You are on page 1of 22

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/361788069

Prediction and visualization of supersonic nozzle flows using OpenFOAM

Article  in  Journal of Visualization · June 2022


DOI: 10.1007/s12650-022-00856-5

CITATION READS

1 571

4 authors, including:

Prasanth P Nair Vinod Narayanan


Indian Institute of Technology Gandhinagar Indian Institute of Technology Gandhinagar
22 PUBLICATIONS   114 CITATIONS    39 PUBLICATIONS   86 CITATIONS   

SEE PROFILE SEE PROFILE

Abhilash Suryan
College of Engineering Trivandrum
89 PUBLICATIONS   482 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

9th Asian Joint Workshop on Thermophysics and Fluid Science View project

Real gas effects in s-CO2 modeling View project

All content following this page was uploaded by Vinod Narayanan on 02 September 2022.

The user has requested enhancement of the downloaded file.


J Vis
https://doi.org/10.1007/s12650-022-00856-5

R E G UL A R P A P E R

Prasanth P. Nair • Vinod Narayanan • Abhilash Suryan • Heuy Dong Kim

Prediction and visualization of supersonic nozzle flows


using OpenFOAM

Received: 31 January 2021 / Revised: 5 May 2022 / Accepted: 22 May 2022


Ó The Visualization Society of Japan 2022

Abstract At low altitudes during rocket flight, the atmospheric pressure is higher compared to the design
pressure of the nozzle at the exit. This leads to the formation of overexpansion shock, and consequently,
flow separation. When the separation is asymmetric, the lateral force acts on the nozzle wall, and the
magnitude of the lateral force depends on the extent of asymmetry. Hence, accurate prediction of the flow
separation is essential to estimate side loading. This study uses OpenFOAM and ANSYS to analyze flow
separation. OpenFOAM offers the flexibility to modify the code as per the requirements of the problem, as
the code is readily available. There is only a limited number of studies conducted on supersonic nozzles
using OpenFOAM. This study addresses the choice of solver, discretization method, and boundary condi-
tions to be implemented for accurately predicting supersonic flow through different nozzle geometries. The
analysis is conducted on cold flow through planar convergent-divergent, planar expansion-deflection, and
conical aerospike nozzle geometries. Reynolds-averaged Navier–Stokes equations are solved along with
turbulence models. Compressible solvers sonicFOAM and rhoCentralFOAM are used for the simulations
with OpenFOAM. Different turbulence models are tested and validated for the planar convergent-divergent
nozzle and compared with the expansion-deflection nozzle and aerospike nozzle. The results are validated
with available experimental data. While comparing the supersonic flow through the different nozzles, it is
observed that rhoCentralFOAM captures flow separation, shocks, shear layer, and pressure profile better in
comparison to sonicFOAM.

Keywords Planar nozzle  Separated flow  OpenFOAM  Supersonic flow  Asymmetry 


Advanced rocket nozzle

Abbreviations

Ae Area at nozzle exit, mm2


At Area at nozzle throat, mm2
cp Specific heat at constant pressure, J/kgK

P. P. Nair  V. Narayanan
Mechanical Engineering Discipline, Indian Institute of Technology Gandhinagar, Gujarat 382055, India
E-mail: prasanth.n@iitgn.ac.in
V. Narayanan
E-mail: vinod@iitgn.ac.in

A. Suryan (&)
Department of Mechanical Engineering, College of Engineering Trivandrum, Kerala 695016, India
E-mail: suryan@cet.ac.in

H. D. Kim
Department of Mechanical Engineering, Andong National University, Andong 1375, Republic of Korea
P. P. Nair

D Exit nozzle diameter, mm


E Total energy, J
FSS Free Shock Separation
I Unit tensor
k Turbulence kinetic energy, J/kg
kT Thermal conductivity, W/(mK)
L Axial length of divergent section of the nozzle, mm
LW Lower Wall
NPR Nozzle Pressure Ratio
Pa Atmospheric pressure, Pa
Pe Pressure at the exit of the nozzle, Pa
Po Jet stagnation pressure, Pa
Pw Wall pressure on nozzle profile, Pa
p Static pressure, Pa
Prt Turbulent Prandtl number
RANS Reynolds-Averaged Navier–Stokes
RSS Restricted Shock Separation
Re Radius of the nozzle exit, mm
SA Spalart–Allmaras
SST Shear Stress Transport
Th Height of the throat, mm
UW Upper Wall
m Molecular kinematic viscosity
ve Velocity at the exit, m/s
X Coordinate along X-axis, mm
Y Coordinate along Y-axis, mm
c Ratio of the specific heats
q Density, kg/m3
l Dynamic viscosity, Pa.s
lt Turbulent viscosity, Pa.s
leff Effective viscosity, Pa.s

1 Introduction

Flow separation in a rocket nozzle at the over-expanded condition is one of the major issues faced by the
aerospace industry. As a rocket ascends from sea level until it reaches design altitude, the flow falls under
the overexpanded regime. In this regime, the flow separates from the nozzle surface due to high ambient
pressure. When the flow separation is asymmetric, the rocket nozzle experiences a lateral load on the
surface. The magnitude of side loads depends on the extent of this lateral load and could lead to catastrophic
consequences (Frey and Hagemann 1999; Verma et al. 2006, 2017). Therefore it is necessary to accurately
predict the flow separation location to improve the nozzle design. In a planar convergent-divergent nozzle,
there are different overexpansion flow conditions, as shown in Fig. 1. Initially, due to high ambient pressure,

Fig. 1 Flow features of a planar nozzle at different NPR (Nair et al. 2020b, a)
Prediction and visualization of supersonic nozzle

