressure (2.5 Mpa)
Fig. 2. Nozzle Mach number contour. (For interpretation of the colors in the fig- zone in front of the combustion chamber to initiate the rotating
ure(s), the reader is referred to the web version of this article.)
detonation. The reaction zone is filled with a mixture of hydrogen
and air at the equivalent ratio shown in Fig. 3(a). The grid struc-
equation. For the convergent part, a Witoszynski curve is used to ture of the RDE model is shown in Fig. 3(b). The grid structure is
obtain a well-distributed and stable flow [42]. For the divergent divided into a structural grid that is locally refined at the front of
part (as shown in Fig. 2), the design uses the minimum-length the combustion chamber and the nozzle wall to improve the accu-
MOC (black curve) and changes the Mach number from Mach 1 racy of the numerical simulations.
to the exit design Mach number along the divergent section (blue
line). The nozzle contour designed in this study adopts the method 2.3. Numerical method and validation
of combining a first-order function of the Mach number change
The governing equations are the three-dimensional Navier–
with the MOC. The minimum-length MOC is adopted in the expan-
Stokes equations with mass conservation equations for four species
sion section near the nozzle outlet, causing the outflow to expand
(H2 , O2 , H2 O, and N2 ). These are solved using the commercial CFD
sufficiently along the axial direction, and the first-order function
software ANSYS Fluent 14.0. The ANSYS Fluent solver is commonly
of the Mach number is adopted for the nozzle contour design near
used for simulations of rotating detonation waves and nozzle gas
the throat, which ensures a smooth nozzle throat transition. By fit- dynamics [14,43–46] and these simulation methods have been ex-
ting the back of the black line to the front of the blue line, the axial tensively verified. Thus, this is a feasible approach for simulat-
distribution of Mach number along the divergent part of the noz- ing detonation-related combustion and flow problems. The tran-
zle is constructed (red curve). Truncating the cowl and the spike sient explicit density-based solver is selected to solve the unsteady
gives the nozzle contour used in this study (as shown in Fig. 1). Navier–Stokes equations, and an implicit generalized second-order
The sectional area of the nozzle is given by: Runge–Kutta scheme is used for the time integration of the flow.
A second-order upwind scheme is used for the flow and turbu-
  2((γγ+−11))
A 1 1 + [(γ − 1)/2] Ma2 lence equations, and a ROE-FDS solver is utilized to construct up-
= (1) wind numerical fluxes. The gradients are calculated through the
A∗ Ma 1 + [(γ − 1)/2]
least-squares cell-based method [47]. The standard k − ε turbu-
where A is the sectional area of the nozzle, A ∗ is the throat area, lence model [48] is enough to capture the basic RDW structure
Ma is the Mach number along the path, and γ is the specific heat and flowfield about the shock wave and expansion characteristics
capacity ratio, which is equal to the specific heat of the mixed [16,43–45], and is employed in the present study. The thermal
exhaust components at the outlet of the combustion chamber. and mass diffusion are both considered in NS equations, and the

Fig. 3. Computational model of rotating detonation combustor with an aerospike nozzle.

3
Y. Huang, H. Xia, X. Chen et al. Aerospace Science and Technology 117 (2021) 106969

Fig. 4. Effect of mesh size on the temperature and pressure circumferential distributions of RDE combustor.

Fig. 5. Nozzle flowfield girds with coarse (0.2 mm), medium (0.1 mm), and fine meshes (0.05 mm).

Table 1
Comparison of numerical and experimental results [1] about typical detonation parameters of RDE model.

Typical parameter Computational result Experimental result [1] Relative error (%)
PC J /P0 13.412 13.695 2.06
T C J /T 0 8.789 8.9375 1.66
V C J /(m/s) 1621 1625.8 0.246

