You are on page 1of 6

THE AERONAUTICAL JOURNAL

JUNE 2008

327

The role of drag prediction in combat aircraft


design and development
J. B. Newton
Aerodynamics Consultant
BAE Systems
Warton, UK

NOMENCLATURE

ABSTRACT

ACA
AST
CAD
CFD
EAP
ECA
ECF
ESDU
FOAS
IT
LO
ODM
RAE
STOVL
A

CD
CD0
CL
CLcrit
D
D/Q
k
K
L
M
PTj
P0
Q
Rn

This paper gives an overview of combat aircraft drag prediction in


the context of the overall design and development process.
Following a brief summary of the authors experience in this field,
the importance of drag prediction during initial configuration
design is discussed, emphasising the need for the drag aerodynamicist to develop a good understanding of the other aerodynamic
disciplines involved, as well as an appreciation of the of the total
design process encompassing structural design, propulsion
integration and systems installation.
A brief description is given of typical simplified prediction
methods used in initial design, followed by an example of drag
synthesis procedures based on wind-tunnel test and analysis, illustrating the need for good understanding of test techniques and the
requirements of other aerodynamic disciplines.
Some future challenges are identified, requiring continual
involvement in research and methods development programmes.

Agile Combat Aircraft


Air Staff Target
computer aided design
computational fluid dynamics
Experimental Aircraft Programme
European Combat Aircraft
European Common Fighter
Engineering Sciences Data Unit
Future Offensive Air System
information technology
low observable
operating data manual (for mission planning)
Royal Aircraft Establishment
short take-off vertical landing
aspect ratio span2/area
angle of incidence
drag coefficient D/(V2S)
drag coefficient at zero lift
lift coefficient L/(V2S)
lift coefficient at the drag break
drag force
drag area CDS
induced drag factor
slope of CD vs CL2 curve
lift force
Mach number
jet total pressure
ambient static pressure
dynamic pressure = V2
Reynolds number
ambient air density
true airspeed

1.0 INTRODUCTION
The purpose of this paper is to give an overview of aircraft drag
prediction in the context of the overall design and development
process. Each airframe manufacturing organisation has its own
detailed methodologies and there are some significant differences
between civil and military aircraft design practices, but the basic
objectives of producing efficient, safe air vehicles through the application of good aerodynamic design are common to both.
Although this paper is primarily based on experience of combat
aircraft design and development, many of the techniques and lessons
learnt are relevant to most air vehicle programmes.
The authors professional aerodynamic experience began at the
Handley Page company based at the Cricklewood factory, north
London, and Radlett aerodrome, Hertfordshire, during the 1960s,

Paper No. 3280. Manuscript received 23 January 2008, revised 3 April 2008, accepted 4 April 2008.
Downloaded from http:/www.cambridge.org/core. IP address: 117.241.140.212, on 06 Oct 2016 at 13:25:13, subject to the Cambridge Core terms of use, available at http:/www.cambridge.org/core/terms.
http://dx.doi.org/10.1017/S0001924000002293

THE AERONAUTICAL JOURNAL

328

Figure 1. Lightning Interceptor.

JUNE 2008

involving drag prediction and performance calculation on the Victor


BMk1/BMk2 V-bombers, the Dart Herald turboprop airliner, and the
Jetstream turboprop business/commuter aircraft. This was followed by a
move to Lancashire to join the Aerodynamics Department at British
Aircraft Corporation (which subsequently became British Aerospace
and now BAE Systems) located at Warton Aerodrome, near Preston.
Thus began a long association with combat aircraft, involving drag
prediction work for in-service aircraft and new projects, covering the
Canberra light bomber, Lightning interceptor fighter (Fig. 1), the AngloFrench Jaguar ground attack aircraft (Fig. 2), Panavia MRCA (later
named Tornado, Fig. 3) multi-role and fighter variants, P59 jet trainer
proposal for the Royal Air Force (rejected in favour of the Hawk),
fighter design studies to meet AST 396 and AST 403 (such as ECF,
ECA, P110, ACA) all leading to the EAP demonstrator and subsequently the Eurofighter, now named Typhoon by the Royal Air Force.
Drag prediction work continued as part of overall aerodynamic
planning and coordination activities covering such new projects as
early studies of supersonic STOVL Strike Fighter configurations
with McDonnell Douglas and Lockheed Martin (leading to the
current JSF F-35 programme), initial design of the Nimrod MRA4
(a derivative of the MR2 maritime patrol aircraft), feasibility
studies of Future Offensive Air Systems (FOAS) manned aircraft
proposals and finally, design, development and flight clearance of
a number of UAV demonstrators.

