Académique Documents
Professionnel Documents
Culture Documents
1
Numerical Simulation of a Supersonic Cruise Nozzle
Balasubramanyam Sasanapuri
1
ANSYS Fluent India Pvt Ltd., Pune - 411057, India
Manish Kumar
2
, Sutikno Wirogo
3
and Konstantin A. Kurbatskii
4
ANSYS Inc., Lebanon, NH 03766, USA
The prototyping and testing of a supersonic cruise nozzle that covers wide range of
nozzle geometry configurations (to help with large operating envelop of an aircraft) is both
time consuming and expensive. Numerical simulations offer quick and less expensive
solution to reduce the design time and cost. In the present study one of the configurations of
a supersonic cruise nozzle is simulated for a range of nozzle pressure ratios and the results
are compared with experimental data. The pressure-based and density-based solvers in
ANSYS Fluent CFD code are used for the validation study and solution based adaption is
examined to determine if the accuracy can be improved by local mesh refinement. The
simulation results show very good agreement with the experimental data, and this study
demonstrates an optimized simulation process which can be used to study the entire
envelope of flow and nozzle geometry conditions.
I. Introduction
A supersonic cruise aircraft must be capable of operating over a wide range of altitude and velocity, which
includes subsonic take-off and landing, subsonic cruise, climb and supersonic cruise. One of the solutions for
meeting these varied requirements is a variable-cycle engine, which uses variable-geometry nozzle and combustion
arrangement to operate like a turbofan or turbojet or a hybrid combination to suite the mission requirement. The
design of such a variable geometry nozzle requires testing for a broad envelope of flow conditions and geometry
variations. Prototyping and testing for such an envelope would be very time consuming and expensive.
Computation Fluid Dynamics (numerical simulation) offers a faster and cheaper solution to reduce the design time
and cost. In the present study one of supersonic cruise nozzle configurations is simulated for a range of Nozzle
Pressure Ratio (NPR) values and the results are compared with experimental data. The pressure-based coupled
solver (PBCS) and density-based coupled solver (DBNS) formulations implemented in the general purpose CFD
code (ANSYS Fluent
1
) are used for the validation study. Solution based adaption is applied to determine if the
accuracy can be improved. Exploratory calculations were done for a free stream Mach number of 0.6 and NPR 2.5,
and the best solution process established from these calculations were then used for a second set of cases with zero
free stream Mach number and a NPR range from 2.5 to 7.0. The problem is described in detail in Sec. II, an
overview of the solver algorithm is given in Sec. III, and finally the numerical predictions are presented and
compared with experimental data in Sec. IV.
II. Problem Description
The problem considered for this study is a 2D axi-symmetric configuration of a supersonic cruise nozzle (Fig. 1).
The geometry and flow conditions correspond to one of the nozzle configurations studied experimentally in Ref. [2].
The study was carried out in two parts: the first part focused on the effects of mesh refinement and different solver
settings to get the most accurate solution in comparison to the experimental data. The second part considered a
parametric study for a series of NPRs using the best settings derived from the first part of the study.
1
Senior Technology Specialist.
2
Senior Technical Services Engineer.
3
Senior Technical Account Manager.
4
Lead Technical Services Engineer, Senior AIAA Member.
51st AIAA Aerospace Sciences Meeting including the New Horizons Forum and Aerospace Exposition
07 - 10 January 2013, Grapevine (Dallas/Ft. Worth Region), Texas
AIAA 2013-0492
Copyright 2013 by ANSYS, Inc. Published by the American Institute of Aeronautics and Astronautics, Inc., with permission.
American Institute of Aeronautics and Astronautics
2
Figure 1. 2D axi-symmetric geometry of a supersonic cruise nozzle.
III. Numerical Model
Most numerical approaches to high speed jet flows, e.g.