thermal and mass diffusion coefficient are obtained based on the zle is almost 16 million, and the calculation is run on 168 cores. It
kinetic-theory [16]. A one-step chemical reaction mechanism of takes about 4 weeks for the RDW to propagate one cycle. The use
H2 /air is used to model the chemical reaction of the rotating deto- of a smaller cell size for 3D simulation would continue to increase
nation [16], and the Arrhenius coefficients are chosen based on the the computational cost, and it was unpractical for the computa-
work of Ma et al. [49]. tional resources available for the current study. It found that the
To validate the accuracy of the detonation combustion calcula- three different grid sizes result in similar flow field structures, and
tion, three grid resolutions are chosen: 61 (axial) × 21 (radial) × every case has only one detonation wave in the combustion cham-
401 (circumferential), 121 × 41 × 801, and 181 × 61 × 1601. ber. Thus, since the present study does not pay attention to the
The minimum grid widths near the rotating detonation area are detailed cell structures of the detonation front, and the grid accu-
0.2 mm × 0.2 mm × 0.6 mm (coarse mesh), 0.1 mm × 0.1 mm × racy of 0.1 mm is fine enough to capture the detonation wave and
0.4 mm (medium mesh), and 0.05 mm × 0.05 mm × 0.2 mm (fine related flow structures appropriately.
mesh). The calculated results are compared with the experimental To verify the grid convergence of the nozzle flow computations,
results from Ref. [1]. Fig. 4 shows the influence of the grid size the wall boundary layer of the RDE nozzle is locally adapted to
on the pressure and temperature distributions in the detonation have minimum mesh sizes of 0.05–0.5 mm. Fig. 5 shows the influ-
combustion chamber. The pressure and temperature distributions ence of different mesh sizes on the pressure distribution along the
with different grid sizes are almost the same, which verifies the nozzle and the flow field velocity at the nozzle outlet. The compu-
grid convergence. Typical parameters of the detonation wave in the tational model of the nozzle with a minimum mesh size of 0.1 mm
numerical simulations and experiment are compared in Table 1, is sufficient to meet the computational requirements and the time
where P C J is the peak pressure of the detonation wave, T C J is the step are set to 0.02 μs. Additionally, According to Ref. [50,51], the
frontal temperature of the detonation wave, P 0 is the unburned accumulation of errors in unsteady process can be estimated by
material pressure, T 0 is the unburned reactant temperature, and the number of time steps, the mesh resolution and the order of ac-
V C J is the velocity of the detonation wave. The results show that curacy of numerical scheme. The allowable error was assumed 5%.
the deviation between the calculated and experimental detonation According to the calculation, the cumulative error is 4.48% when
wave parameters is less than 2.5%, which further verifies the ef- the grid scale and time step are set to 0.1 mm and 0.02 μs, respec-
fectiveness of the numerical simulation method used in this study. tively. It is lower than the upper limit of cumulative error, which
The total cell number for the RDE domain with an aerospike noz- ensures a reliable result on such an unsteady reactive problem.

4
Y. Huang, H. Xia, X. Chen et al. Aerospace Science and Technology 117 (2021) 106969

Fig. 6. Wave system structure in the rotating detonation combustor with an aerospike nozzle.

Fig. 7. (a) Schematic view of the shock wave and expansion wave inside aerospike nozzle; (b) expanded view of cylindrical surface along dividing line.

Fig. 8. Change of oblique shock wave structure with the axial position of RDE.

3. Results and discussion oblique shock waves are then formed downstream of the combus-
tion chamber. All those oblique shock waves pass into the RDE
3.1. Unsteady shock dynamics and flowfield in RDE with aerospike nozzle and form a complex spiral shock wave system. With the
nozzle movement of the detonation wave, the spiral wave system moves
into the RDE nozzle at high speed along the circumference. The
Fig. 6 illustrates the complex shock structure and flowfield in position and structure determine the flow field characteristics and
an RDE combustor with an aerospike nozzle. As shown in Fig. 6, working performance of the RDE. The distribution of the detona-
there are several detonation waves moving along the inner cir- tion waves in the RDE combustor exhibits central symmetric char-
cumferential wall in the head of the detonation chamber. Several acteristics [as shown in Fig. 6(b)]. The oblique shock wave structure

5
Y. Huang, H. Xia, X. Chen et al. Aerospace Science and Technology 117 (2021) 106969

Fig. 9. Shock structure with respect to the flow characteristics in front of the shock wave.

Fig. 10. Spatial and temporal evolution of total pressure of flow field in axial section of RDE.

Fig. 11. Pressure distribution of flow field in radial section of RDE.

downstream of the detonation also shows the characteristics of ordinates. The angles between the oblique shock wave and the cir-
central symmetry. Therefore, one of them is selected to analyze cumferential direction (Shock-Cir angle, θsc ), between the oblique
the characteristics of the shock wave, as shown in Fig. 7. shock wave and the airflow direction (Shock angle, θs ), and be-
A schematic view of the shock wave and expansion wave inside tween the airflow direction and the axial direction (Flow angle,
the aerospike nozzle based on the simulated flow field with a sin- θ f ) are defined in Fig. 7(b). These angles are used to characterize
gle detonation wave head is shown in Fig. 7(a). The oblique shock the dynamic changes of the spiral shock wave.
wave downstream of the detonation combustion chamber is trans- It can be seen from Fig. 8(a) that, with an increase in the axial
mitted into the nozzle, and a complex shock wave system, spirally position, θsc first drops and then rises. Upstream of the RDE flow
distributed around the nozzle wall, is formed in the aerospike noz- field, θsc is mainly affected by the detonation waves. The closer
zle; this system continuously rotates with the propagation of the the oblique shock wave is to the front of the detonation wave, the
detonation wave. Fig. 7(b) is drawn by fixing the central axis of closer the value of θsc is to the angle between the detonation wave
rotation, expanding the cylindrical surface along the dividing line, and the circumferential direction, so θsc shows a downward trend.
and then converting the cylindrical coordinates to rectangular co- Downstream of the RDE flow field, θsc is mainly affected by the

