Some Applications of Computational Aerodynamics to Support Guided

Weapon Design and Development

Scott SHAW, Bernard DENCHFIELD, Jon ARMSTRONG, Ben HINCHLIFFE,

Jed HOLLOM
MBDA UK Limited
UNITED KINGDOM
[email protected]

ABSTRACT
Understanding the external aerodynamics of a missile airframe is important to establishing overall system
performance. The aerodynamics of the configuration influences kinematic performance, stability, control and
manoeuvrability of the weapon and its safety during carriage and release. Characterizing the aerodynamic
loads generated by a missile airframe is therefore a fundamental aspect in the development of the missile
system.

Within the MBDA-UK aerodynamics department, the use of computational fluid dynamics to characterize the
aerodynamics of the airframe is ubiquitous. The use of computational fluid dynamics through the missile
development life cycle is illustrated by three use-cases related to the carriage and launch of an internally
carried air-to-air missile launched from a platform bay. The three use cases (i) Free air aerodynamics (ii)
characterization of the aero-acoustics of payload bays and (iii) interaction between the platform and missile
during release illustrate different aspects of the engineering use of CFD analyses at MBDA and reveal the
inherent tensions between the development of analysis tools to support engineering practice and the
development of the science of engineering simulation.

1.0 INTRODUCTION
Understanding the external aerodynamics of a missile airframe is important for establishing overall system
performance. The aerodynamic characteristics of the configuration influence kinematic performance, stability,
control and manoeuvrability of the weapon and its safety during carriage and release. Characterizing the
aerodynamic loads generated by a missile airframe is therefore a fundamental aspect in the development of the
missile system.

1.1 Historical Perspective

In the late 1980s, it was imagined that future computing performance increases would not only allow us to run
many more existing Computational Fluid Dynamics (CFD) simulations much more quickly, but also increase
the fidelity of those calculations. The vision was for a “virtual wind-tunnel” enabling rapid numerical analysis
and assessment of imaginative novel airframe configurations. It was even widely expected that such a
numerical simulation capability would eventually remove the need for wind tunnels entirely.

The example multi-block structured mesh shown in Figure 1-1 illustrates a typical industrial application of
(Euler) CFD for missiles dating from around 1988. For the (limited) flight conditions of interest, results
obtained at that time were considered acceptable for engineering purposes; proving to be better than semi-
empirical methods, whilst also quicker and cheaper than wind-tunnel tests.

© Copyright MBDA UK Limited 2022. All rights reserved.

STO-MP-AVT-366 3-1
Some Applications of Computational Aerodynamics to Support
Guided Weapon Design and Development

Figure 1-1: Weapons Multi-Block Suite structured mesh

(38,146 cells, Euler solution required around 1 CPU hour on Cray-1S, c.1988)

As available computing power increased, however, so too did industrial ambitions for exploring ever more
complex airframe geometries (increasingly beyond the scope of semi-empirical methods) at more challenging
flight conditions. This expansion in usage exposed the fundamental limitations of all CFD methods arising
from the inherent numerical and physical modelling approximations needed to represent a continuous world
using discrete points. So began the seemingly endless quest for mesh resolution and flow solution fidelity,
with each new generation of computational capacity encouraging ever-finer meshes and increasing flow solver
complexity. The original vision of the “virtual wind-tunnel” remains tantalisingly just beyond reach.

The realisation that simply increasing computing power will not overcome fundamental limitations of current
industrially usable CFD methods is now driving renewed research and development efforts in several
competing directions. Ironically, the continuing need for real wind tunnels is also becoming increasingly
apparent. The relatively low entry bar for producing unsubstantiated CFD predictions has increased the
perceived value and credibility of physical test results. Far from replacing the wind tunnel, it is clear that
successful CFD method development remains critically dependent upon high quality experimental data, and
close collaboration between theoreticians and experimentalists.

1.2 Use of Computational Fluid Dynamics in Design

Historically the use and importance of CFD within aerodynamic design can be organised into three main eras,
illustrated in Figure 1-2.

1.2.1 Analysis of Single Configurations, Out-of-iteration

In its early use, the time required to prepare and perform CFD analyses was too long for the results to be used
within a single design iteration. Results generated by analysis were useful in helping to providing a detailed
understanding of observed performance but did not contribute significantly to design decisions.