the flow experiences a free shock separation (FSS), as shown in Fig. 1a. As the ambient pressure starts
decreasing, a strong shock is observed, as shown in Fig. 1b. On the further decrease in ambient pressure, an
overexpansion shock is observed. Due to asymmetry in the formation of the k-shock, the flow starts to bend
and gets attached to either one of the surfaces, as shown in Fig. 1c. Restricted shock separation (RSS) is
seen at one side of the nozzle wall and FSS on the other, as shown in Fig. 1d. As the ambient pressure
decreases, the shock due to overexpansion moves downstream of the nozzle, and due to the end effect, the
flow on both sides of the nozzle becomes FSS, as shown in Fig. 1e. Many numerical and experimental
studies had attempted to understand the flow separation and the resultant side loads (Nave and Coffey 1973;
Frey and Hagemann 2000; Damgaard et al. 2004; Ruf et al. 2010; Nair et al. 2020a, b).
Several studies were conducted in the past, both experimentally and computationally, on flow separation
phenomena in an over-expanded nozzle. Hunter (1998, 2004) conducted experimental, analytical, and
numerical study regarding the flow separation in a planar convergent-divergent nozzle. The experimental
study pointed out that at overexpansion condition, the shock/boundary layer interaction (SBLI) separation
was dominant. In the overexpanded condition, two discrete flow separation regions were identified. When
the nozzle pressure ratio (NPR) is less than 1.8, unsteady and asymmetric flow separation is observed.
However, for the overexpanded condition with NPR greater than 2, a steady and symmetric flow separation
was observed. The NPR is the ratio of jet stagnation pressure to the ambient pressure. For experimental and
numerical purposes, ambient pressure remains constant, and jet stagnation pressure is varied. A transitional
zone between two flow separation regions was observed in the range from NPR 1.8 to NPR 2. This transition
zone was a result of natural thermodynamic balance instead of SBLI. Another experimental study conducted
on a planar convergent-divergent nozzle by Papamoschou and Zill (2004) in which the area ratio was varied
from 1.0 to 1.5. The experiment was conducted for the range of NPR from 1.2 to 1.7. It was observed that
when the NPR increased from low to high, three modes of flow separation occurred. At first, symmetric flow
separation was observed, followed by asymmetric flow separation, and finally again returning to symmetric
flow separation. It was found that shocks in these flow regimes were unsteady. Xiao et al. (2007) conducted
a numerical study on a planar convergent-divergent nozzle having an area ratio of 1.5, based on the
experimental study conducted by Papamoschou and Zill (2004). They found asymmetric flow separation for
the NPR range of 1.5 to 2.4. For higher NPR, the flow was found to be symmetric. However, unsteady
shocks were not captured in the study. In a different experimental study conducted by Papamoschou et al.
(2009) on a planar convergent-divergent nozzle, it was seen that the flow attachment for asymmetric flow
separation was sensitive to the initial condition. Once the flow adheres to one side of the nozzle, it would
stick to that side for the complete test duration. If the flow attaches to one side of the nozzle, it preserves the
position for the entire duration of the test. If the same experiment was restarted, then the flow was observed
to attach on the other side of the nozzle wall. This phenomenon was explained by the Coanda effect.
There have been studies to use advanced rocket nozzles to reduce or eliminate the flow separation at
underexpansion conditions. These nozzles are also known as altitude compensation nozzles as they work
efficiently for a wide range of altitudes. Many studies have been focused on different types of advanced
rocket nozzles, such as the dual-bell nozzle (Choudhury et al. 2018; Zmijanovic et al. 2018), plug nozzle,
expansion deflection (ED) nozzle, dual throat (Wang et al. 2019), and extendible nozzle (Hagemann et al.
1998). These nozzles could be used for Single Stage to Orbit or Reusable Launch Vehicle missions.
Aerospike nozzles are a type of plug nozzle, which is similar to a bell nozzle inverted inside out. Since
there is no outer surface like that of the bell nozzle, the thrust generated is imparted on the center spike.
Therefore, for the varying altitude, the nozzle works efficiently. ED nozzle is a type of bell nozzle with a
pintle inside the nozzle that controls the expansion process. The altitude adaptability of the ED nozzle is
achieved by the position of the pintle just downstream of the throat. The pintle plays a passive role in
changing the effective area ratio relative to ambient pressure as the altitude varies. The pintle deflects the
supersonic gas coming from the throat of the nozzle radially outwards toward the nozzle walls, hence
reducing or altogether avoiding flow separation at low NPR. There have been many recent studies, both
experimentally and numerically, to test the performance parameters on aerospike as well as ED nozzles
(Taylor et al. 2010, 2011; Wagner et al. 2011; Wagner and Schlechtriem 2011; Verma and Viji 2011;
Karthikeyan et al. 2013; Nair et al. 2017; Chutkey et al. 2018; Nair et al. 2019a, b; George et al. 2021;
Soman et al. 2021).
Computational fluid dynamics (CFD) over the years has become one of the necessary tools for designing
and analysis of a wide variety of problems happening both in the domain of engineering and understanding
the fundamental phenomena occurring in nature (Agarwal 1999; Wang et al. 2006; Suzuki and Koyaguchi
2013). The Open Source Field Operation and Manipulation (OpenFOAM) was first released in 2004, since
P. P. Nair

then, OpenFOAM has been widely used in the engineering and scientific communities. The OpenFOAM
(Greenshields 2020) is an open-source CFD software in which the codes are written in C ? ? . The
OpenFOAM provides a variety of finite volume solvers for both structured and unstructured grids. It has
significantly expanded and improved based on the available solvers and its pre- and post-processing
capabilities. OpenFOAM currently incorporates more than 90 solvers, ranging from DNS, combustion, heat
transfer, incompressible, and compressible flows to electromagnetics, finance, stress analysis of solids, etc.
OpenFOAM being open-source has some discrete advantages over commercially available CFD toolkits.
The codes can be modified according to the user’s need to solve a problem, and also, new solvers can be
created by modifying the existing solvers. Therefore, several CFD analysis has been conducted and pub-
lished based on the OpenFOAM package (Lysenko et al. 2010; Flores et al. 2014; Vuorinen et al. 2014;
Rabbani et al. 2016; Constant et al. 2017; Tunstall et al. 2017; Zang et al. 2018a, b; Putra et al. 2019; Abed
et al. 2020; Ali et al. 2021).
Doolan (2009) had performed a two-dimensional unsteady simulation on NASA tandem cylinder
experiment using OpenFOAM. The results of the computational simulation of unsteady and mean flow were
compared with the experimental results and found to be reasonably accurate. The accuracy of the noise-level
predictions was increased by implementing statistical models. It was found that noise-level predictions were
in good overall agreement with the experimental acoustic power spectra. The unsteady three-dimensional
swirling flow simulation was performed on a conical diffuser with a precessing vortex rope (Muntean et al.
2009). The standard k-e turbulence model was solved with the URANS equations. The results of Open-
FOAM and ANSYS fluent were compared and validated with the experimental results. It was concluded that
the higher harmonics and fundamental frequency were accurately captured. Only the last part of the cone
had a discrepancy with regard to the experimental result. D’Alessandro et al. (2016) conducted detached
eddy simulation (DES) on flow over a cylinder at Reynolds number 3900 to examine the DES capability of
OpenFOAM. Standard SA-DES and improved Delayed-Detached-Eddy Simulation (IDDES) were tested. It
was concluded that both were in good agreement with the experimental data and recommended the uses of
OpenFOAM for simulating turbulent flow with DES. Large-eddy simulations (LES) were conducted by Cao
and Tamura (2016) on flow over a square cylinder. Unstructured LES were examined using Linear-Upwind
Stabilized Transport (LUST), limitedLinear, and linearUpwind schemes. It was observed that the 1st and the
2nd order of schemes did not influence the accuracy of the time-averaged quantities. The accuracy of the
time-averaged quantities using LUST and linearUpwind schemes was similar. However, for limitedLinear
non-physical distribution was observed in the inertial subrange.
Palharini et al. (2015) conducted a benchmark simulation on high- and low-speed non-reacting flows
using dsmcFOAM for simple as well as complex geometries and noted that the results matched well with the
previous experimental and numerical results. White et al. (2018) further incorporated molecular vibrational
and electronic energy modes, chemical reactions, and subsonic pressure boundary conditions within the
dsmcFOAM and named the new solver dsmcFoam ? . Cao et al. (2021, 2022) had further modified the
dsmcFoam ? code to simulate both steady and transient flows for any two or three dimensional, two-phase
rarefied flows that also includes a phase change model for materials. Yang et al. (2015) had validated a
viscoelastic flow solver based on OpenFOAM. Robertson et al. (2015) conducted an extensive study on the
validation of OpenFOAM for the bluff body in incompressible flows. The study included a backward-facing
step, sphere, and delta wing. The results of the OpenFOAM simulation were compared with the experiment
and ANSYS Fluent. It was concluded that OpenFOAM was a dependable solver; nevertheless, it was
affected by the quality of the mesh in comparison to ANSYS Fluent. Han et al. (2015) had used LES with
PaSR (Partially Stirred Reactor) model in OpenFOAM for the prediction of combustion instability. Han
et al. (2020) had also used the model for simulating stratified swirl burner in order to understand the flame
stabilization, combustion instabilities, and beating oscillations. It can also be noted that OpenFOAM results
in some studies were also used as a precursor simulation for mapping the data to DNS (Kadu et al.
2019, 2020). Nakao et al. (2014) had conducted a numerical study on NACA 64A010 using OpenFOAM and
compared the solvers sonicFOAM and rhoCentralFoam. It was found that sonicFOAM had performed better
in comparison to rhoCentralFOAM in the subsonic regime. Ashton and Skaperdas (2019) conducted a
comparative study of OpenFOAM and STAR-CCM ? using a flat plate, NACA0012 airfoil, DSMA661
airfoil, NASA HLRM, and JAXA JSM model. It was found that the computation result of OpenFOAM
matches with the STAR-CCM ? , and the experimental data. It was concluded that the unstructured hybrid
prismatic-tetrahedral grid took a more significant time to converge as the inability of OpenFOAM to run in
higher Courant number as compared to STAR-CCM ? . Adaptive shock control on the bumps of transonic
wings was conducted using sonicFOAM (Jinks et al. 2014) and rhoCentralFOAM solver (Jinks et al. 2018;
Prediction and visualization of supersonic nozzle