2.0 DRAG PREDICTION AS PART OF


OVERALL AERODYNAMIC DESIGN
ACTIVITIES

Figure 2. Jaguar ground attack aircraft.

Within any airframe manufacturing business, the discipline of


aerodynamics plays a fundamental part in meeting a customers
performance, safety and cost requirements for a particular air
vehicle. For a given engine choice, the performance and handling
characteristics of an aircraft depend primarily on the airframe
external shape (or mould lines). It follows that aerodynamic
design activities are on the critical path from the start of a project,
and after lines freeze any changes deemed necessary due to later
discovery of aerodynamic short-comings will usually incur serious
programme cost and timescale penalties.
In the initial feasibility phase of a new combat aircraft, a set of
target operational requirements is usually established with a
potential customer. One of the fascinating aspects of combat
aircraft design is the wide range of possible airframe and
propulsion system configurations which need to be considered,
and it is during this selection process that the capability to make
realistic and timely drag predictions plays an important part in
arriving at balanced design decisions. For newcomers to drag
prediction, it is very beneficial to acquire a good working
knowledge of the overall design, manufacturing and development
processes in their organisation in order to fully appreciate the
relevance and implications of their drag prediction/reduction work
for the project as a whole.

3.0 DRAG COMPONENT BREAKDOWN


INITIAL PREDICTIONS

Figure 3. Tornado combat aircraft.

All airframe design organisations have developed their own simplified


drag prediction methods for use during early design studies, usually
based on a mix of empirical techniques/data sheet methods, simple
aerodynamic theory and experimental data from past projects. At
Warton, this approach was developed over a number of aircraft
programmes and resulted in a reasonably quick assessment method for
feasibility and initial design phases, and provides a means of
accounting for and comparing the contributions to overall drag from
all the airframe components and features.

Downloaded from http:/www.cambridge.org/core. IP address: 117.241.140.212, on 06 Oct 2016 at 13:25:13, subject to the Cambridge Core terms of use, available at http:/www.cambridge.org/core/terms.
http://dx.doi.org/10.1017/S0001924000002293

NEWTON

THE ROLE OF DRAG PREDICTION IN COMBAT AIRCRAFT DESIGN AND DEVELOPMENT

Unlike aircraft designed for long-range cruise such as airliners


and long endurance surveillance air vehicles, which tend to be
optimised for efficient cruise at a design lift coefficient CLdes or
ML/D, combat aircraft have a wide operational envelope and it has
proved useful to define the total drag as the sum of profile drag at
low or zero lift (corresponding to low-level dash or 1g supersonic
acceleration), and drag due to lift, covering the full range of lift
coefficients appropriate to cruise and manoeuvring flight. For
such high-speed designs the flying surfaces usually feature thin
aerofoil sections with little or no camber to minimise profile drag
at low CL, with camber-changing devices such as leading- and
trailing-edge flaps to minimise drag during sustained turn
manoeuvres and maximise usable lift.
A useful simplification often adopted for initial drag predictions
is to linearise the CD vs CL2 curve as shown in Fig. 4, usually
involving two or more segments of increasing slope (with
dCD/dCL2 = K = k/A) as CL is increased, with the breaks defined
by values of critical lift coefficient, CLcrit. The total drag is thus
given by:
CD = CD0 + K1CL2

329

Figure 4. Simplified drag representation.