1.2.2 Analysis of Single Configurations, Within-iteration

In the mid 1990s dramatic reductions in the cost of computing, primarily due to the development of high-
performance computing infrastructure based upon commodity hardware and message passing communication
paradigms, significantly increased the utility of CFD computations. Results could now be obtained at

3-2 STO-MP-AVT-366
 Some Applications of Computational Aerodynamics to Support
Guided Weapon Design and Development

acceptable cost within a timeframe that allowed insights from the computations to be used within the design
iteration. Design decisions were now informed by analysis.

1.2.3 Analysis of Multiple Configurations, Within-iteration

In this final era, which began towards the middle of the first decade of this century, further reductions in the
time and cost of performing CFD simulations allowed multiple configurations to be assessed within each
design iteration. The use of CFD simulation in aerodynamic design was transformed from a supporting activity
to the main means of driving design.

Figure 1-2: Historical Usage of CFD

In recent times, the situation has become less clear. Reynolds-Averaged Navier-Stokes (RANS) based
simulations continue to be widely used to characterize multiple configurations within single design iterations.
Indeed, further reductions in the time and cost of RANS simulations is allowing them to be used earlier and
earlier in the design activity. At the same time, the advent of scale-resolving methods such as large and
detached eddy simulations has provided increased physical fidelity but at much greater cost and results
obtained in this way are used either out of iteration or sparingly within iteration.

In order to illustrate some important aspects of how MBDA-UK uses computational aerodynamics, and where
we believe developmental effort can have greatest impact, we confine our discussion to three key phases of
the launch of an internally carried air-to-air missile. The three phases are shown in Figure 1-3. The remainder
of the paper deals with these three phases in turn. In the next section, we discuss the computation of the so-
called free-air aerodynamics, the aerodynamic characterization of the weapon in free flight far from the
platform. We then consider the case of the weapon in its carriage position before finally considering the case
where the displacement effect of the aircraft on the local airflow around the weapon is significant, weapon-
platform interaction. The paper then concludes with our recommendations on how developers can best use
their efforts to support our industrial processes.

STO-MP-AVT-366 3-3
Some Applications of Computational Aerodynamics to Support
Guided Weapon Design and Development

(a) Internal Carriage (b) Interaction (c) Free-air

Figure 1-3: Key phases during the launch of an internally carried weapon.

2.0 FREE-AIR AERODYNAMICS

The use of CFD to compute the free-air aerodynamics of missile configurations occurs throughout the
development life cycle. In early stage pre-conceptual design, the need for detailed aerodynamic models is less
important and typically, only a small number of cases need to be computed. However, the number of
configurations that need to be considered in order to arrive at a balanced design is large. As the development
process continues, the number of configurations that need to be considered diminishes rapidly, but the number
of cases that must be evaluated for each configuration increases significantly. Finally, in detailed design and
during integration with a platform only a very small number of configurations need to be considered, but many
hundreds of evaluations may be required to populate the aerodynamic model. Large numbers of CFD
evaluations are therefore required at all stages of the development process and consequently the cost of a single
calculation can be a significant driver for time and cost.

Consequently, MBDA-UK typically employs simulations based upon the solution of the steady Reynolds
Averaged Navier-Stokes equations together with a turbulence closure. In common with much of the external
aerodynamics community, the Spalart-Allmaras [1] and Menter SST [2] turbulence models and their
developments are used.

Shaw et al [3] present a detailed assessment of the application of RANS turbulence models for the MBDA
open test case OTC1 [4] that illustrates the challenges faced computing the supersonic flow around missile
configurations. Although the OTC1 test case is geometrically simple, the resulting flow field is rich involving
multiple shockwaves, boundary layers, wakes, vortices and their interactions. Figure 2-1 shows contours of
total pressure ratio obtained at various locations along the axis of the OTC1 missile viewed from front to aft
at a nominal flow condition of Mach number M = 1.4, total incidence σ = 15o at sea-level conditions. The
results were obtained by solving the RANS equations with a turbulence closure based on the SA-neg
turbulence model.