6
Y. Huang, H. Xia, X. Chen et al. Aerospace Science and Technology 117 (2021) 106969

Fig. 12. Total pressure recovery evolution along the RDE nozzle.

flow characteristics in front of the shock wave. As the gas flow ex-
pands and accelerates in the combustor, the flow angle θ f and the
velocity in front of the shock wave change, which leads to a change
in the shock waves. Fig. 8(b) shows that, as the axial position in-
creases, the oblique shock wave intensity, which is determined by
the ratio of  P (pressure difference before and after the shock
wave) to P 1 (pressure in front of the shock wave), first decreases
and then rises. Similar to θsc , the shock wave intensity is mainly
affected by the state parameters of the detonation wave upstream
of the RDE flow field and the relative velocity of the shock wave
downstream of the RDE flow field.
The law governing the changes in the shock dynamics with the
flow characteristics in front of the shock wave is shown in Fig. 9.
θsc increases with the flow angle θ f in front of the shock wave. The
magnitude of the shock angle θs and the flow velocity have an ap-
proximately inverse correlation. The shock angle decreases as the
flow velocity increases. It can be seen that the flow angle θ f and
the flow velocity are the main factors determining the structure of
the shock wave.
The total pressure along the path of the RDE airflow direction
can be calculated as:
 γ −1 V 02 γ γ−1
P 0 (1 + 2 γ RT
) dA
Ṗ t = (2)
A
where Ṗ t is the total pressure of the flow field, P 0 is the static
pressure of the flow field, γ is the specific heat capacity ratio, V 0
is the velocity of the flow field, and T is the temperature of the
flow field. Equation (2) provides a means of calculating the average
total pressure along the RDE.
To further analyze the shock dynamics and unsteady flow field
in the RDE, Figs. 10 and 11 show the motion of the shock waves
in the RDE. As the shock wave in the RDE has a stable, symmet-
rical structure, the motion of the shock wave in the axial section
exhibits periodic variations. Fig. 11 indicates that the spatial and
temporal pressure wave from the RDE combustor attenuates along
the RDE, and the azimuthal component in the exit plume of the
Fig. 13. Distribution of flow field inside and at exit plane of the RDE combustor.
RDE is dampened by the aerospike nozzle. According to the prop-
agation process of shock waves in the axial section of RDEs, as
shown in Fig. 10, we can identify four working stages. Fig. 12 nozzle, the shock wave only exists in the combustor, and the inten-
shows the periodic total pressure fluctuations along the nozzle sity decreases continuously. At this stage, the nozzle works at the
and the total pressure evolution at these four different stages of design point state, and the total pressure recovery of the nozzle is
the nozzle during a cycle. The total pressure recovery of the noz- high [as shown in Fig. 12(b)].
zle is defined as the ratio of the average pressure of the nozzle Second working stage: the shock wave propagates in the con-
inlet plane to that of the exit plane [as shown in Fig. 10(a)] at verging section of the nozzle (about 4 μs). The characteristics of
θ = 360/330◦ (time change rate t = 10e−6 s) intervals of cross- this working state are that, as the shock wave propagates in the
sectional area. converging section, the intensity of the shock wave increases and
First working stage: with the initiation of detonation, a wave reaches a maximum at the nozzle throat. At this stage, the nozzle
propagates in the RDE combustor (about 30 μs). The characteris- works in the over-expansion state and the total pressure recovery
tics of this working state are that there is no wave structure in the of the nozzle is low.

7
Y. Huang, H. Xia, X. Chen et al. Aerospace Science and Technology 117 (2021) 106969

Fig. 14. Shock wave structures under different backpressures.

Third working stage: the shock wave propagates in the nozzle

throat, interacting with the transonic flow, and gradually trans-
forms into a pressure wave (about 3 μs). The characteristics of this
working state are that the shock waves are affected by the tran-
sonic flow and the intensity fluctuates with the flow rate. At this
stage, the working state of the nozzle is gradually changing from
over-expansion to under-expansion. The total pressure recovery of
the nozzle is increasing.
Fourth working stage: a pressure wave enters the nozzle expan-
sion section and interacts with the supersonic flow. This pressure
wave strengthens into a shock wave under the effect of supersonic
flow. At this stage, the working state of the nozzle is gradually
changing from under-expansion to over-expansion. The total pres-
sure recovery of the nozzle is decreasing.
The real working state may be a superposition of the above
working states, influenced by the propagation speed of the shock
wave in the RDE. The flow field is characterized by the fact that
waves from the previous cycle have not yet left the nozzle by the
time the waves of the next cycle have entered.