Gramola et al. 2020). Wojewodka et al. (2022) conducted numerical simulations on the Wellborn s-duct
using steady-state and transient solvers, the rhoSimpleFOAM and rhoPimpleFOAM, respectively, for a
Mach number of 0.6 at the inlet. They had also conducted Unsteady Reynolds Averaged Navier–Stokes
(URANS) and Detached Delayed Eddy Simulation (DDES) for the transient simulations. It was noted that
the DDES simulations closely matched the experimental data.
There are many studies using OpenFOAM for incompressible and subsonic flow. However, only limited
studies have been conducted using OpenFOAM in the supersonic flow regime. Droeske et al., (2014) created
a new solver in OpenFOAM to investigate the supersonic flow inside a hot gas channel flow. The simulation
was conducted in both two dimensions as well as three dimensions. The result of the simulation was
compared with the commercial code results and experimental data. It was found to be a good match. It was
also found that van-Driest transformation had improved the simulation result in comparison to standard wall
function. However, the temperature gradient near the wall lacked accuracy. Mukundhan and Kumar (2017)
performed the design and optimization of a two-dimensional supersonic intake using OpenFOAM. It was
concluded that variable geometry intake had better performance in comparison to fixed geometry in off-
design condition. Stold et al. (2020) performed a validation test on rhoCentralFOAM, using shock tube,
supersonic turbulent flat plate, laminar oblique shock, and delta wing. The results showed that van Albada
flux limiter was best optimized for stability, speed, and accuracy. The solver rhoCentralFOAM was in good
agreement with the analytical solution. There was a slight disagreement in the supersonic delta wing. It was
concluded that the disagreement was due to the SST k-x turbulence model. It was further proposed that DES
could be a better option for solving such flows to reduce the error between experimental and numerical
results. Zang et al. (2018a, b) evaluated rhoCentralFOAM solver using a supersonic jet coming from a round
nozzle for different NPR. The simulation was performed in the region of overexpansion, design, and
underexpansion conditions. The computed results were in good match with the experimental data. It was
observed that the SST k-x turbulence model had over-estimated the spread of the shear layer. However, it
was within the acceptable limits. When the solution obtained by rhoCentralFOAM was compared with
ANSYS Fluent, it was observed that the performance of rhoCentralFOAM was comparable with ANSYS
Fluent.
In the present paper, a comparative study has been conducted on a 5.7° and 10.7° planar CD nozzle,
planar ED nozzle, and conical aerospike nozzle. The results of the simulation using OpenFOAM are
compared with the results of a previously conducted experimental and numerical study using ANSYS Fluent
(2013). The parametric study of turbulence models and solvers is first done on a planar CD nozzle geometry.
Then, the turbulence model and solver are implemented on the planar ED nozzle and conical aerospike
nozzle. This paper serves as a guideline for selecting a solver, discretization method, and turbulence model
for simulating flow conditions within different supersonic nozzles having complex geometries using
OpenFOAM. The solvers compared in this paper are sonicFOAM and rhoCentralFOAM.

2 Numerical methodology

This section describes governing equations used for the simulation. The geometry and grid details, boundary
conditions applied, and the CFD solver used are discussed in depth. The solver sonicFOAM is an unsteady
solver for trans-sonic/supersonic compressible flow with turbulence, whereas rhoCentralFOAM is a density-
based unsteady solver for compressible flow with turbulence. The rhoCentralFOAM is based on Kurganov
and Tadmor scheme (2000), and Kurganov et al. (2001) central-upwind scheme. The simulation in the
current paper has been conducted using OpenFOAM 5.0. It should be noted that sonicFOAM has been
integrated into rhoPimpleFOAM. In order to use the integrated sonicFOAM, the transonic has to be true in
the PIMPLE sub-dictionary in the fvSolution.

2.1 Governing equations

Numerical simulations are performed on two-dimensional planar CD nozzles of 5.7° and 10.7° divergence
angle, planar ED nozzle, and axisymmetric aerospike nozzle. Two-dimensional, unsteady, Reynolds-aver-
aged Navier–Stokes (RANS) equations are solved along with turbulence models. The system of equations is
closed with the ideal gas equation of state, and viscosity is defined by Sutherland’s law (1893). The
governing equations which are utilized for the numerical simulation are given below:
Continuity Equation:
P. P. Nair

oq o
þ ðqui Þ ¼ 0 ð1Þ
ot oxi
Momentum equation:
  
o o  op o oui ouj 2 oul o  0 0

ðqui Þ þ qui uj ¼  þ l þ  dij þ qui uj ð2Þ
ot oxj oxi oxj oxj oxi 3 oxl oxj
0 0
where qui uj is the Reynolds stresses. The turbulence models used in the current study use the Boussinesq
hypothesis that relates the Reynolds stresses, which is given by:
   
0 0 oui ouj 2 ouk
qui uj ¼ lt þ  qk þ lt dij ð3Þ
oxj oxi 3 oxk
Energy equation:
  
o o o cp lt oT 
ðqEÞ þ ½ui ðqE þ pÞ ¼ kT þ þ ui sij eff ð4Þ
ot oxi oxj Prt oxj

where sij eff
is the deviatoric stress tensor, which is given by:
 
 ouj oui 2 ouk
sij eff ¼ leff þ  leff dij ð5Þ
oxi oxj 3 oxk
where lt is the turbulent viscosity,Pr t is the turbulent Prandtl number and was taken as 0.85, leff is the
effective viscosity that is leff ¼ l þ lt . For Spalart–Allmaras (SA) model, one additional transport equation
that represents the turbulent viscosity is solved, whereas in the case of k-e and k-x turbulence models, two
additional transport equations for the turbulence kinetic energy, k, and either the turbulence dissipation rate,
e, or the specific dissipation rate, x, are solved, and the turbulent viscosity is computed as a function of k-e
or k-x.

2.2 Discretization

For sonicFOAM, the discretization for the temporal term is done by Euler’s scheme. Euler’s scheme is a
first-order implicit bounded scheme. The gradient scheme used is Gauss linear. The Gauss in the Gauss
linear scheme specifies the discretization of Gaussian integration. It requires the interpolation of values from
cell centers to face centers. The scheme used for the interpolation of values from cell centers to face centers
is linear interpolation or central differencing. The divergence schemes used for all the terms except tur-
bulence quantities are Gauss limitedLinear 1. It is a second-order scheme that limits upwind in areas where
the gradient change is rapid, and the ‘‘1’’ represents strong limiting. For the discretization of turbulence
quantities, upwind scheme was chosen. The upwind scheme is a first-order bounded scheme. The corrected
scheme is used for the surface normal gradient schemes for a maximum non-orthogonality of 70°. The
Laplacian terms are discretized using Gauss linear corrected scheme. Linear scheme is used for the inter-
polation scheme.
For rhoCentralFOAM, the flux scheme is chosen as Kurganov et al. (2001). The discretization for the
temporal term is done by Euler’s scheme. The gradient scheme used is Gauss linear. The divergence
schemes used for all the terms except turbulence quantities are Gauss linear. Gauss linear is an unbounded
second-order scheme. For discretization of turbulence quantities, upwind scheme is chosen. The corrected
scheme is used for the surface normal gradient schemes. The Laplacian terms are discretized using Gauss
linear corrected scheme. Linear scheme was used as the default interpolation scheme. For the reconstruction
of U, T, and rho, vanLeer interpolation is used.

2.3 Nozzle geometry and computational domain

Three cases are considered for the comparative study. Case 1 consists of two planar CD nozzles having
different divergence angles and the same area ratio. Case 2 and Case 3 consist of a planar ED nozzle and a
conical aerospike nozzle, respectively. The geometry for the planar CD planar nozzle is based on the
configuration used by Verma and Manisankar (2014) for their experimental study.
Prediction and visualization of supersonic nozzle

The geometry is divided into two segments, the first segment consists of a converging section up to the
throat, and the second segment consists of a small rounded contour part followed by the diverging section, as
shown in Fig. 2a. The second segment consists of the straight diverging segment, as shown in Fig. 2a. For
the evaluation of two planar CD nozzles, the area ratio (Ae/At = 1.79) is kept the same. The throat height
(Th) is taken as 28 mm. Two different diverging angles of 5.7° and 10.7° are taken for the analysis. As the
diverging angle increases, the length (L) is reduced to keep the area ratio same.
The geometry used for the ED planar nozzle is based on the configuration used by Wagner et al. (2011).
The model considered for the analysis is symmetric. As a result, only one-half of the nozzle is modeled for
running the simulations. The throat height (Th) was taken as 4 mm. The nozzle inlet height is taken to be
2.5Th. The nozzle geometry is as shown in Fig. 2b. The throat is inclined at an angle of 40.6° to the x-axis in
the direction of the flow. The nozzle contour is designed as per the Prandtl–Meyer expansion, which resulted
in an area ratio of 4. The computational domain for the planar ED nozzle varied from 50Th along the x-axis
to 34Th along the y-axis, as shown in Fig. 3b.