3-4
, employ density-based coupled formulations where the
governing equations of continuity, momentum, energy and (where appropriate) species transport are solved
simultaneously as a set, or vector, of equations. In this approach, density is used as a primary variable found from
the continuity equation, and then pressure is deduced from it using an equation of state. Density-based techniques
are found to be efficient when used for high subsonic, transonic or supersonic flows; however they require
modifications, such as preconditioning
5-6
in low Mach number flow regions (e.g. stagnation region outside the jet in
the outer domain) to overcome the problem of the system matrix becoming singular in the incompressible limit. The
density-based double-precision implicit solver
1
with preconditioning is used in this work.
As an alternative to the density-based approach, a number of coupled pressure-based methods have been
proposed
7 11
to extend applicability of pressure-based segregated techniques to problems where the inter-equation
coupling is strong. Unlike a segregated algorithm, in which the momentum equations and pressure correction
equation are solved one after another in a decoupled manner, a pressure-based coupled algorithm solves a coupled
system of equations comprising the momentum equations and pressure correction equation. Since the momentum
and pressure equations are solved in a closely coupled manner, the rate of solution convergence significantly
improves when compared to a segregated solver. The coupling also makes pressure-based coupled algorithms
applicable to supersonic and hypersonic problems, which can be very difficult to solve by a segregated approach. A
pressure-based coupled double-precision solver
1
is employed in this study to compare with the conventional density-
based solver. The solver algorithms are briefly discussed in the sections below.
A. Density-based Coupled Solver (DBNS)
The governing equations for the conservation of mass, momentum and energy are discretized using a control-
volume-based technique. The system of governing equations for a single-component fluid is cast in integral
Cartesian form for an arbitrary control volume V with differential surface area dA as follows:
} } }
= +
c
c
V V
dV d dV
t
H G F W A ] [ (1)
where the vectors W, F and G are defined as,
T
] , , , , 1 [ E w u v = W ,
T
] ,
[ v v k v j v i v v F p E p w p p u + + + + = v , ,
T
] , , , , 0 [ q G + =
j ij zi yi xi
v t t t t (2)
and the vector H contains source terms such as body forces and energy sources. Here , v, E, and p are the density,
velocity, total energy per unit mass, and pressure of the fluid, respectively, t is the viscous stress tensor, and q is the
heat flux. Total energy E is related to the total enthalpy H as E = H p / where H = h + |v|
2
/2 and h is sensible
enthalpy. The Navier-Stokes equations (1) become numerically very stiff at a low Mach number due to the disparity
between the fluid velocity and the acoustic speed of sound. The numerical stiffness of the equations under these
conditions results in poor convergence rates. This difficulty is overcome by employing time-derivative
preconditioning
6
, which modifies the time-derivative term in (1) by pre-multiplying it with a preconditioning matrix.
This has the effect of re-scaling the acoustic speed (eigenvalue) of the system of equations being solved in order to
alleviate the numerical stiffness encountered at low Mach numbers and in incompressible flows. Face values
American Institute of Aeronautics and Astronautics
3
required for computing the convection terms are interpolated from the cell centers using a second-order upwind
scheme
12
.
The inviscid flux vector F appearing in (1) is evaluated by Advection Upstream Splitting Method
13
(AUSM).
ANSYS Fluent utilizes an all-speed AUSM+ scheme
14
based on the low Mach number preconditioning. The coupled
set of governing equations (1) is discretized in time using an implicit time-marching algorithm. In the implicit
scheme, an Euler implicit discretization in time is combined with a Newton-type linearization of the fluxes to
produce a linearized system in delta form
15
. The system is solved using Incomplete Lower Upper (ILU) factorization
in conjunction with an algebraic multigrid (AMG) method
1, 16
adapted for coupled sets of equations. Time marching
proceeds until a steady-state solution is reached. Explicit relaxation is applied to improve the convergence to steady
state by controlling the amount that the solution vector changes between iterations after the end of the AMG cycle.
Gradients needed for constructing values of a scalar at the cell faces and for computing secondary diffusion
terms and velocity derivatives are calculated using the least squares cell-based gradient evaluation
1
which preserves
a second-order spatial accuracy.