3.2. Unsteady shock dynamics and flowfield in nozzleless RDE

combustor with different backpressures

The mounted nozzle increases the backpressure at the outlet

of the combustion chamber to a certain extent. The shock mo-
tion and unsteady flow characteristics in the combustion chamber
under different backpressures are shown in Fig. 13. As shown in
Fig. 13(a), as the backpressure of the combustor outlet increases,
the detonation structure in the RDE combustor gradually changes
from a single wave to multiple waves, and the propagation mode
becomes unstable with the continuous multi-wave collisions.
As the RDE combustor outlet pressure increases [as shown in
Figs. 13(b) and 13(c)], the axial velocity of the whole combustor
decreases, and the pressure of the reactants in front of the det-
onation wave increases. As a result, the detonation wave is more
easily generated and self-sustained, and the number of detonation
waves increases. When the backpressure of the RDE combustor ex-
ceeds some critical value (such as 3 atm), the combustor reaches
a thermally choked state and the detonation wave mode presents
continuous multi-wave collisions.
Fig. 15. Flow field comparison between the RDE combustor with/without aerospike
Fig. 14 shows the shock wave structure under different back- nozzle.
pressures. As shown in Fig. 14, the shock wave angle and intensity
both exhibit a downward trend and reach minimum values near
the combustor outlet. The change in the backpressure at the com-
bustor outlet mainly affects the amplitude of the periodic shock,
with larger values of the backpressure producing a smaller ampli-
tude of shock wave intensity and a smaller shock wave angle.

8
Y. Huang, H. Xia, X. Chen et al. Aerospace Science and Technology 117 (2021) 106969

Fig. 16. Shock wave structure comparison in RDE combustor with/without aerospike nozzle.

3.3. Comparison between RDE combustors with/without aerospike

nozzle

Fig. 15 compares the flow field in the cases of the RDE combus-
tor with and without the aerospike nozzle. As shown in Fig. 15(a),
under the design pressure (2 atm) of the RDE combustor outlet, the
plug nozzle changes the propagation mode of detonation waves in
the RDE combustor: the detonation flow field without the noz-
zle has three detonation waves traveling in the same direction,
whereas the detonation flow field with the nozzle has seven deto-
nation waves traveling in the same direction. Therefore, in addition
to changing the backpressure of the combustor, the RDE nozzle
has other unknown effects on the upstream RDE flow field. From
Fig. 15(b, c), it can be clearly observed that the pressure and Mach
number distribution at the RDE combustor exit plane with an noz-
zle is more uniform than nozzleless case. As a result, the mounted
nozzle could damp out the periodic flow oscillations at the rotating
detonation combustor exit, which is consistent with the previous
studies [7,9,24].
The shock wave structures in the RDE combustor with and
without the nozzle are compared in Fig. 16. Similar to the effect
of backpressure, the mounted aerospike does not change the trend
of the shock wave angle and intensity. Rather, it only changes the
amplitude of the shock wave angle and intensity. The shock wave
intensity and the shock angle θs are smaller with the aerospike
nozzle than in the nozzleless model, which is consistent with the
effect of backpressure on the shock dynamics. In other words, the
shock wave intensity of the RDE combustor with the nozzle is
weaker at the entrance and stronger at the exit. The RDE com-
bustor exit backpressure and the interaction between the nozzle
flow field and the shock wave are important factors in the propa-
gation mode of the detonation waves, as they affect the amplitude
of the shock wave angle and intensity.

3.4. Shock dynamics and expansion characteristics of integrated nozzle

with/without RDE combustor

Due to the unsteady time and space characteristics of the flow

at the RDE combustor outlet, the combustion chamber and nozzle
design need to be integrated. However, it is necessary to determine
the expansion characteristics of the RDE nozzle design by decou- Fig. 17. The temperature and pressure change profile of the measurement point of
pling the nozzle from the RDE combustor. The three measurement combustor exit plane under different working conditions.
points R1 , R2 , and R3 [see Fig. 17(a)] at the combustion cham-
ber outlet are used to analyze the influence of the RDE operating
state on the flow field parameters. The pressure and temperature the distribution and the magnitude of loss at each stage can be
at point 2 are shown in Figs. 17(b) and 17(c), respectively. The modified by changing the intensity and angle of the exit excitation
working state does not affect the periodic characteristics of the wave from the combustion chamber. Therefore, accurate simula-
wave propagation in the RDE, and the four stages of wave prop- tions of the structure and motion of the exit waves are of great
agation are still apparent (as shown in Fig. 12). The duration of importance for the design of rotating nozzles.

9
Y. Huang, H. Xia, X. Chen et al. Aerospace Science and Technology 117 (2021) 106969

Fig. 19. Pressure distribution of RDE nozzles with (left) and without (right) combus-
tor.