The flow development over the forebody closely matches that expected for slender bodies. Despite a
significant thickening of the leeside boundary layer, there is little evidence of crossflow separation over the
nose (x/D > -2.0) and it is not until x/D = -3.0, a calibre aft of the nose body junction that crossflow separation
first becomes evident. The separated shear layers roll up to form a pair of coherent counter rotating vortices in
the leeside wake of the body. The vortex cores grow in size and the mutual interaction between the vortices
and body causes them to be pushed away from the feeding sheet. The resultant stretching produces a non-
circular core. The effects of the slight flow asymmetry produced by the roll angle are clearly visible with the
starboard side separation occurring slightly later than on the port side. This difference in separation location is
evident in the relative position of the vortex cores. There is some evidence of secondary separation below the
primary vortices.

3-4 STO-MP-AVT-366
 Some Applications of Computational Aerodynamics to Support
Guided Weapon Design and Development

x/D = -2.0 x/D = -3.0 x/D = -4.0 x/D = -6.0

x/D = -7.9 x/D = -8.1 x/D = -9.0 x/D = -10.0

x/D = -11.0 x/D = -14.0 x/D = -17.1 x/D = -19.0

x/D = -21.1 x/D = -21.9 x/D = -22.1 x/D = -23.0

Figure 2-1: Evolution of the flow around the OTC1 missile at the nominal flow condition,
contours of normalized total pressure (SA-neg model, structured mesh).

The immediate effect of the wings is to detach the forebody vortices from their feeding sheets (see for example
x/D = -8.1) resulting in an outwards movement of the vortex core. The forebody vortices continue to grow
becoming more circular in shape. At the same time, additional vortices begin to form along the tips of the
wings. The development of the windward and leeward tip vortices is very different. On the leeward side, the

STO-MP-AVT-366 3-5
Some Applications of Computational Aerodynamics to Support
Guided Weapon Design and Development

behaviour of the tip vortex is dominated by its interaction with the forebody vortex. The vortices are co-rotating
and initially well separated, the individual cores remain coherent and rotate about one another. As the vortices
age further, their cores spread by viscous diffusion, the ratio of core diameter to separation distance increases
and the vortices begin to merge. This process is relatively rapid and results in the forebody and tip vortex
merging into a single vortex, see for example x/D = -11.0. Once the merging of the forebody and wing tip
vortices on the leeward-side has completed the vortex grows rapidly and develops an elliptical shape. On the
windward side the vortex development is similar to that observed for low aspect ratio wings, and once it has
formed the vortex grows rapidly and develops an elliptical shape.

Downstream of the wings there are two features of note. Firstly, there is evidence of a strong interaction
between the leeward and windward vortices. It appears that the vortices rotate around one another, although
the structures are indistinct. In addition, the wake structure that develops behind the wings (x/D = -17.1)
interacts with the free vortices and is entrained into the vortical flow (x/D = -19.0). By the time the flow reaches
the missile fins, the wing wake structure is no longer distinct. The interaction of the remnants of the forebody
vortex with fin 4 appears strong with the fin tip appearing to pass through the core (x/D = -21.9, x/D = -22.1).

The flowfield associated with the OTC1 test case is typical of that for many missile configurations. Indeed
vortices and their interactions are fundamental to the aerodynamic characterisation of nearly all missile
airframes. Unfortunately, the prediction of vortex-dominated flows is challenging for conventional turbulence
model based approaches due to the sensitivity of the flow to the rate of decay of the vortex. The rate at which
the vortex decays is fundamental to the characterisation of its behaviour downstream, in particular its
interaction with other features, such as solid surfaces and vortices. In numerical simulations based upon the
RANS equations, the decay of the vortex occurs due to the action of:

(i) Diffusion related to the effective viscosity of the fluid.

(ii) Dissipation due to truncation error.

(iii) Dissipation due to other features of the numerical method

(iv) Diffusion due to the over prediction of turbulent eddy viscosity by the turbulence model

The action of (ii), (iii) and particularly (iv) can overwhelm the action of the physical viscosity resulting in
vortices that decay and grow more rapidly than is observed in physical experiments, for further discussion see
Bradshaw [5].