(for CL CLcrit)

= CD0 + K1CLcrit2 + K2(CL2 CLcrit2)

(for CL > CLcrit)

A typical estimated drag breakdown at zero-lift for a combat aircraft


in clean configuration (i.e. zero control deflections, no external stores)
is shown in Fig. 5, with drag being expressed in terms of drag area
(drag divided by free-stream dynamic pressure, D/Q) with units of m2
or ft2. It has been found that this approach, as opposed to the use of CD,
or drag counts (i.e. CD = 0001), gives the aerodynamicist a better
feel for the magnitude of various drag components, especially when
comparing different aircraft configurations with different wing
reference areas.
At subsonic and supersonic Mach numbers, a fully turbulent
boundary layer is assumed on all surfaces (i.e. transition at leadingedge/nose apex). Subsonic pressure or form drag is accounted for
mainly by semi-empirical form factors (functions of t/c) applied to
flying surface skin friction, an afterbody pressure drag term for the
fuselage and individual pressure drag terms for items such as cockpit
canopies, intake boundary-layer diverters and excrescences such as
auxiliary air intakes/outlets, antennae, local fairings, skin steps and
gaps etc., all based on empirical data from relevant data sheets or windtunnel tests (e.g. Ref. 1).
At supersonic speeds, the pressure drag of the fuselage, in the form
of wave drag, is calculated using quasi-cylinder/slender body theory
(Ref. 2) embodying Wards area transfer rule (Ref. 3) to account for
flying surface/fuselage interference, with additional allowances for
propulsion nozzle base and jet interference effects where appropriate,
using generalised empirical/theoretical data. This far-field method
allows a rapid assessment and refinement of the fuselage cross-section
area distribution during initial layout design to approach the theoretical
total area distribution for minimum wave drag. The linearised theory
used in this method treats the fuselage as an equivalent body of
revolution, and hence generally under-estimates the wave drag of
locally bluff areas, particularly in the region of the front fuselage,
intake and canopy, and is dealt with in this method by adding a
bluffness correction term based on comparisons of zero-lift drag
predictions with wind-tunnel and flight data for a number of
strike/fighter aircraft. The isolated flying-surface wave drag is based on
correlations of experimental data (e.g. Ref. 4). Diverters and excrescences are again accounted for empirically.
For initial predictions of lift-dependent drag, values of K1, K2, and
CLcrit and their variation with Mach number are usually based on
relevant available wind-tunnel data or data sheet methods (e.g. Ref. 5),
supported by calculations using vortex lattice and lifting surface panel
methods, with sanity checks against theoretical minimum and
maximum values, e.g. min. K = 1/A corresponding to a planar wing
with elliptic spanwise loading; max. K = 1/(dCL/d) assuming zero

Figure 5. Zero-lift drag breakdown (new project combat aircraft).

Figure 6. Lift-dependent drag factors vs Mach number.

Downloaded from http:/www.cambridge.org/core. IP address: 117.241.140.212, on 06 Oct 2016 at 13:25:13, subject to the Cambridge Core terms of use, available at http:/www.cambridge.org/core/terms.
http://dx.doi.org/10.1017/S0001924000002293

330

THE AERONAUTICAL JOURNAL

Figure 7. Tornado high-speed wind-tunnel model (1/11th scale).

JUNE 2008

leading-edge suction. Typical variations of these parameters with Mach


number are shown in Fig. 6.
The drag of externally carried stores such as weapons, fuel tanks,
targeting pods etc. has a very significant effect on aircraft performance,
and predictions are again initially based on empirical methods and windtunnel data for similar installations. For a new store, a typical approach
is to estimate the free-air drag and allow for airframe/store interference
effects by applying a suitable installation factor derived from analysis of
wind-tunnel data for similar store/aircraft configurations.
Although the overall accuracy of this simplified approach to
airframe and store drag prediction is generally no better than 10%, it
does give the aerodynamicist a feel for the magnitudes of the
component contributions, enabling quantitative inputs to decisions on
choice of configuration and the need for use of more rigorous
prediction methods and wind-tunnel testing.
It should be noted that there are a number of commercial codes now
available based on similar semi-empirical methods.

4.0 DESIGN AND DEVELOPMENT


ACTIVITIES

Figure 8. Tornado high-speed afterbody model (1/21st scale).