the nozzle on the flow field structure is mainly reflected in the

shock wave structure at the nozzle inlet. The θsc value of ∼ 64.3◦
at the entrance of the nozzle with the combustor is greater than
the value of ∼ 47.9◦ without the combustor.
Fig. 18. Monitoring curves of the RDE combustor with aerospike nozzle model. Fig. 20(b) shows the change in the shock wave intensity in the
RDE nozzle. The coupling of the detonation combustor and nozzle
has a significant influence on the shock wave intensity, as can be
The monitoring results at the rotating combustor exit with the
inferred from the following two observations: 1) As the flow field
aerospike are shown in Fig. 18. The time-averaged parameters from
parameters change at the outlet of the RDE combustor, the shock
the monitoring data can be used as the nozzle inlet conditions by
wave intensity at the nozzle inlet changes accordingly. At the en-
defining a suitable Fluent user-defined function [48]. The effect
trance of the RDE nozzle, the initial shock intensity of the model
of the combustor on the flow field characteristics and expansion
without the combustor is much lower than that with the com-
performance of the aerospike nozzle can then be calculated and
bustor. However, as the shock wave intensity in the RDE nozzle is
analyzed.
mainly affected by the flow field velocity (as shown in Fig. 8), the
Fig. 19 shows the pressure distribution of the flow field in the
shock wave intensity at the outlet of both aerospike nozzle mod-
RDE nozzle with and without the combustor. As shown in Fig. 19,
els is nearly the same. 2) Different from the variation in θsc , in-
regardless of whether the nozzle is connected to the combustor,
creasing the axial position produces an upward trend in the shock
the movement of the oblique shock wave in the aerospike noz-
wave intensity of the nozzle model without the combustor, but a
zle can again be divided into four working stages. The difference
brief decrease in the shock wave intensity appears near the noz-
between the two models is mainly reflected in the intensity and
zle throat. The aerospike nozzle improves the exit uniformity of
angle of the shock wave. The intensity of the shock wave in the
the RDE combustor and enhances the stability of the expansion
nozzle with the combustor is slightly stronger than that in the
and acceleration process in the nozzle. Therefore, the shock wave
nozzle without the combustor. intensity in the model with the combustor also exhibits a steady
Fig. 20 compares the shock wave structures inside the nozzle increase.
between the models with and without the combustor. As shown
The thrust and fuel specific impulse of the aerospike nozzle are
in Fig. 20(a), changes in the flow field structure within the RDE
given by:
combustor do not affect the trend in θsc , which exhibits an overall

upward trend in the RDE nozzle and reaches a maximum at the  
nozzle outlet. The influence of coupling the RDE combustor and F exit = ρ V 2 + P − P inf d A exit (3)

10
Y. Huang, H. Xia, X. Chen et al. Aerospace Science and Technology 117 (2021) 106969

Fig. 20. Shock wave dynamics inside RDE nozzle integrated with/without an RDE combustor.

and without an integrated combustor has been investigated. This

enabled a comprehensive analysis of the coupling characteristics of
the rotating detonation combustor and the aerospike nozzle. The
new features of the shock wave dynamics and expansion charac-
teristics in the nozzle with/without an integrated RDE combustor
were compared, and a nozzle design method based on the time-
averaged parameters of the rotating detonation combustor exit was
proposed by decoupling the nozzle from the RDE. Finally, the reli-
ability of this design method was analyzed.
Under the same conditions at the rotating detonation com-
bustor exit, the propagation mode of the detonation waves and
backpressure in the RDE combustor exit were found to be affected
by the aerospike nozzle. The RDE with an aerospike nozzle gen-
Fig. 21. Total pressure recovery varies with the axial position in RDE nozzle. erates more detonation wave heads than without the nozzle. Spi-
ral oblique shock waves form downstream of the RDE combustor
F and move into the nozzle at high speed along the circumference,
I sp = (4)
ṁg and their number corresponds to the number of detonation wave
heads. The spiral oblique shock wave positions and structures de-
where F exit is the nozzle outlet thrust, ρ is the flow field density, termine the flow field characteristics and working performance of
V is the nozzle outlet velocity, P is the nozzle outlet pressure, P inf the RDE.
is the external far-field pressure, A exit is the nozzle outlet area, The angles between the oblique shock waves and the circum-
I sp is the specific impulse for the mass flow of fuel, and g is the ferential direction (θsc ), between the shock wave and the airflow
acceleration due to gravity (9.81 m/s2 ). direction (θs ), and between the airflow direction and the axial di-
Fig. 21 compares the shock wave structures inside the nozzle rection (θ f ) were examined, and the shock intensity (ratio of pres-
between models with and without the combustor. Regardless of sure difference before and after the shock to the pressure before
whether the nozzle is connected to the combustor, the total pres- the shock), airflow velocity around the shock, and total pressure
sure recovery inside the nozzle first ascends and then descends. recovery along nozzle were analyzed to describe the shock wave
The coupling characteristics of the RDE combustor and nozzle do dynamics and unsteady flowfield in the rotating detonation com-
not affect the total pressure recovery trend inside the nozzle, al- bustor and nozzle. The magnitude of θs and the flow velocity are
though the model with the combustor gives a higher amplitude approximately inversely correlated. That is, θs decreases as the flow
and more dramatic changes in the convergent section of the RDE velocity increases. Similar to θsc , the shock wave intensity is mainly
nozzle (peak value of 116%). Therefore, the model with the com- affected by the state parameters of the detonation wave upstream
bustor has a higher total pressure recovery (88.1%) than the model of the RDE flow field and by the relative velocity of the shock wave
without the combustor (83.4%). downstream of the RDE flow field. The flow angle and flow velocity
Fig. 22 shows the effect of coupling the RDE combustor and the are the main factors determining the structure of the shock wave.
nozzle on the thrust performance of the RDE nozzle. As shown in The intensity of the shock wave and the angle with the direction
Fig. 22, the thrust and fuel specific impulse of the model with the of the air flow are smaller with the aerospike nozzle than in the
combustor are 1057.2 N and 6409.9 s−1 , respectively, which are nozzleless model, which is consistent with the effect of backpres-
both higher than in the model without the combustor. The rea- sure on the shock dynamics. The RDE combustor exit backpressure
son is that the inlet flow of the nozzle model with the combustor and the interaction between the nozzle flow field and shock wave
is more uniform [as shown in Fig. 13(a)], so the nozzle with the are the most important factors affecting the propagation modes of
detonation combustor undergoes greater expansion. The inlet flow the detonation wave, as they change the amplitude of the shock
field is closer to the design point of the nozzle, which is conducive wave angle and intensity.
to the expansion acceleration of the detonation nozzle. Regardless of whether it is connected to the combustor, the
nozzle exhibits similar working states of dynamic alternation from
4. Conclusions the design point to over-expansion and/or under-expansion along
the circumferential direction. The shock wave intensity and total
A thorough three-dimensional numerical simulation of a hydro- pressure ratio inside the nozzle with the combustor are slightly
gen–air rotating detonation combustor with and without an stronger than for the nozzle without the combustor. Consequently,
aerospike nozzle has been conducted, and the nozzle flow with the thrust and total pressure recovery of the nozzle integrated with