Fig. 2 Schematic diagram of nozzle geometry: a Planar CD nozzles, b Planar ED nozzle, and c Conical aerospike
P. P. Nair

Fig. 3 Computational domain: a Planar CD nozzles, b Planar ED nozzle, c and d Conical aerospike nozzle
Prediction and visualization of supersonic nozzle

For the conical aerospike nozzle, the configuration used for analysis is the same as that used by Verma
(2009). The conical aerospike nozzle is having 15° half-angle, as shown in Fig. 2c. The radius of the nozzle
exit (Re) was taken as 25 mm. The throat height (Th) is taken as 9 mm, and the length L of the conical
aerospike nozzle is 59.71 mm taken from the throat. The cowl length was taken as 9 mm from nozzle throat
to cowl tip, and the inner nozzle area ratio was taken as 1.19. The computational domain varied from 60D on
the x-axis to 10D on the y-axis, where D is the exit nozzle diameter. The after body is extended up to 5D on
upstream of the throat, and the inlet for the simulation is placed at 2.5D from the throat. The computational
domain used in the analysis is shown in Fig. 3c. For the two-dimensional axisymmetric case in OpenFOAM,
the domain had to be wedged to an angle of 2.5° as shown in Fig. 3d.
The computational grid for the current study is generated using ANSYS ICEM CFD. The details of grids
and grid independence studies for conical aerospike nozzle, planar ED nozzle, and planar CD nozzle are
provided in the previous studies (Nair et al. 2019a, b, 2020a; Paul et al. 2020).

2.4 Boundary conditions

The stagnation temperature and pressure are taken as 300 K and 1 bar, respectively. At walls, adiabatic and
no-slip boundary conditions are implemented. The simulations for different NPR are conducted by changing
the inlet stagnation pressure. The ambient conditions at the far-field were assumed to be stagnation tem-
perature and pressure. Non-reflecting boundary condition is applied at the outlet. The details of boundary
conditions implemented in OpenFOAM are provided in Table 1.

2.5 Flow solver

In the present study, compressible solvers sonicFOAM and rhoCentralFOAM are chosen to simulate flow
within different nozzle configurations. For sonicFOAM, the diagonal solver was used to solve rho, and
smoothSolver, with symGaussSeidel smoother, and was used to solve the rest of the discretized terms. For
rhoCentralFOAM, the diagonal solver was used to solve rho. The matrix of turbulence terms is solved using
Preconditioned bi-conjugate gradient (PBiCGStab) solver with Diagonal-based Incomplete LU (DILU)
smoother, with L being lower unitriangular and U being upper triangular. The rest of the terms were solved
using Geometric agglomerated Algebraic MultiGrid (GAMG) solver with GaussSeidel smoother. The time
step is taken as 1e-8 s, and the end time is taken as 0.02 s. The solutions are saved for every 1e-4 s. The time
averaging is done from 0.01 to 0.02 s.

3 Results and discussion

Three cases of two-dimensional supersonic nozzles are considered in the study. Case 1 consists of two
planar CD nozzles having divergence angles of 5.7° and 10.7°, but having the same area ratio. Case 2 and
Case 3 consist of a planar ED nozzle and conical aerospike nozzle, respectively. The significance of Case 2
is to test the ability of OpenFOAM solvers to predict the flow characteristics in complex geometries, and
that of the Case 3 is to test the ability of OpenFOAM solvers to solve supersonic flow through a two-
dimensional axisymmetric nozzle. A comparison of turbulence models has been conducted on 5.7° planar
CD nozzle and consequently implemented on the 10.7° planar CD nozzle, planar ED nozzle, and conical
aerospike nozzle. Two unmodified compressible flow solvers, namely sonicFOAM, and rhoCentralFOAM,
have been used for the performance comparison with ANSYS Fluent. The Mach number and density
gradient magnitude contours are post-processed using ParaView 5.4.1.

Table 1 Pressure, velocity, and temperature boundary conditions

Pressure (P) Velocity (U) Temperature (T)


Inlet totalPressure zeroGradient totalTemperature
Outlet waveTransmissive waveTransmissive waveTransmissive
Wall zeroGradient noSlip zeroGradient
P. P. Nair

Fig. 4 Comparison of wall pressure measurements with different turbulence models and the experimental results (Verma and
Manisankar 2014) at NPR 2.9 for a sonicFOAM, and b rhoChentralFOAM

Fig. 5 Comparison of experimental (Verma and Manisankar 2014) and numerical wall pressure measurements in 5.7° planar
nozzle at different NPR

3.1 Case 1 – planar CD nozzle

The turbulence models have been compared for 5.7° planar CD nozzle and validated using experimental
data by Verma (2019) for NPR 2.9. At NPR 2.9, the flow separation is symmetric within the nozzle.
Figure 4a shows the ratio of wall pressure to the stagnation jet pressure plotted with respect to the ratio of
the x-axis to throat height for the sonicFOAM solver. The flow separation in the experiment is observed at
X/Th of 2.63. The flow separation location captured by standard k-e, realizable k-e, SST k-x, and SA models
had a difference of 0.69, 0.22, 0.21, 1.26, respectively. It is observed that SST k-x predicts the flow
separation satisfactorily in comparison to standard k-e, realizable k-e, and SA models. Figure 4b shows the
ratio of wall pressure to the stagnation jet pressure plotted with respect to the ratio of the x-axis to throat
height for the rhoCentralFOAM solver. The flow separation location captured by standard k-e, realizable
Prediction and visualization of supersonic nozzle

Fig. 6 Flow structure in 5.7° planar nozzle; Experimental schlieren image (Verma and Manisankar 2014): a NPR 2.4 b NPR
2.9, Mach number contour of Fluent: c NPR 2.4 d NPR 2.9, Mach number contour of rhoCentralFOAM: e NPR 2.4 f NPR 2.9,
Mach number contour of sonicFOAM: g NPR 2.4 h NPR 2.9

k-e, SST k-x, and SA models had a difference of 0.28, 1.12, 0.05, and 0.39, respectively. It is observed that
SST k-x predicts the flow separation location accurately. Therefore, the SST k-x turbulence model is
chosen for further study.
Figure 5 shows wall pressure comparison for NPR 2.4 and NPR 2.9 for 5.7° planar CD nozzle using
ANSYS Fluent (Nair et al. 2020ab), sonicFOAM, and rhoCentralFOAM. The numerical simulations using
ANSYS Fluent, sonicFOAM, and rhoCentralFOAM are in good match with the experimental data. It is
observed that although Fluent and rhoCentralFOAM capture flow separation location accurately for NPR 2.4
and 2.9, but it overstated the pressure rise after the flow separation. The difference in pressure rise for fluent
was 0.055, and 0.021 for NPR 2.4, and 2.9, respectively. The difference in pressure rise for rhoCen-
tralFOAM was 0.0356, and 0.014 for NPR 2.4, and 2.9, respectively. In the case of sonicFOAM, flow
separation is not accurately captured, but pressure rise after the flow separation is accurately captured. The
difference in pressure was 0.17 and 0.21 for NPR 2.4 and 2.9, respectively. Fluent had overstated the
P. P. Nair