As a project progresses from feasibility to the main design and


development phases, it is necessary to keep track of the effects of
detail changes in external shape on drag, and to refine the overall
drag prediction by the use of more rigorous theoretical methods, e.g.
CFD, and the use of appropriate wind-tunnel model testing.
One aspect of these activities is to log all excrescence items as
they are defined, together with their estimated drag (with the aid of
Ref. 1), and to maintain this log through to the flight test stage. As
an example, on the Tornado aircraft, some 130 excrescence items
were accounted for. This not only facilitates more realistic drag
predictions, but also helps to focus design effort on the items with
most potential for drag savings.
The aerodynamicist should also become familiar with the capabilities of available CFD codes in drag calculation, both for full aircraft
configurations and their components. In some cases it may be
necessary to use full viscous Navier-Stokes codes to correctly model
the surface flows over complex geometric shapes, although these can
be very time-consuming and may still be subject to uncertainties.
Alternatively, a simpler code such as inviscid Euler is capable of
calculating transonic/supersonic wave drag and lift-dependent drag
for wing/body configurations within reasonably short timescales and
with acceptable accuracy, whilst also providing the aerodynamicist
with valuable insight into local drag-producing flow features. In
planning the efficient use of such methods it is necessary to be aware
of the needs of other aerodynamic disciplines, as it is usually
possible to utilise the same basic geometry definition and CFD code
to generate useful information in other areas such as stability and
control, aerodynamic loads and air data, as well as drag.
At an early stage in a project programme, the aerodynamicist must
define requirements for wind-tunnel testing and be able to justify the
associated timescales and costs to the project managers! It is
therefore essential to achieve the widest possible utilisation of each
model and test programme, requiring the active cooperation of all
members of the aerodynamics team to ensure all needs for windtunnel data are satisfied within the overall project timescale.

5.0 DRAG SYNTHESIS

Figure 9. Tornado high-speed afterbody model rear view (1/21st scale).

In the context of drag prediction, an approach which has been implemented on several combat aircraft programmes is that of drag
synthesis, using a number of wind-tunnel models together with
prediction methods to allow for differences in geometry and flow
conditions between model and full-scale aircraft. An essential task is
to establish a clearly-defined force and moment accounting system
to ensure correct interpretation of wind-tunnel measurements.

Downloaded from http:/www.cambridge.org/core. IP address: 117.241.140.212, on 06 Oct 2016 at 13:25:13, subject to the Cambridge Core terms of use, available at http:/www.cambridge.org/core/terms.
http://dx.doi.org/10.1017/S0001924000002293

NEWTON

THE ROLE OF DRAG PREDICTION IN COMBAT AIRCRAFT DESIGN AND DEVELOPMENT

The Tornado multi-role combat aircraft provides a good illustration of this procedure. A 1/11th. scale high-speed 6-component
force and moment model of the full aircraft configuration (Fig. 7),
used primarily to establish stability and control data, formed the
basis for drag synthesis before first flight. The model featured
accurate mould line representation apart from the afterbody/nozzle
region, which was necessarily distorted to accommodate a rear sting
mounting in the tunnel and large-area outlets for the internal free airflow from the open air intakes. This distortion resulted in elimination
of inter-nozzle gullies on the upper surface, shallower boattail angles
and unrepresentative nozzle details and base areas. Drag corrections
for internal flow through the open air intakes, ducts and outlets were
obtained by total- and static-pressure rake measurements at the
outlets.
The design of a close-spaced nozzle/afterbody with adjacent
tailplane and fin surfaces presented a significant airframe/engine
integration challenge, and to contribute to the design process an
accurate 1/21st scale high-speed afterbody model and associated
mounting rig was developed (illustrated in Figs 8 and 9). This model
not only represented the detailed external boattail and nozzle
geometry including tailplanes and fin, but also featured nozzle jet
exhaust flow at representative jet pressure ratios PTj /P0.
A forebody fairing incorporated thick wings to supply the high
pressure air to the nozzles and was earthed to the model support
system. The afterbody was live, mounted on internal strain-gauge
balances to measure thrust, drag and thrust-minus-drag. The splitline between earthed and live components was chosen to be as close
as possible to the location of maximum cross-sectional area in order
that the measured pressure force on the live afterbody corresponded
closely with the required afterbody pressure drag. Note that to obtain
afterbody pressure drag, estimates of afterbody skin friction drag
(and tailplane/fin friction plus form drag, when fitted) had to be
subtracted from the total measured afterbody drag. An equivalent
model of the distorted afterbody, featured on the 6-component
model, was also tested on this rig to obtain the required afterbody
pressure drag correction term.
Intake spill drag variation with intake mass-flow was partially
obtained from the 6-component model (using blockers in the ducts to
reduce mass flow below the fully-open condition) and also from
larger-scale subsonic and supersonic intake development models
(e.g. Fig. 10, consisting of a forebody/intake configuration mounted
on a sting and balance and powered by an ejector to obtain a full
range of mass flow conditions).
An important part of planning such a suite of models is the choice
of intake and exhaust nozzle reference conditions so that model test
results can be unambiguously corrected to flight operating conditions.
The zero-lift drag synthesis process illustrated in Fig. 11, thus
consisted of:
a)
b)
c)
d)