11
Y. Huang, H. Xia, X. Chen et al. Aerospace Science and Technology 117 (2021) 106969

Fig. 22. Thrust performance of different RDE nozzle models.

the RDE combustor are 7% and 5.6% higher, respectively, than in [16] N. Zhao, Q. Meng, H. Zheng, Z. Li, F. Deng, Numerical study of the influence of
the nozzle without the integrated combustor. Thus, decoupling the annular width on the rotating detonation wave in a non-premixed combustor,
Aerosp. Sci. Technol. 100 (2020) 105825.
nozzle from the RDE is feasible for testing the design performance
[17] H. Peng, W. Liu, S. Liu, H. Zhang, S. Huang, The competitive relationship be-
of unsteady nozzles. tween detonation and deflagration in the inner cylinder-variable continuous
rotating detonation combustor, Aerosp. Sci. Technol. 107 (2020) 106263.
Declaration of competing interest [18] X. Huang, C.J. Teo, B.C. Khoo, Experimental investigation on coexisting wave
components in an optically accessible rotating detonation combustor, Aerosp.
Sci. Technol. 111 (2021) 106538.
The authors declare that they have no known competing finan-
[19] S. Zhou, H. Ma, Y. Ma, C. Zhou, N. Hu, Experimental investigation on detonation
cial interests or personal relationships that could have appeared to wave propagation mode in the start-up process of rotating detonation turbine
influence the work reported in this paper. engine, Aerosp. Sci. Technol. (2021) 106559.
[20] T.H. Yi, J. Lou, C. Turangan, B.C. Khoo, P. Wolanski, Effect of nozzle shapes
Acknowledgements on the performance of continuously-rotating detonation engine, in: 48th AIAA
Aerospace Sciences Meeting Including the New Horizons Forum and Aerospace
Exposition, Orlando, FL, 2010, AIAA 2010-152.
This work was supported by the General program of Na-
[21] Y. Shao, M. Liu, J.-P. Wang, Continuous detonation engine and effects of dif-
tional Natural Science Foundation of China (No. 51876182) and ferent types of nozzle on its propulsion performance, Chin. J. Aeronaut. 23 (6)
the Fundamental Research Funds for the Central Universities [No. (2010) 647–652.
20720180058]. [22] Y. Wang, J. Le, C. Wang, Y. Zheng, S. Huang, The effect of the throat width of
plug nozzles on the combustion mode in rotating detonation engines, Shock
Waves 29 (4) (2019) 471–485.
References
[23] H. Zhang, W. Liu, S. Liu, Experimental investigations on H2 /air rotating deto-
nation wave in the hollow chamber with Laval nozzle, Int. J. Hydrog. Energy
[1] M. Hishida, T. Fujiwara, P. Wolanski, Fundamentals of rotating detonations, 42 (5) (2017) 3363–3370.
Shock Waves 19 (1) (2009) 1–10. [24] H. Peng, W. Liu, S. Liu, H. Zhang, Experimental investigations on ethylene-air
[2] K. Kailasanath, The rotating-detonation-wave engine concept: a brief status re- Continuous Rotating Detonation wave in the hollow chamber with Laval nozzle,
port, in: 49th AIAA Aerospace Sciences Meeting Including the New Horizons Acta Astronaut. 151 (2018) 137–145.
Forum and Aerospace Exposition, Orlando, FL, 2011.
[25] B.A. Rankin, M.L. Fotia, A.G. Naples, C.A. Stevens, J.L. Hoke, T.A. Kaemming, F.R.
[3] P. Wolański, Detonative propulsion, Proc. Combust. Inst. 34 (1) (2013) 125–158.
Schauer, Overview of performance, application, and analysis of rotating deto-
[4] E.M. Braun, F.K. Lu, D.R. Wilson, J.A. Camberos, Airbreathing rotating detonation
nation engine technologies, J. Propuls. Power 33 (1) (2017) 131–143.
wave engine cycle analysis, Aerosp. Sci. Technol. 27 (1) (2013) 201–208.
[26] M.L. Fotia, F. Schauer, T.A. Kaemming, J. Hoke, Experimental study of the per-
[5] F.K. Lu, E.M. Braun, Rotating detonation wave propulsion: experimental chal-