B. Pressure-based Coupled Solver (PBCS)
An implicit discretization of the pressure gradient terms in the momentum equations, and an implicit
discretization of the face mass flux, including the Rhie-Chow pressure dissipation terms, provide fully implicit
coupling between the momentum and continuity equations. This discretization yields a system of algebraic equations
whose matrix depends on the discretization coefficients of the momentum equations
1
, which is then solved using the
coupled algebraic multigrid (AMG) scheme
1, 16
. An ILU smoother is applied to smooth the residuals between levels
of the AMG. The ILU smoother is more expensive than standard Gauss-Seidel, but has better smoothing properties,
especially for block-coupled systems solved by the coupled AMG, which permits more aggressive coarsening of
AMG levels.
Either a second-order upwind scheme
12
or QUICK-type scheme
1, 17
scheme is used for interpolating face values
of velocities and energy. The QUICK scheme implementation in ANSYS Fluent is based on a weighted average of
second-order-upwind and second-order central differencing of the variable. It uses a variable, solution-dependent
value of the weight factor, chosen so as to avoid introducing new solution extrema.
Face values of pressure are reconstructed using a second-order or PRESTO! scheme. The second-order
implementation is similar to a multidimensional linear reconstruction approach
12
. In this approach, higher-order
accuracy is achieved at cell faces through a Taylor series expansion of the cell-centered solution about the cell
centroid. The PRESTO! (Pressure Staggering Option) scheme
1
uses the discrete continuity balance for a staggered
control volume about the face to compute the staggered (i.e., face) pressure. This procedure is similar in spirit to
the staggered-grid schemes used with structured meshes
18
. For triangular, tetrahedral, hybrid, and polyhedral
meshes, comparable accuracy is obtained using a similar algorithm.
C. Gradient Limiters
Both DBNS and PBCS solver formulations take advantage of gradient (or slope) limiters used on the second-
order upwind scheme to prevent spurious oscillations, which would otherwise appear in the solution flow field near
shocks, discontinuities, or near rapid local changes in the flow field. The gradient limiter attempts to invoke and
enforce the monotonicity principle by prohibiting the linearly reconstructed field variable on the cell faces to exceed
the maximum or minimum values of the neighboring cells. A non-differentiable limiter
12
based on the Minmod
function (Minimum Modulus) is utilized in this study to limit and clip the reconstructed solution overshoots and
undershoots. Cell to face limiting direction is chosen, where the limited value of the reconstruction gradient is
determined at cell face centers.
D. Physical Models and Boundary Conditions
Air is modeled as a single-species ideal gas. For the NPRs considered in this study, the maximum Mach number
is expected to be below 4.0, hence real-gas thermodynamic non-equilibrium processes are not expected to have a
strong effect on aerodynamic heating, and therefore an aerothermochemical model is not taken into account in the
simulation. The numerical code ANSYS FLUENT provides capabilities to include chemical and vibrational non-
equilibrium effects when they cannot be neglected. Paterna et al
19
discussed one such example application of the
code, with chemical non-equilibrium enabled, to the Martian atmosphere entry problem.
The effects of turbulence are modeled using the shear-stress transport SST k-e model proposed by Menter
20
. It
effectively blends the robust and accurate formulation of the k-e model
21
in the near-wall region with the free-
stream independence of the k-c model
22
in the far field. To achieve this, the k-c model is converted into a k-e
American Institute of Aeronautics and Astronautics
4
formulation. The standard k-e model and the transformed k-c model are both multiplied by a blending function, and
both models are added together. The blending function is one in the near-wall region, which activates the standard k-
e model, and zero away from the surface, which activates the transformed k-c model.
The pressure inlet boundary condition at the nozzle inlet specifies static and total pressure, total temperature and
flow direction which correspond to the test conditions
2
. Boundary values of turbulent kinetic energy and its specific
dissipation rate e at the nozzle inlet are derived from the turbulence intensity I = 5% and the nozzle inlet diameter D
as,
where U
i
is calculated inlet velocity and C