Fig. 7 Comparison of experimental (Verma and Manisankar 2014) and numerical wall pressure measurements in 10.7° planar
nozzle at different NPR

pressure decrease due to shock reflection by a difference of 0.04, but it is accurately predicted by sonic-
FOAM and rhoCentralFOAM.
Figure 6 shows a qualitative study by comparing the experimental schlieren image, and the Mach
number contour for NPR 2.4 and 2.9. The internal shocks originating from the throat where the divergent
section starts have been strongly captured in Fluent and rhoCentralFOAM. However, in the case of
sonicFOAM internal shocks are feebly captured. The strength of the reflected shock in Fluent is stronger in
comparison to the rhoCentralFOAM, and hence, the overstated dip in pressure is observed in the pressure
profile. The flow features, asymmetric flow separation, Mach stem, and shocks caught by both the CFD
packages are similar to that of the experimental schlieren image. The CFD packages have accurately
captured the flow asymmetry, shock reflections, overexpansion shock, FSS, and RSS. In NPR 2.4, RSS is
seen at the bottom wall and FSS on the top wall. At NPR 2.9, due to the end effect, FSS is on the top and
bottom walls. It should be noted that at a different run, the flow could attach to either top or the bottom wall.
It is observed that once the flow attaches to the upper or lower surface, it keeps on attaching to it for the
remaining time of the simulation. This was stated as the Coanda effect by Papamoschou et al. (2009).
Figure 7 shows wall pressure comparison for NPR 1.82 and NPR 2.4 for 10.7° planar CD nozzle using
ANSYS Fluent, sonicFOAM, and rhoCentralFOAM. The computational results are compared with an
experimental study conducted by Verma and Manisankar (2014). Fluent has accurately captured the flow
separation region both at low NPR and high NPR. Yet again, it overestimated the pressure rise by 0.036 for
NPR 2.4 after the flow separation. In the case of sonicFOAM, flow separation is not accurately captured,
having a difference of 0.145, and 0.11 for NPR 1.82, and 2.4, respectively; however, the pressure rise after
the flow separation was accurately captured. The rhoCentralFOAM is seen to be overestimating the sepa-
ration location by a difference of 0.107, and 0.15 for NPR 1.82, and 2.4, respectively.
The flow visualization using Mach number contour of Fluent (Nair et al. 2020a, 2020 ), sonicFOAM, and
rhoCentralFOAM for NPR 1.82 and 2.4 is compared with the experimental (Verma and Manisankar 2014)
schlieren image in Fig. 8. The internal shocks similar to 5.7° planar CD nozzle are strongly captured in
Fluent and rhoCentralFOAM, but, in the case of sonicFOAM, internal shocks are weakly captured. The flow
features, Mach stem, and shocks caught by both the CFD packages are similar to that of the experimental
schlieren image. The asymmetry of flow separation at low NPR is well captured by both the CFD packages.
A similar phenomenon is observed in the case of 10.7° planar CD nozzle. When the simulation was
restarted, the flow is be seen to be attached to either top or the bottom wall. Once it attaches to the top or
bottom wall, it retains its position till the end of the simulation. At NPR 1.82, the flow begins to bend toward
the upper wall due to the Coanda effect. The FSS is observed in the bottom wall, and the overexpansion
shock that is seen to be originating from the bottom wall. The k-shock is observed on the top wall that
causes the reattachment of the flow on to the nozzle wall leading to entrapment of the recirculation bubble.
This leads to the formation of RSS on the top wall. At NPR 2.4, the flow becomes symmetric. Due to the end
effect, the RSS changes to FSS on the upper portion of the wall.
Prediction and visualization of supersonic nozzle

Fig. 8 Flow structure in 10.7° planar nozzle; experimental schlieren image (Verma and Manisankar 2014): a NPR 1.82 b NPR
2.4, Mach number contour of Fluent: c NPR 1.82 d NPR 2.4, Mach number contour of rhoCentralFOAM: e NPR 1.82 f NPR
2.4, Mach number contour of sonicFOAM: g NPR 1.82 h NPR 2.4
P. P. Nair

Fig. 9 Comparison of experimental (Wagner and Schlechtriem 2011) and numerical wall pressure measurements for a nozzle
wall, and b pintle wall in planar ED nozzle at different NPR

3.2 Case 2 – planar ED nozzle

Planar ED nozzle is tested using Fluent (Paul et al. 2020), sonicFOAM, and rhoCentralFOAM for low and
high NPR of 10 and 30, respectively. These flow regimes fall under open wake and closed wake regimes of
ED nozzle. At low altitudes, high ambient pressure at the exit of the nozzle in comparison to the nozzle
design pressure, leads to overexpansion. Due to the presence of the pintle, the flow is deflected toward the
nozzle wall. Therefore, the flow separation on the nozzle walls is avoided. During this process, the base of
the pintle is exposed to the ambient pressure, which affects the pintle base pressure, consequently affecting
the performance of the nozzle. As the altitude increases, the shear layer at the edge of the pintle moves close
to the centerline of the nozzle. This cuts off the interaction of ambient pressure with the pintle base. This
leads to a closed wake regime. Therefore, the base of the pintle is no longer affected by ambient pressure.
For the comparison of planar ED nozzle, NPR from each flow regime is considered. The result of the
numerical simulation has been compared with the experimental result of Wagner and Schlechtriem (2011).
Figure 9a shows the non-dimensional nozzle wall pressure with respect to the non-dimensional x-axis. It is
observed that Fluent and rhoCentralFOAM accurately match the experimental result. In the case of
sonicFOAM, till X/Th 1.7, the pressure profile matches with that of the experimental data. The sonicFOAM
incorrectly predicts the shock originating from the pintle, which impinges on the nozzle surface, leading to
an increase in the nozzle pressure for the remaining part of the nozzle profile, as shown in Fig. 9a. It should
be noted that the initial condition for the simulation of rhoCentralFOAM was taken from the end result of
sonicFOAM as rhoCentralFOAM was encountering divergence issues at the initial stages of iteration when
the flow is initialized from the inlet.
While comparing the pressure at the pintle and centerline, it is found that Fluent accurately predicted the
pintle and centerline pressure in comparison to sonicFOAM, and rhoCentralFOAM. In the case of Open-
FOAM solvers, the shear layer impingement is incorrectly predicted for the closed wake regime. This leads
to an increase in pressure toward the end of the centerline for the sonicFOAM, and rhoCentralFOAM as
observed in Fig. 9b. The pressure dip due to the expansion fan over the pintle is also overestimated by the
sonicFOAM, and rhoCentralFOAM. It could be noted that rhoCentralFOAM had predicted the pintle
pressure better in comparison to sonicFOAM.
Figure 10 provides a qualitative view of the flow through a planar ED nozzle. The density gradient
obtained from Fluent, sonicFOAM, and rhoCentralFOAM is compared with the experimental schlieren
image. The flow features are accurately captured in Fluent and rhoCentralFOAM. Flow features such as
shock reflection, shear layer impingement, and expansion fan are incorrectly captured in the sonicFOAM
Prediction and visualization of supersonic nozzle

Fig. 10 Flow structure in planar ED nozzle; experimental schlieren image (Wagner and Schlechtriem 2011): a NPR 10 b NPR
30, density gradient contour of Fluent: c NPR 10 d NPR 30, density gradient contour of rhoCentralFOAM: e NPR 10 f NPR 30,
density gradient contour of sonicFOAM: g NPR 10 h NPR 30
P. P. Nair

Fig. 11 Comparison of experimental Verma (2009) and numerical wall pressure measurements in conical aerospike nozzle at
different NPR

both for NPR 10 as well as NPR 30. In Fig. 10, it can be seen that the shear layer impingement happens near
the end of the centerline just before the nozzle exit. In Fig. 10, it can be observed that the shear layer
impingement happens ahead of the end of the centerline.