formance of a rotating detonation engine with nozzle, J. Propuls. Power 32 (3)
lenges, modeling, and engine concepts, J. Propuls. Power 30 (5) (2014)
(2016) 674–681.
1125–1142.
[27] M.L. Fotia, J. Hoke, F. Schauer, Performance of rotating detonation engines
[6] R. Zhou, D. Wu, J. Wang, Progress of continuously rotating detonation engines,
for air breathing applications, in: Detonation Control for Propulsion, Springer,
Chin. J. Aeronaut. 29 (1) (2016) 15–29.
Cham, 2018, pp. 1–21.
[7] V. Anand, E. Gutmark, Rotating detonation combustors and their similarities to
[28] M.L. Fotia, T.A. Kaemming, J. Hoke, J.R. Codonic, F. Schauerd, Experimental
rocket instabilities, Prog. Energy Combust. Sci. 73 (2019) 182–234.
thrust sensitivity of a rotating detonation engine to various aerospike plug-
[8] Z. Wang, K. Wang, Q. Li, Y. Zhu, M. Zhao, W. Fan, Effects of the combustor
nozzle configurations, AIAA Scitech 2019 Forum. AIAA 2019-1743, https://
width on propagation characteristics of rotating detonation waves, Aerosp. Sci.
doi.org/10.2514/6.2019-1743.
Technol. 105 (2020) 106038.
[9] A.J. Harroun, Investigation of Nozzle Performance for Rotating Detonation [29] Y. Zhu, K. Wang, Z. Wang, M. Zhao, Z. Jiao, Y. Wang, W. Fan, Study on the
Rocket Engines, Master’s thesis, Purdue University Graduate School, 2019. performance of a rotating detonation chamber with different aerospike nozzles,
[10] Q. Zhang, K. Wang, J. Wang, X. Qiao, W. Fan, Experimental research on vector Aerosp. Sci. Technol. 107 (2020) 106338.
control features of a pulse detonation tube with fluidic nozzle, Aerosp. Sci. [30] J.W. Bennewitz, B. Bigler, S. Danczyk, W.A. Hargus, R.D. Smith, Performance of
Technol. (2021) 106456. a rotating detonation rocket engine with various convergent nozzles, in: AIAA
[11] J. Kindracki, P. Wolański, Z. Gut, Experimental research on the rotating detona- Propulsion and Energy 2019 Forum, Indianapolis, IN, 2019, AIAA 2019-4299.
tion in gaseous fuels–oxygen mixtures, Shock Waves 21 (2) (2011) 75–84. [31] K. Goto, J. Nishimura, A. Kawasaki, K. Matsuoka, J. Kasahara, A. Matsuo, I. Fu-
[12] N.N. Smirnov, V.F. Nikitin, L.I. Stamov, E.V. Mikhalchenko, V.V. Tyurenkova, Ro- naki, D. Nakata, M. Uchiumi, K. Higashino, Propulsive performance and heating
tating detonation in a ramjet engine three-dimensional modeling, Aerosp. Sci. environment of rotating detonation engine with various nozzles, J. Propuls.
Technol. 81 (2018) 213–224. Power 35 (1) (2019) 213–223.
[13] N.N. Smirnov, V.F. Nikitin, L.I. Stamov, E.V. Mikhalchenko, V.V. Tyurenkova, [32] J. Braun, J. Saavedra Garcia, G. Paniagua, Evaluation of the unsteadiness
Three-dimensional modeling of rotating detonation in a ramjet engine, Acta across nozzles downstream of rotating detonation combustors, in: 55th AIAA
Astronaut. 163 (2019) 168–176. Aerospace Sciences Meeting, Grapevine, TX, 2017, AIAA 2017-1063.
[14] J. Sun, J. Zhou, S. Liu, Z. Lin, W. Lin, Numerical investigation of a non- [33] J. Braun, B.H. Saracoglu, G. Paniagua, Unsteady performance of rotating deto-
premixed hollow rotating detonation engine, Int. J. Hydrog. Energy 44 (31) nation engines with different exhaust nozzles, J. Propuls. Power 33 (1) (2017)
(2019) 17084–17094. 121–130.
[15] F. Wang, C. Weng, Y. Wu, Q. Bai, Q. Zheng, H. Xu, Numerical research on [34] D.A. Schwer, C.M. Brophy, R.H. Kelso, Pressure characteristics of an aerospike
kerosene/air rotating detonation engines under different injection total tem- nozzle in a rotating detonation engine, in: 2018 Joint Propulsion Conference,
peratures, Aerosp. Sci. Technol. 103 (2020) 105899. Cincinnati, OH, 2018, AIAA 2018-4968.