Dissipation due to truncation error and other features of the numerical model can often be addressed by careful
construction and local refinement of the computational mesh. However, dealing with over prediction of
viscosity by the turbulence model is more problematic. The fundamental problem is clearly illustrated in Figure
2-2 which shows computed contours of turbulent Reynolds number, the ratio of the turbulent eddy viscosity
to the molecular viscosity, for the OTC1 test case at two streamwise stations. We observe that in the region of
the vortices, the turbulent eddy viscosity is 3-4 orders of magnitude greater than that of the molecular viscosity.
This seems unreasonable on physical grounds and results in computed vortices that rapidly dissipate as they
travel downstream. In the case of the OTC 1 test case, this affects downstream vortex-vortex interactions that
changes the strength and geometry of the vortex field close to the fins.

3-6 STO-MP-AVT-366
 Some Applications of Computational Aerodynamics to Support
Guided Weapon Design and Development

(a) x/D = -7.9 (b) x/D = -16.9

Figure 2-2: Evolution of the flow around the OTC1 missile at the nominal flow condition,
contours of normalized total pressure (SA-neg model, structured mesh).

Several efforts have been made to remedy this behaviour including attempts to sensitize the model equations
to curvature and rotation. The use of such palliatives results in significant changes to the structure of the
vortices and their downstream interactions that in turn result in significant increases in rolling moment. This
is illustrated in Figure 2-3 which shows comparisons of rolling moment coefficient computed using several
models and several codes.

Figure 2-3 Influence of rotation correction on computed rolling moment coefficient (unstructured
mesh).

Scale resolving methods, see for example Tormalm [6] and Barakos [7], appear to offer a resolution to this
issue. Figure 2-4 provides a comparison of the convergence histories of the peak turbulent Reynolds number
for calculations employing the Spalart-Allmaras model and two scale-resolving IDDES methods. The IDDES
methods appear to behave in a much more physically meaningful way, resulting in much stronger vortices and
a significant (more than double) rolling moment coefficient.

STO-MP-AVT-366 3-7
Some Applications of Computational Aerodynamics to Support
Guided Weapon Design and Development

Fig. 2-4 Iterative convergence of maximum turbulent Reynolds number (ARL, unstructured
mesh).

The apparent improvements offered by scale resolving simulations is promising and has motivated significant
recent efforts amongst the CFD research community at the expense of more traditional turbulence modelling
approaches. However, it must be recognised that with even relatively poorly resolved detached eddy
simulations costing O(1000) more than related RANS calculations and requiring days of wall-clock time rather
than hours, the use of scale resolving simulations in aerodynamics would turn the clock back to the first or
second eras discussed in the introduction. Such approaches are therefore unlikely to displace RANS as the
main means of producing aerodynamic data in the short to medium term.

Instead, we argue that there is an industrial need for development efforts to be focussed on understanding and
extending the utility of RANS based models for vortical flows. Adopting some of the ideas expounded by
Spalart [8], we believe that this effort should be both systematic and more openly empirical.

Although progress has been made in sensitizing standard turbulence models to the effects of streamline
curvature and system rotation, the models still perform inadequately for vortex-dominated flows. A more
systematic approach based upon the solution of the Reynolds stress transport equations seems to offer the most
promising avenue for definitive improvement.

We should also realise that the search for a single, all-purpose turbulence model is likely to end in failure and
disappointment. Being more systematic in our modelling is therefore not sufficient in itself. Instead, industry
must actively embrace a more openly empirical approach in which we work together with research and
development partners to calibrate models for our specific needs. This approach is not new; industry has long
used semi-empirical aerodynamic predictions built upon simple theoretical considerations. The problem has
been that as the models have become more complicated the knowledge and understanding required to calibrate
them has typically exceeded that available within an individual company. Moreover, commercial imperatives
have prevented effective transfer of experimental data from industry to the research and development
community. If we are to overcome current limitations, a new collaboration is required. To facilitate this
collaboration on calibration there is a need for high-quality validation and calibration experiments to be
performed.