3.3 Case 3 – conical aerospike nozzle

Conical aerospike nozzle is tested for a two-dimensional axisymmetric model using Fluent (Nair et al.
2019a, b), sonicFOAM, and rhoCentralFOAM for two overexpanded cases of NPR 2.1 and 3.82, respec-
tively. The simulated results have been compared with the experimental study conducted by Verma (2009).
Since the flow is in the overexpansion regime, the flow experiences overexpansion shock on the surface of
the nozzle, leading to RSS. The ratio of wall pressure to the stagnation jet pressure with respect to axial
length to the length of the nozzle is plotted in Fig. 11. The wall pressure has been accurately predicted by
the Fluent for NPR 2.1. The flow separation location captured by sonicFOAM is the same as Fluent, and the
pressure rise due to flow separation is also accurately captured. However, the pressure rise due to a k-shock
is overestimated. The rhoCentralFOAM did not accurately captured the k-shock, which led to a further dip
in pressure on the nozzle wall surface. Nevertheless, the pressure rise in the recirculation region and
corresponding separation regions are accurately captured. For NPR 3.82, Fluent under predicted the com-
pression region, which is seen as the hump between X/L 0.4 and 0.6. Nonetheless, it accurately predicted the
pressure rise due to flow separation and the recirculation region. In contrast to Fluent, sonicFOAM, and
rhoCentralFOAM accurately captured the pressure rise due to the compression region. However, sonic-
FOAM and rhoCentralFOAM, over-predicted the pressure rise due to flow separation and the recirculation
region.
Figure 12 shows the comparison with flow features of the experiment by Verma (2009), and numerical
results obtained through Fluent (Nair et al. 2019a, b), sonicFOAM, and rhoCentralFOAM at NPR 2.1, and
3.82. At NPR 2.1, the overexpansion shock originating from the nozzle surface interacts with the overex-
pansion shock from the tip of the cowl, forming a k-shock pattern. This leads to RSS, as a trapped
recirculation bubble is observed. Fluent and sonicFOAM accurately capture the k-shock pattern, as seen in
the experiment. The k-shock pattern is not accurately captured by the rhoCentralFOAM.
Nevertheless, the overexpansion fan both from the nozzle surface and cowl tip is precisely captured. The
slipstream at the triple point is also captured. At NPR 3.82, the internal shocks are seen arising from the
throat of the nozzle, due to underexpansion of the inner nozzle leading to the formation of a compression
region on the nozzle surface. Expansion fan can be seen emerging from the cowl tip leading to an
Prediction and visualization of supersonic nozzle

Fig. 12 Flow structure in conical aerospike nozzle; experimental schlieren image Verma (2009): a NPR 2.1 b NPR 3.82, Mach
number contour of Fluent: c NPR 1.82 d NPR 2.4, Mach number contour of rhoCentralFOAM: e NPR 2.1 f NPR 3.82, Mach
number contour of sonicFOAM: g NPR 2.1 h NPR 3.82
P. P. Nair

overexpansion shock. The overexpansion shock leads to flow separation, and reattachment of flow conse-
quently leading to the formation of a recirculation region. The overexpansion shock is captured by Fluent,
sonicFOAM, and rhoCentralFOAM. For sonicFOAM, it is found that flow did not reattach to the surface of
the nozzle leading to FSS rather than RSS. The RSS is captured by Fluent as well as rhoCentralFOAM.

4 Conclusions

Numerical simulation is performed on 5.7° and 10.7° planar CD nozzle, planar ED nozzle, and conical
aerospike nozzle for low and high NPR. The numerical results have been compared both qualitatively and
quantitatively with the experimental results. The three geometries are evaluated to test the ability of
OpenFOAM compressible solvers to solve and capture the flow feature of simple, complex, and axisym-
metric geometries. Unmodified compressible solvers were selected in OpenFOAM, namely sonicFOAM,
and rhoCentralFOAM. The numerical results of sonicFOAM and rhoCentralFOAM were compared with the
results from ANSYS Fluent and experimental data. The code validation was done by comparing the
experimental result with fluent, sonicFOAM, and rhoCentralFOAM. It was observed that for simple planar
geometries, the flow features were accurately captured by sonicFOAM and rhoCentralFOAM. However, the
details of shock reflections were crisp in rhoCentralFOAM. For 5.7° planar CD nozzle, wall pressure was in
good agreement with the experimental data. For 10.7° planar CD nozzle, it was observed that flow pattern,
slip lines, and shock structures were well captured, yet the flow separation location was over predicted.
While comparing the case of planar ED nozzle, it was observed that sonicFOAM incorrectly predicts the
flow pattern and shock structure and under-predicted the impingement of the shear layer onto the centerline
at higher NPR. The wall pressure predicted by sonicFOAM was also inaccurate. However, it may be noted
that the end solution of the sonicFOAM can be supplied as an initial value for rhoCentralFOAM. The
rhoCentralFOAM had better predicted the shock structure, expansion fan, and the impingement of the shear
layer. The nozzle wall pressure was accurately predicted by rhoCentralFOAM; however, the pressure over
the pintle was over predicted.
In the case of the conical aerospike nozzle, at low NPR, rhoCentralFOAM fails to capture the k-shock
arising due to overexpansion shock accurately. However, it accurately predicts the nozzle wall pressure
before and after separation. The sonicFOAM captures the k-shock but over-predicts the pressure rise due to
the recirculation zone in contrast to Fluent. At higher NPR, flow separation, compression region, and
recirculation regions were accurately predicted by rhoCentralFOAM in contrary to Fluent. Nevertheless, the
pressure rise due to the recirculation zone was over predicted. The sonicFOAM did not capture the recir-
culation zone; instead, FSS was predicted at higher NPR.
The overall prediction of rhoCentralFOAM was comparable with that of the ANSYS Fluent. Therefore,
rhoCentralFOAM is suggested for supersonic nozzle flows. Future studies could be based on using LES or
RANS-LES hybrid models. Modified or user-defined solvers could also be tested for better prediction of
flows.

References

Abed N, Afgan I, Cioncolini A, Iacovides H, Nasser A, Mekhail T (2020) Thermal performance evaluation of various
nanofluids with non-uniform heating for parabolic trough collectors. Case Stud Therm Eng 22:100769. https://doi.org/10.
1016/j.csite.2020.100769
Agarwal R (1999) Computational fluid dynamics of whole body aircraft. Annu Rev Fluid Mech 31:125–169. https://doi.org/10.
1146/annurev.fluid.31.1.125
Ali AE, Afgan I, Laurence D, Revell A (2021) A dual-mesh hybrid RANS-LES simulation of the buoyant flow in a
differentially heated square cavity with an improved resolution criterion. Comput Fluids 224:104949. https://doi.org/10.
1016/j.compfluid.2021.104949
Ashton N, Skaperdas V (2019) Verification and validation of OpenFOAM for high-lift aircraft flows. J Aircr 56(4):1641–1657.
https://doi.org/10.2514/1.C034918
Cao Y, Tamura T (2016) Large-eddy simulations of flow past a square cylinder using structured and unstructured grids.
Comput Fluids 137:36–54. https://doi.org/10.1016/j.compfluid.2016.07.013
Cao Z, White C, Kontis K (2021) Numerical investigation of rarefied vortex loop formation due to shock wave diffraction with
the use of rorticity. Phys Fluids 33(6):067112. https://doi.org/10.1063/5.0054289
Cao Z, Agir MB, White C, Kontis K (2022) An open source code for two-phase rarefied flows: rarefiedMultiphaseFoam.
Comput Phys Commun 276:108339. https://doi.org/10.1016/j.cpc.2022.108339
Prediction and visualization of supersonic nozzle