12
Y. Huang, H. Xia, X. Chen et al. Aerospace Science and Technology 117 (2021) 106969

[35] M.C. Schnabel, C.M. Brophy, Pressure distribution and performance impacts of [43] Y. Wang, Rotating detonation in a combustor of trapezoidal cross section for
aerospike nozzles on rotating detonation engines, in: 2018 AIAA Aerospace Sci- the hydrogen–air mixture, Int. J. Hydrog. Energy 41 (2016) 5605–5616.
ences Meeting, Kissimmee, FL, 2018, AIAA 2018-1626. [44] Q. Meng, N. Zhao, H. Zheng, J. Yang, L. Qi, Numerical investigation of the effect
[36] N. Jourdaine, N. Tsuboi, K. Ozawa, T. Kojima, A.K. Hayashi, Three-dimensional of inlet mass flow rates on H2 /air non-premixed rotating detonation wave, Int.
numerical thrust performance analysis of hydrogen fuel mixture rotating det- J. Hydrog. Energy 43 (2018) 13618–13631.
onation engine with aerospike nozzle, Proc. Combust. Inst. 37 (3) (2019) [45] Y. Wang, J. Wang, W. Qiao, Effects of thermal wall conditions on rotating deto-
3443–3451. nation, Comput. Fluids 140 (2016) 59–71.
[37] N. Kurita, N. Jourdaine, N. Tsuboi, K. Ozawa, A.K. Hayashi, T. Kojima, Three- [46] W.A. Stoddard, E.J. Gutmark, Numerical investigation of centerbodiless RDE
dimensional numerical simulation on hydrogen/air rotating detonation engine design variations, in: 53rd AIAA Aerospace Sciences Meeting, Kissimmee, FL,
with aerospike nozzle: effects of nozzle geometries, in: AIAA SciTech 2020 Fo- 2015, AIAA 2015-0876.
rum, Orlando, FL, 2020, AIAA 2020-0688. [47] J. Sun, J. Zhou, S. Liu, Z. Lin, W. Lin, Plume flowfield and propulsive perfor-
[38] D.P. Stechmann, S.D. Heister, A.J. Harroun, Rotating detonation engine perfor- mance analysis of a rotating detonation engine, Aerosp. Sci. Technol. 81 (2018)
mance model for rocket applications, J. Spacecr. Rockets 56 (3) (2019) 887–898. 383–393.
[39] A.J. Harroun, S.D. Heister, S.V. Sardeshmekh, J.H. Ruf, Effect of aerospike nozzles [48] ANSYS FLUENT 14.0, User’s and theory guide, Canonsburg, Pennsylvania, USA:
on rotating detonation engine performance for rocket applications, in: AIAA ANSYS, Inc., 2011.
Scitech 2019 Forum, San Diego, CA, 2019, AIAA 2019-0197. [49] F. Ma, J.Y. Choi, V. Yang, Propulsive performance of airbreathing pulse detona-
[40] A.J. Harroun, S.D. Heister, J.H. Ruf, Experimental validation of nozzle flow sim- tion engines, J. Propuls. Power 22 (2016) 1188–1203.
ulations for rocket application rotating detonation engines, in: AIAA Scitech [50] N.N. Smirnov, V.B. Betelin, V.F. Nikitin, L.I. Stamov, D.I. Altoukhov, Accumulation
2020 Forum, Orlando, FL, 2020, AIAA 2020-0198. of errors in numerical simulations of chemically reacting gas dynamics, Acta
[41] G. Angelino, Approximate method for plug nozzle design, AIAA J. 2 (10) (1964) Astronaut. 117 (2015) 338–355.
1834–1835. [51] N.N. Smirnov, V.B. Betelin, R.M. Shagaliev, V.F. Nikitin, I.M. Belyakov, Y.N.
[42] E. Witoszynski, Über Strahlerweiterung und Strahlablenkung, Springer, Berlin- Deryuguin, S.V. Aksenov, D.A. Korchazhkin, Hydrogen fuel rocket engines sim-
Heidelberg, 1924, pp. 248–251. ulation using LOGOS code, Int. J. Hydrog. Energy 39 (20) (2014) 10748–10756.

13