3-8 STO-MP-AVT-366
 Some Applications of Computational Aerodynamics to Support
Guided Weapon Design and Development

3.0 INTERNAL CARRIAGE

When dealing with the weapon in its carriage position we are particularly concerned with characterizing the
aero-acoustic environment within the weapon bay. Although the dynamic pressure within the bay is low when
the weapon bay doors are open, the aerodynamic environment is harsh due to the unsteady flow associated
with the resulting cavity flow. This has implications for the platform, the weapon and any adjacent stores
within the payload bay. Tonal behaviour at lower frequencies can be important for the structural integrity of
the missile airframe while higher frequency broadband noise can limit the useful service life of internal sub-
systems particularly energetic materials (for example the warhead and rocket motor) and electronics

The mechanism governing the aero-acoustic environment within a cavity is shown schematically in Figure 3-
1. As the flow passes over the leading edge, it separates, creating a shear layer between the relatively stagnant
air contained within the cavity, and the fast flowing air outside it. A feedback mechanism is created with the
rear wall as pressure perturbations are communicated back upstream, eddies form in the shear layer near the
leading edge, growing in size as they travel downstream. This leads to a highly unsteady flow regime.

Fig. 3-1 Schematic of the flow over an empty cavity.

The inherently unsteady nature of the problem and the importance of length and time scales similar to those of
the largest turbulent eddies requires the adoption of a scale-resolving strategy to the modelling of turbulence.
In the simulations discussed in this paper, a direct eddy simulation model has been employed.

Some example results for the empty cavity are shown in Figure 3-2 and Figure 3-3. Figure 3-2 shows
instantaneous contours of density gradient. The complexity of the flow and its inherent unsteadiness are clearly
visible. The schematic shown in Figure 3-3 shows the potential insights that can be obtained from such
visualisations.

STO-MP-AVT-366 3-9
Some Applications of Computational Aerodynamics to Support
Guided Weapon Design and Development

Fig. 3-2 Instantaneous density gradient contours (empty cavity).

Fig. 3-3 Sketch of inferred flow behaviour (empty cavity).

In order to characterise the aero-acoustic environment within the bay and around the missile, time averaged
contour plots of overall sound pressure were created by extracting pressure slices at each time-step and
calculating space discretised values with Equation 1.

𝑃𝑃 2
𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂 = 10 log10 � 𝑃𝑃𝑟𝑟𝑟𝑟𝑟𝑟2 � (1)
𝑟𝑟𝑟𝑟𝑟𝑟

An example result is shown in Figure 3-4 for the empty payload bay at a Mach number M = 0.85.

Figure 3-4 Computed OASPL in an empty payload bay.

3 - 10 STO-MP-AVT-366
 Some Applications of Computational Aerodynamics to Support
Guided Weapon Design and Development

(a) Computed PSD (b) Computed OASPL (cavity ceiling)

Figure 3-5 Computed Power Spectral Density (PSD) and Overall Sound Pressure Level
(OASPL) for the empty payload bay.

Figure 3-5 shows the computed power spectral density and the overall sound pressure level on the cavity
ceiling.

The results of such simulations, together with related computations of the payload bay with the missile in its
carriage position provide help to characterise not only the aerodynamics of the problem but also the response
of the missile to the unsteady loads. Working with other engineering disciplines within the business, we are
able to understand how the weapon bay environment affects the structural integrity and life of the missile.
Moreover, the insights into the hydrodynamic and acoustic phenomena offered by the simulations provide
important information that can help us to devise effective flow management strategies to mitigate or eliminate
the consequences of the noisy environment in which the missile must operate.

4.0 PLATFORM-MISSILE INTERACTION

Our final example is associated with the launch of a missile from the platform payload bay. The objective of
this activity is to establish the conditions over which a weapon can be safely and successfully launched or
jettisoned from the platform. The process is built on a 6 degrees of freedom rigid body representation of the
missile and an aerodynamic model that employs the free-air aerodynamics of the platform and missile and
their mutual interactions. The aerodynamic model can be generated using wind tunnel measurements, results
from CFD simulations or more usually using a combination of data from both sources. Aerodynamic
interference between the platform and missile are produced using a grid survey in which data are obtained for
a combination of Mach numbers, missile attitudes (total incidence and roll), control deflections and positions
relative to the aircraft.