Choudhury SP, Suryan A, Pisharady JC, Jayashree A, Rashid K (2018) Parametric study of supersonic film cooling in dual bell
nozzle for an experimental air–kerosene engine. Aerosp Sci Technol 78:364–376. https://doi.org/10.1016/j.ast.2018.04.
038
Chutkey K, Viji M, Verma SB (2018) Interaction of external flow with linear cluster plug nozzle jet. Shock Waves
28(6):1207–1221. https://doi.org/10.1007/s00193-018-0849-6
Constant E, Favier J, Meldi M, Meliga P, Serre E (2017) An immersed boundary method in OpenFOAM: verification and
validation. Comput Fluids 157:55–72. https://doi.org/10.1016/j.compfluid.2017.08.001
D’Alessandro V, Montelpare S, Ricci R (2016) Detached–eddy simulations of the flow over a cylinder at Re= 3900 using
OpenFOAM. Comput Fluids 136:152–169. https://doi.org/10.1016/j.compfluid.2016.05.031
Damgaard T, Östlund J, Frey M (2004) Side-Load phenomena in highly overexpanded rocket nozzles. J Propuls Power
20(4):695–704. https://doi.org/10.2514/1.3059
Doolan C (2009) Flow and noise simulation of the NASA tandem cylinder experiment using OpenFOAM, Proc. of 15th AIAA/
CEAS Aeroacoustics Conference (30th AIAA Aeroacoustics Conference), Miami, Florida, USA, AIAA Paper 2009–3157,
2009 https://doi.org/10.2514/6.2009-3157
Droeske N, Makowka K, Nizenkov P, Vellaramkalayil JJ, Sattelmayer T, von Wolfersdorf J (2014) Validation of a novel
OpenFOAM solver using a supersonic, non-reacting channel flow, Proc. of 19th AIAA International Space Planes and
Hypersonic Systems and Technologies Conference, Atlanta, GA, USA, AIAA Paper 2014–3088 https://doi.org/10.2514/6.
2014-3088
Flores F, Garreaud R, Muñoz RC (2014) OpenFOAM applied to the CFD simulation of turbulent buoyant atmospheric flows
and pollutant dispersion inside large open pit mines under intense isolation. Comput Fluids 90:72–87. https://doi.org/10.
1016/j.compfluid.2013.11.012
Fluent user’s guide, (2013) ANSYS, Inc., Canonsburg, PA:1–1146
Frey M, Hagemann G (1999) Flow separation and side-loads in rocket nozzles, Proc of 35th Joint Propulsion Conference and
Exhibit, Los Angeles, CA, USA, AIAA Paper 1999–2815 https://doi.org/10.2514/6.1999-2815
Frey M, Hagemann G (2000) Restricted shock separation in rocket nozzles. J Propuls Power 16(3):478–484. https://doi.org/10.
2514/2.5593
George J, Nair PP, Soman S, Suryan A, Kim HD (2021) Visualization of flow through planar double divergent nozzles by
computational method. J vis 24(4):711–732. https://doi.org/10.1007/s12650-020-00729-9
Gramola M, Bruce PJK, Santer M (2020) Off-design performance of 2D adaptive shock control bumps. J Fluids Struct
93:102856. https://doi.org/10.1016/j.jfluidstructs.2019.102856
Greenshields CJ (2020) OpenFOAM User Guide v8.0. OpenFOAM Foundation Ltd. http://foam.sourceforge.net/docs/Guides-
a4/OpenFOAMUserGuide-A4.pdf. Assessed on 1 August 2020
Hagemann G, Immich H, Nguyen TV, Dumnov GE (1998) Advanced rocket nozzles. J Propuls Power 14(5):620–634. https://
doi.org/10.2514/2.5354
Han X, Li J, Morgans AS (2015) Prediction of combustion instability limit cycle oscillations by combining flame describing
function simulations with a thermoacoustic network model. Combust Flame 162(10):3632–3647. https://doi.org/10.1016/j.
combustflame.2015.06.020
Han X, Laera D, Yang D, Zhang C, Wang J, Hui X, Lin Y, Morgans AS, Sung CJ (2020) Flame interactions in a stratified swirl
burner: flame stabilization, combustion instabilities and beating oscillations. Combust Flame 212:500–509. https://doi.org/
10.1016/j.combustflame.2019.11.020
Hunter CA (2004) Experimental investigation of separated nozzle flows. J Propuls Power 20(3):527–532. https://doi.org/10.
2514/1.4612
Hunter C (1998) Experimental, theoretical, and computational investigation of separated nozzle flows, Proc. of 34th AIAA/
ASME/SAE/ASEE Joint Propulsion Conference and Exhibit, Cleveland, OH, USA, AIAA Paper 98–3107 https://doi.org/
10.2514/6.1998-3107
Jinks ER, Bruce PJK, Santer M (2018) Optimisation of adaptive shock control bumps with structural constraints. Aerosp Sci
Technol 77:332–343. https://doi.org/10.1016/j.ast.2018.03.018
Jinks ER, Bruce PJK, Santer M, (2014) Adaptive shock control bumps, In: 52nd Aerospace Sciences Meeting, AIAA
2014–0945, (p 0945) https://doi.org/10.2514/6.2014-0945
Kadu PA, Sakai Y, Ito Y, Iwano K, Sugino M, Katagiri T, Nagata K (2019) Numerical investigation of passive scalar transport
and mixing in a turbulent unconfined coaxial swirling jet. Int J Heat Mass Transf 142:118461. https://doi.org/10.1016/j.
ijheatmasstransfer.2019.118461
Kadu PA, Sakai Y, Ito Y, Iwano K, Sugino M, Katagiri T, Hayase T, Nagata K (2020) Application of spectral proper
orthogonal decomposition to velocity and passive scalar fields in a swirling coaxial jet. Phys Fluids 32(1):015106. https://
doi.org/10.1063/1.5131627
Karthikeyan N, Kumar A, Verma SB, Venkatakrishnan L (2013) Effect of spike truncation on the acoustic behavior of annular
aerospike nozzles. AIAA J 51(9):2168–2182. https://doi.org/10.2514/1.J052139
Kurganov A, Tadmor E (2000) New high-resolution central schemes for nonlinear conservation laws and convection–diffusion
equations. J Comput Phys 160(1):241–282
Kurganov A, Noelle S, Petrova G (2001) Semidiscrete central-upwind schemes for hyperbolic conservation laws and Hamilton-
Jacobi equations. SIAM J Sci Comput 23(3):707–740. https://doi.org/10.1137/S1064827500373413
Lysenko DA, Ertesvåg IS, Rian KE (2010) Modeling of turbulent separated flows using OpenFOAM. Comput Fluids
80:408–422. https://doi.org/10.1016/j.compfluid.2012.01.015
Mukundhan D, Kumar R, (2017) Preliminary design and optimization of 2D supersonic intake using OpenFOAM, Proc. of 30th
International Symposium on Shock Waves 2, Springer, Cham: 1047–1051 https://doi.org/10.1007/978-3-319-44866-4_46
Muntean S, Nilsson H, Susan-Resiga R (2009) 3D numerical analysis of the unsteady turbulent swirling flow in a conical
diffuser using Fluent and OpenFOAM, Proc. of 3rd IAHR International Meeting of the Workgroup on Cavitation and
Dynamic Problem in Hydraulic Machinery and Systems, Brno, Czech Republic
P. P. Nair