331

Figure 10. Tornado high-speed intake model (1/18th scale).

Figure 11. Zero-lift drag synthesis.

calculating the skin-friction drag Reynolds number correction


(wind-tunnel to flight) to apply to the 6-component model data.
predicting the difference in excrescence drag between 6component model and full-scale aircraft.
applying the measured afterbody pressure drag corrections.
applying the measured intake spill drag corrections.

To obtain the total clean-aircraft drag at typical flight conditions,


lift-dependent drag, including trim drag increments, was taken from
the high-speed 6-component wind-tunnel model data. No corrections
for Reynolds number effects were made, partly due to a lack of
reliable prediction methods and the probability that the effects would
be small, particularly at the higher wing sweeps and low
thickness/chord ratios.
For take-off and landing configurations, data from low-speed
wind-tunnel models was used to establish the lift/drag characteristics
with high-lift devices and undercarriage deployed.
The drag synthesis was completed by the addition of external

Figure 12. Tornado with a typical store load.

store drag values from the high-speed 6-component wind-tunnel


model. This provides another illustration of the need for the drag
aerodynamicist to maintain awareness of the wind-tunnel test plans
of other aerodynamic disciplines such as stability and control and
aerodynamic loads. Although they may cover most of the relevant
store configurations, their test requirements may not include suffi-

Downloaded from http:/www.cambridge.org/core. IP address: 117.241.140.212, on 06 Oct 2016 at 13:25:13, subject to the Cambridge Core terms of use, available at http:/www.cambridge.org/core/terms.
http://dx.doi.org/10.1017/S0001924000002293

332

THE AERONAUTICAL JOURNAL

cient datum configurations to allow extraction of installed store drag,


without the active and timely intervention of the drag specialist.
The end result of such a drag synthesis is a lift/drag data-set in a
form compatible with the input requirements of the particular performance program suite used by the design organisation to produce the
aircraft ODM. During planning and preparation of this data-set it is
essential to liaise closely with the stability and control/flight
mechanics specialists to ensure mutual compatibility with the overall
aircraft aerodynamics data set (often called the flight mechanics data
set) used for flight control-law development and flight simulation
studies. In particular, the flight mechanics data-set usually employs a
body-axis system (e.g. using normal and axial force components),
rather than the wind-axis system with lift and drag components used
for performance calculation. As an example, the axial force values in
the flight mechanics data-set must reflect the level of zero-lift drag
established from the synthesis process, and the values of CLmax in the
lift/drag data-set must be compatible with the maximum usable lift
established from overall aircraft handling characteristics based on
the flight mechanics data-set.
Another important role of the drag specialist during the design and
development phase is to establish a realistic build standard for
external surfaces, in terms of allowable steps, gaps, waviness and
surface finish. This requires an understanding of the manufacturing
processes involved, and close liaison with relevant production
engineering and manufacturing personnel. Any exceedences of build
standard, either by design necessity or during manufacture, will
require the raising of concessions, the clearance of which will need
consideration of all aerodynamic implications, e.g. possible effects
on aircraft handling, stall characteristics, panel loads, vibration and
buffet, as well as drag and mission performance.
On a major aircraft programme the final predicted drag level will
usually be verified or refined during dedicated flight testing. This
again requires the active involvement of the drag aerodynamicist to
ensure that test procedures and instrumentation are adequate for the
extraction of accurate drag data. Typically the Tornado programme
involved the use of instrumented engines, calibrated in an altitude
test facility at RAE Pyestock, and extensive airframe instrumentation. This allowed the airframe drag to be derived from flightmeasured engine thrust and aircraft accelerations during wind-up
turns and roller coaster manoeuvres, with a resulting accuracy of
around 5%. As in the wind-tunnel synthesis process, predictions
were required of excrescence drag differences between flight-test
and production aircraft due to instrumentation such as flight-test air
data probes and differences in equipment fit.
Finally, during the service life of a combat aircraft, there will
usually be several upgrades involving carriage of new weapons,
sensor pods and addition of external blisters/fairings (e.g. Fig. 12),
all of which require assessment of drag as well as other aerodynamic
effects. There is therefore a need for continuous monitoring of total
aircraft drag, which requires good documentation of the derivation
of the drag data set, including prediction methods, calculations,
geometric definitions and wind-tunnel data. With the inevitable
changes in aerodynamics personnel during a typical project lifecycle, controlled documentation allows some degree of continuity
and consistency to be achieved.