In determining the aerodynamics using computational fluid dynamics, a quasi-unsteady approach is adopted.
Steady computations based upon solution of the Reynolds-averaged Navier-Stokes equations are performed
for each location relative to the aircraft platform. The need to consider the relative position of the missile to
the platform requires that the mesh contain both bodies and developing an appropriate meshing workflow is
key to managing the time and cost required to establish the aerodynamic database. Formerly a remeshing
approach was used, in which a single mesh was generated for each combination of missile attitude and position
relative to the platform aircraft. More recently, the overset meshing capability available within the flow solver
that we use has been exploited to simplify the meshing problem. In this approach, a mesh containing the
platform aircraft is generated independently of the mesh around the missile. The two meshes are then overlaid

STO-MP-AVT-366 3 - 11
Some Applications of Computational Aerodynamics to Support
Guided Weapon Design and Development

and an appropriate interpolation region established in the overlap between the background (platform) and
foreground (missile) meshes. This process is shown schematically in Figure 4-1. Calculations are then
performed on the two meshes and information is exchanged in the overlap region. This process has proven to
be both simple to manage and robust.

Fig. 4-1 Schematic of the Overset Meshing Strategy.

The resulting computed data form the basis of the aerodynamic model used by the six-degrees of freedom
trajectory model. An example trajectory envelope is shown in Figure 4-2 for a generic missile with geometric
and mass properties representative of those of a current generation air-to-air missile jettisoned from the open-
access fifth-generation fighter aircraft described in [9]. The nominal trajectory (red line) is indicated, together
with a large number of related trajectories that show the sensitivity of the launch to input data (for example
variability in the mass properties of the missile or the initial accelerations imparted by the ejector unit) and
model form uncertainties.

Fig. 4-2 Jettison of a generic missile from a representative platform.

More recently, there has been interest amongst the research and development community in performing
computations that couple the unsteady aerodynamics of the problem with a rigid body trajectory simulation.
Loupy and Barakos [10] used the Glasgow University solver by HMB3 and a Scale-Adaptive Simulation to
simulate a representative missile configuration being launched from a shallow rectangular cavity at a Mach
number of 0.85 and the Reynolds number based on the cavity length of 6.5 million. Loupy and Barakos’ work

3 - 12 STO-MP-AVT-366
 Some Applications of Computational Aerodynamics to Support
Guided Weapon Design and Development

shows that in principal such simulations are within the realm of possible, indeed in later work Loupy has
performed simulations which include the effect of aeroelasticity [11]. The results also illustrate the potential
costs of employing such simulations in the assessment of safe release. Even in the case where the inertia of the
missile dominates the dynamics one must still perform 10s of simulations to achieve statistically converged
trajectories.

The results shown in Figure 4-2 illustrate the importance of understanding the sensitivity of the trajectory to
uncertainty in the process used to establish a safe release. The nominal trajectory is not enough. This
observation reveals a key aspect of the industrial use of simulation in decision-making processes – the need to
model and manage risk.

As the use of engineering simulation to make technical and business decisions broadens, it is becoming
increasingly important to employ an effective engineering simulation risk model. This model accounts not
only for the accuracy of the modelling used, but also for the sensitivity of results to model input and model
form uncertainties together with the criticality and risk associated with the decision being made. This is
fundamental to using modelling and simulation to support the assessment of safe release of a weapon from a
platform where the criticality of a wrong decision may have catastrophic consequences for the platform and
its pilot. Individual heroic calculations that couple multiple physical models to perform single highly accurate
simulations are unlikely to be helpful in this situation. They are too expensive and too time consuming to be
useful for the many hundreds of evaluations that are required to properly characterize and understand the
sensitivity of the model outcome to model input and model form uncertainties.

From an industrial end user perspective, significant progress is required to improve the community’s
understanding of error and uncertainty in CFD simulations. The tools at hand lack the generality and robustness
that industry requires to be able to make effective use of simulation results. This leads to the need to take large
margins that compromise potential performance and a requirement for extensive physical trials. Heroic
simulations that couple multiple physical models, in expensive time accurate calculations will be important,
but their impact is likely to be limited to helping to understand the sensitivity of trajectories to individual
physical effects (for example flow unsteadiness or structural flexibility) and quantifying associated
uncertainties.

5.0 CONCLUSIONS
In this paper, we have presented three examples related to the carriage and release of an air-to-air missile from
a weapons bay that illustrate some of the challenges related to the use of computational fluid dynamics in an
industrial context. The examples illustrate the inherent tensions between industry’s need to make more
effective use of existing models, largely based on the solution of the Reynolds-averaged Navier-Stokes
equations, and the perceived focus of the research and development community on increasingly accurate (at
the expense of time and cost) simulations, whether of the basic fluid mechanics or by coupling multiple
physical models.