Nair PP, Suryan A, Kim HD (2017) Computational study of performance characteristics for truncated conical aerospike
nozzles. J Therm Sci 26(6):483–489. https://doi.org/10.1007/s11630-017-0965-0
Nair PP, Suryan A, Kim HD (2019a) Study of conical aerospike nozzles with base-bleed and freestream effects. J Spacecr
Rockets 56(4):990–1005. https://doi.org/10.2514/1.A34256
Nair PP, Suryan A, Kim HD (2019b) Computational study on flow through truncated conical plug nozzle with base bleed.
Propuls Power Res 8(2):108–120. https://doi.org/10.1016/j.jppr.2019.02.001
Nair PP, Suryan A, Kim HD (2020a) Computational study on reducing flow asymmetry in over-expanded planar nozzle by
incorporating double divergence. Aerosp Sci Technol 100:105790. https://doi.org/10.1016/j.ast.2020.105790
Nair PP, Suryan A, Chandran R (2020b) A numerical study on planar nozzles with different divergence angles, Recent asian
research on thermal and fluid sciences, Springer, Singapore, 133-146 https://doi.org/10.1007/978-981-15-1892-8_12
Nakao S, Kashitani M, Miyaguni T, Yamaguchi Y (2014) A study on high subsonic airfoil flows in relatively high reynolds
number by using openfoam. J Therm Sci 23(2):133–137. https://doi.org/10.1007/s11630-014-0687-5
Nave LH, Coffey GA (1973) Sea level side loads in high-area-ratio rocket engines, Proc of 9th Propulsion Conference, Las
Vegas, NV, USA, AIAA Paper 1973–1284. https://doi.org/10.2514/6.1973-1284
Palharini RC, White C, Scanlon TJ, Brown RE, Borg MK, Reese JM (2015) Benchmark numerical simulations of rarefied non-
reacting gas flows using an open-source DSMC code. Comput Fluids 120:140–157. https://doi.org/10.1016/j.compfluid.
2015.07.021
Papamoschou D, Zill A, Johnson A (2009) Supersonic flow separation in planar nozzles. Shock Waves 19(3):171–183. https://
doi.org/10.1007/s00193-008-0160-z
Papamoschou D, Zill A (2004) Fundamental investigation of supersonic nozzle flow separation, Proc. of 42nd AIAA
Aerospace Sciences Meeting and Exhibit, Reno, Nevada , USA, AIAA Paper 2004–1111 https://doi.org/10.2514/6.2004-
1111
Paul PJ, Nair PP, Suryan A, Martin MJP, Kim HD (2020) Numerical simulation on optimization of pintle base shape in planar
expansion-deflection nozzles. J Spacecr Rockets 57(3):539–548. https://doi.org/10.2514/1.A34559
Putra YS, Beaudoin A, Rousseaux G, Thomas L, Huberson S (2019) 2D numerical contributions for the study of non-cohesive
sediment transport beneath tidal bores. Comptes Rendus Mécanique 347(2):166–180. https://doi.org/10.1016/j.crme.2018.
11.004
Rabbani HS, Joekar-Niasar V, Shokri N (2016) Effects of intermediate wettability on entry capillary pressure in angular pores.
J Colloid Interface Sci 473:34–43. https://doi.org/10.1016/j.jcis.2016.03.053
Robertson E, Choudhury V, Bhushan S, Walters DK (2015) Validation of OpenFOAM numerical methods and turbulence
models for incompressible bluff body flows. Comput Fluids 123:122–145. https://doi.org/10.1016/j.compfluid.2015.09.
010
Ruf, J., McDaniels, D. & Brown A (2010) Details of side load test data and analysis for a truncated ideal contour nozzle and a
parabolic contour nozzle, Proc of 46th AIAA/ASME/SAE/ASEE Joint Propulsion Conference & Exhibit, Nashville, TN,
USA, AIAA Paper 2010–6813 https://doi.org/10.2514/6.2010-6813
Soman S, Suryan A, Nair PP, Kim HD (2021) Numerical analysis of flowfield in linear plug nozzle with base bleed. J Spacecr
Rockets 58(6):1786–1798. https://doi.org/10.2514/1.A34992
Stoldt H, Johansen, CT, Korobenko A, Ziade P (2020) Verification and validation of a high-fidelity open-source simulation tool
for supersonic aircraft aerodynamic analysis, Proc. of 19th AIAA aviation 2020 forum, Virtual event, AIAA Paper
2020-2758. https://doi.org/10.2514/6.2020-2758
Sutherland W (1893) The viscosity of gases and molecular force, philosophical magazine series 5, 36 (223):507–531 https://
doi.org/10.1080/14786449308620508
Suzuki YJ, Koyaguchi T (2013) 3D numerical simulation of volcanic eruption clouds using the 2011 Shinmoe-dake eruptions.
Earth Planets Space 65(10):581–589. https://doi.org/10.5047/eps.2013.03.009
Taylor NV, Hempsell CM, Macfarlane J, Osborne R, Varvill R, Bond A, Feast S (2010) Experimental investigation of the
evacuation effect in expansion deflection nozzles. Acta Astronaut 66(3–4):550–562. https://doi.org/10.1016/j.actaastro.
2009.07.016
Taylor N, Steelant J, Bond R (2011) Experimental comparison of dual bell and expansion deflection nozzles, Proc. of 47th
AIAA/ASME/SAE/ASEE Joint Propulsion Conference & Exhibit, San Diego, California, USA, AIAA Paper 2011–5688
https://doi.org/10.2514/6.2011-5688
Tunstall R, Laurence D, Prosser R, Skillen A (2017) Towards a generalised dual-mesh hybrid LES/RANS framework with
improved consistency. Comput Fluids 157:73–83. https://doi.org/10.1016/j.compfluid.2017.08.002
Verma SB (2009) Performance characteristics of an annular conical aerospike nozzle with freestream effect. J Propuls Power
25(3):783–791. https://doi.org/10.2514/1.40302
Verma SB, Manisankar C (2014) Origin of flow asymmetry in planar nozzles with separation. Shock Waves 24(2):191–209.
https://doi.org/10.1007/s00193-013-0492-1
Verma SB, Viji M (2011) Freestream effects on base pressure development of an annular plug nozzle. Shock Waves
21(2):163–171. https://doi.org/10.1007/s00193-011-0305-3
Verma SB, Stark R, Haidn O (2006) Relation between shock unsteadiness and the origin of sideloads inside a thrust optimized
parabolic rocket nozzle. Aerosp Sci Technol 10(6):474–483. https://doi.org/10.1016/j.ast.2006.06.004
Verma SB, Stark R, Haidn O (2017) Origin of side-loads in a subscale truncated ideal contour nozzle. Aerosp Sci Technol
71:725–732. https://doi.org/10.1016/j.ast.2017.10.014
Vuorinen V, Keskinen JP, Duwig C, Boersma BJ (2014) Boersma, On the implementation of low-dissipative Runge-Kutta
projection methods for time dependent flows using OpenFOAM. Comput Fluids 93:153–163. https://doi.org/10.1016/j.
compfluid.2014.01.026
Wagner B, Stark R, Schlechtriem S (2011) Experimental study of a planar expansion-deflection nozzle. Progress Propul Phys
2:641–654. https://doi.org/10.1051/eucass/201102641
Prediction and visualization of supersonic nozzle

Wagner B, Schlechtriem S (2011) Numerical and Experimental Study of the Flow in a Planar Expansion-Deflection Nozzle,
Proc. of 47th AIAA/ASME/SAE/ASEE Joint Propulsion Conference & Exhibit, San Diego, California, USA, AIAA Paper
2011–5942 https://doi.org/10.2514/6.2011-5942
Wang M, Freund B, Lele SK (2006) Computational prediction of flow-generated sound. Annu Rev Fluid Mech 38:483–512.
https://doi.org/10.1146/annurev.fluid.38.050304.092036
Wang Y, Xu J, Huang S, Lin Y, Jiang J (2019) Computational study of axisymmetric divergent bypass dual throat nozzle.
Aerosp Sci Technol 86:177–190. https://doi.org/10.1016/j.ast.2018.11.059
White C, Borg MK, Scanlon TJ, Longshaw SM, John B, Emerson DR, Reese JM (2018) dsmcFoam?: an OpenFOAM based
direct simulation Monte Carlo solver. Comput Phys Commun 224:22–43. https://doi.org/10.1016/j.cpc.2017.09.030
Wojewodka MM, White C, Shahpar S, Kontis K (2022) Numerical study of complex flow physics and coherent structures of
the flow through a convoluted duct. Aerosp Sci Technol 121:107191. https://doi.org/10.1016/j.ast.2021.107191
Xiao Q, Tsai HM, Papamoschou D (2007) Numerical investigation of supersonic nozzle flow separation. AIAA J
45(3):532–541. https://doi.org/10.2514/1.20073
Yang WJ, Yi W, Ren XG, Xu LY, Xu XH, Yuan XF (2015) Toward large scale parallel computer simulation of viscoelastic
fluid flow: a study of benchmark flow problems. J Non-Newton Fluid Mech 222:82–95. https://doi.org/10.1016/j.jnnfm.
2014.09.004
Zang B, Vevek US, Lim HD, Wei X, New TH (2018a) An assessment of OpenFOAM solver on RANS simulation of round
supersonic free jet. J Comput Sci 28:18–31. https://doi.org/10.1016/j.jocs.2018.07.002
Zang B, Vevek US, Lim HD, Wei X, New TH (2018b) An assessment of OpenFOAM solver on RANS simulations of round
supersonic free jets. J Comput Sci 28:18–31. https://doi.org/10.1016/j.jocs.2018.07.002
Zmijanovic V, Leger L, Sellam M, Chpoun A (2018) Assessment of transition regimes in a dual-bell nozzle and possibility of
active fluidic control. Aerosp Sci Technol 82:1–8. https://doi.org/10.1016/j.ast.2018.02.003

Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional
affiliations.

View publication stats

You might also like