6.0 CHALLENGES OF FUTURE AIRCRAFT


CONFIGURATIONS
Most of the methodology for drag prediction described above in the
context of so-called legacy aircraft should still be applicable to
novel aircraft configurations, and the need for efficient, low drag

JUNE 2008

airframes will increase with future demands for long range/long


endurance air-vehicles. The advent of low-observability requirements, in particular low radar signature, has proved to be a powerful
ally to the drag aerodynamicist in achieving smooth, clean external
surfaces with minimal excrescences, whilst the associated configuration constraints pose more of a headache for the stability and
control/flight mechanics community!
For LO configurations with blended wing-bodies and highlyintegrated propulsion systems (air intakes and exhaust nozzles more
embedded in the airframe), accurate force and moment accounting
becomes even more critical, and will benefit from the application of
advanced/robust CFD codes.
Long endurance air vehicles will require minimisation of liftdependent drag through low wing loading, low span loading, and the
careful design of the wing-tip region (canted tips, winglets etc).
Also, the achievement of extensive regions of laminar flow may be
necessary, and accurate drag predictions will require reliable
methods for determination of boundary-layer transition position and
separation characteristics.
Appropriate force and moment accounting will be of paramount
importance for air-vehicles employing extensive flow control such as
surface suction, leading-edge and trailing-edge blowing, and various
degrees of thrust-vectoring, all of which will have an impact on the
accurate definition and prediction of drag.

7.0 CONCLUDING REMARKS


This paper has attempted to illustrate the role of the drag aerodynamicist in a typical combat aircraft programme, emphasising the need for
awareness of the many other design disciplines and their interactions
and dependencies. Many major programmes over the last forty years
have been characterised by international collaborations, which necessarily involve the drag aerodynamicist in assessment and comparison of
in-house methods of the partner companies. This stimulates critical
reviews of the strengths and weaknesses of different approaches, which
can promote a deeper understanding of drag-producing flow mechanisms, and lead to proposals for basic research and method development. Over the years this has also resulted in fruitful research
collaborations between industry, government establishments and
universities, with the drag aerodynamicist taking a central role. Long
may it continue!

ACKNOWLEDGEMENTS
The author wishes to thank BAE Systems for the opportunity and
privilege to take part in so many varied and interesting aircraft projects,
and for providing funding and facilities to produce this paper. Special
thanks go to my aerodynamics and wind-tunnel colleagues at Warton
for their generous assistance, IT advice and forbearance.

REFERENCES
1.
2.
3.
4.

5.

HOERNER, S.F. Fluid Dynamic Drag, 1965.


WARREN, C.H.E. and FRAENKEL, L.E. A combination of quasi-cylinder and
slender body theory, Aeronaut J, April 1955, 59.
WARD, G.N. The drag of source distributions in linearised supersonic
flow, College of Aeronautics, Cranfield Report No 88.
CALLAN, M.M. and TEMPLIN, R.J. A collection and analysis of experimental data on the aerodynamic characteristics of finite wings. Part II
Zero-lift drag of swept and straight wings at Supersonic speeds, NAE
(Canada) LR 93.
ESDU Data Items Aerodynamics Series.

Downloaded from http:/www.cambridge.org/core. IP address: 117.241.140.212, on 06 Oct 2016 at 13:25:13, subject to the Cambridge Core terms of use, available at http:/www.cambridge.org/core/terms.
http://dx.doi.org/10.1017/S0001924000002293

You might also like