With respect to the modelling of the fundamental fluid physics we recognise that in some situations, such as
the example of unsteady cavity flows described in the paper, scale-resolving simulations employing detached
or large eddy simulations are required. Nevertheless, in most of our applications RANS models should suffice.
Turbulence modelling improvements are required for the vortical flows that are of direct interest to missile
aerodynamicists. We advocate a strategy that combines a more systematic approach to the modelling of
turbulence using the Reynolds Stress Transport equations as a basis. However, this should be combined with
a more openly empirical approach that recognises the futility of seeking a single model that can be applied in
all circumstances. This strategy should be based upon a new collaboration between industry and researchers
that would see the community work together to design and conduct appropriate calibration experiments.

STO-MP-AVT-366 3 - 13
Some Applications of Computational Aerodynamics to Support
Guided Weapon Design and Development

More generally, there is a need for the community to bring renewed focus to the development of the
understanding and infrastructure to characterise error and uncertainty in simulation results. There is a related
need to develop robust, simple processes that allow such data to be used to develop effective engineering risk
management models. It is our belief that in most cases having a more robust understanding of the error and
uncertainty associated with a large number of less accurate results is likely to be more beneficial than a handful
of calculations performed with highly accurate models.

6.0 ACKNOWLEDGEMENTS
The authors would like to thank Michael Anderson, Erdem Dikbaş, Jim DeSpirito, Christian Schnepf, and
Magnus Tormalm for use of the data contained in Figures 2-3 and 2-4 which appeared originally in Reference
[3].

3 - 14 STO-MP-AVT-366
 Some Applications of Computational Aerodynamics to Support
Guided Weapon Design and Development

7.0 REFERENCES
[1] Spalart, P.R., Allmaras, S.R.,. “A One-Equation Turbulence Model For Aerodynamic Flows,” AIAA Paper
92-0439, January 1992.

[2] Menter, F.R., “Two-Equation Eddy-Viscosity Turbulence Models for Engineering Applications”, AIAA
Journal Vol. 32 No. 8, pp. 1598-1605, August 1994.

[3] Shaw S., Anderson, M., Barakos, G., Boychev, K., Dikbaş, E., DeSpirito, J., Loupy, G., Schnepf, C. and
Tormalm, M., “The Influence of Modelling in Predictions of Vortex Interactions About a Generic Missile
Airframe: RANS”, paper to be presented at AIAA SciTech 2022, January 2022.

[4] Taylor, N. et al., “The Prediction of Vortex Interactions on a Generic Missile Configuration Using CFD:
Current Status of Activity in NATO AVT-316,” NATO STO-MP-AVT-307, Paper 24, October 2019.

[5] Bradshaw, P. "Effects of streamline curvature on turbulent flow," AGARDograph 169, 1973.

[6] Tormalm, M., et al., “The Influence of Scale Resolving Simulations in Predictions of Vortex Interaction
about a Generic Missile Airframe”, paper to be presented at AIAA SciTech 2022, January 2022.

[7] Boychev, K., Barakos, G. and Steijl, R., “Simulations of flows around complex and simplified supersonic
store geometries at high incidence angles using statistical and scale resolving turbulence models”, paper to be
presented at AIAA SciTech 2022, January 2022.

[8] Spalart, P., “Philosophies and fallacies in turbulence modeling”, Progress in Aerospace Sciences , Vol 74,
pp. 1-15, 2015.

[9] Bykerk, T., Giannelis, N.F. and Vio, G., “Static Aerodynamic Analysis of a Generic Fifth Generation
Fighter Aircraft”, AIAA Science and Technology Forum, AIAA Paper 22-1951, 2022.

[10] Loupy, G. and Barakos, G., “Modelling of Transonic Shallow Cavity Flows, and Store Release
Simulations from Weapon Bays” 35th AIAA Applied Aerodynamics Forum, AIAA 17-3252, 2017.

[11] Loupy, G. “High fidelity, multi-disciplinary analysisof flow in realistic weapon bays”, PhD Thesis,
Glasgow University, 2018.

STO-MP-AVT-366 3 - 15
Some Applications of Computational Aerodynamics to Support
Guided Weapon Design and Development

3 - 16 STO-MP-AVT-366