1

1 Introduction
During the past few years there has been an increased
interest of the aircraft community on design loads for
aircraft. Consequently there was a workshop in 1996
SC73 on Loads and Requirements for Military Aircraft
(AGARD Report 815). Elastic effects on design loads
were presented at a Workshop: Static Aeroelastic
Effects on High Performance Aircraft.
Also an Agadogragh was written on Gust Loads:
AGARDograph 317: Manual on the Flight of Flexible
Aircraft in Turbulence. All these topics are covered in
this manual.
With the increased use of active control systems on
aircraft, there is currently a strong need to revisit some
concepts used for conventional aircraft and to identify the
correction to be brought forward to existing procedures to
compute the several loads affecting a military aircraft and
the effect of the active control system. Special attention
has been given to cover these items.
This report contains the following:
Maneuver Loads
Under this topic, design loads derivation covers the
following aspects:
Aerodynamic/inertia loads
Aeroservoelastic effects
Effects of control system failure on design envelope
Dynamic loads
Gust loads
Although not a major concern for fighter aircraft, gust
loads play an important role on aircraft that are designed
under civil requirements. A complete description of the
methods used is presented along with recommendations
on their use. The effect of control system failure is
described for the case of gust alleviation systems in
Appendix A.
Aircraft/Landing Gear Loads
The specification of a landing gear as a system is shown
in the Appendix B.
Limit Loads Concept
Limit load concepts and design loads criteria are explored
for actively controlled aircraft.
CONCLUSIONS
In this manual several approaches are presented how to
calculate design loads for existing and future aircraft.
There is a description of requirements included with
some historical background.
It very soon becomes clear that for fly by wire, agile,
inherently unstable aircraft, these requirements as far as
manoeuvres are concerned are obsolete.
Therefore, an approach as described for the Eurofighter,
where flight parameters are restricted and care free
handling of the aircraft is provided, is a possible solution.
Gust loads are also presented with some very interesting
comparisons of methods dealing with non-linear aircraft.
There is also an extensive compendium of dynamic loads
which may be designing the aircraft structure.
A more global approach is also shown which tries to
avoid insufficiencies of classical load regulations.
It is hoped that this manual can be helpful for aircraft
designers to produce realistic flight loads which will
result in optimum weight structures.
2 Loads Requirements Review
The design of modern fighter aircraft is becoming an
increasingly complex process, and the establishment of
design criteria is an extremely important element in that
process. The Structures and Materials Panel of AGARD
have noted with concern that the existing design
maneuver load regulations in the NATO nations a ) are
not uniform in content and b) do not generally reflect the
actual service experience of the aircraft.
Therefore an AGARD manual was prepared which tries
to put together the latest requirement and methods which
have been used for the design of recent modern airplanes.
As an introduction to the present situations two
contributions to military requirements are given. The first
one gives a suggestion how maneuver loads criteria could
be developed for modern agile aircraft.
In the second one the changes in the USAF Structural
Load Requirements are presented which show the
evolution of general load criteria valid for every aircraft
to a specific document which is part of the overall
specification.
Similarly a specification for undercarriage is shown in
the Appendix B. The third set of specifications is for civil
airplanes and is laid down in JAR25 (not included in this
report).
2
2.1 The development of maneuver load
criteria for agile aircraft
Max Hacklinger
Munich, FRG
AGARD Report 746, May 1987
2.1.1 Introduction
The flight maneuver loads are major design criteria for
agile aircraft (aerobatics, trainer, fighter aircraft), because
large portions of their airframe are sized by these loads.
They also belong traditionally to the most elusive
engineering criteria and so far engineers never succeeded
in precisely predicting what pilots will eventually do with
their machines. One extreme solution to this problem
would be to put so much strength into the structure that
the aerodynamic and pilot tolerance capabilities can be
fully exploited by maneuvering without failure. This is
more or less the case with aerobatics aircraft, but modern
fighters would grow far too heavy by this rule.
To keep things lucid in this overview, I shall try to
generalize or simplify the Problems but retain the
essential interrelations. Fig. 1 serves to illustrate this:
Figure 1
Box 1 contains the pilot's sensomotoric capabilities, that
is, his production of time, force and frequency dependent
inputs into the aircraft controls.
Box 2 resembles the complete flight control system
function from the sensors down to powered actuators. It
has to satisfy not only aircraft stability but also
man-machine stability criteria among others.
Box 3 stands for the airframe with its aerodynamic and
structural capabilities to produce and withstand maneuver
loads.
Box 4 contains the physiological limitations of the pilot -
his tolerance of high g, angular acceleration etc. Box 4
acts as a single limiting function on box 3 and can be
treated independently, but all other boxes are strongly
coupled with multiple feedback paths.
In the course of an aircraft development programme, box
4 is given a priori, and apart from special training effects,
box 1 is also given at the start in average form. Box 3 is
frozen relatively early by definition of the aircraft
configuration and so is the architecture of box 2. But then
for a long period of simulation and flight testing the
functions of 2 are optimized, not only for the clean
aircraft but for a variety of external stores. To a lesser
degree corrections are also possible in this period for box
3. This optimization process concerns both handling
qualities and maneuver loads, but the approaches are
different. The handling specialist has to analyze the
whole spectrum of possible flight maneuvers with main
emphasis an stability and achievement of performance.
Design load investigations are a search for maximal and
an experienced loads analyst can narrow down the vast
spectrum of possible flight cases to relatively few which
become load critical. However, this process is becoming
increasingly difficult with modern active control systems
and the control system departments have to live with a
new burden - the responsibility for causing exotic loads.
As a basis for a return to safe ground when the following
discussions of advanced maneuver systems leads us too
far astray, the next chapter gives a summary of the
present status of maneuver load regulations for agile
aircraft.
2.1.2 Status of present Criteria
The easiest way of obtaining maneuver loads is to assume
abrupt control surface movement to the stops, limited
only by pilot or actuator force, and to derive the resulting
airloads without aircraft motion analysis. This cheap
method is still in use for certification of some civil
aircraft but all the military regulations now require
sequences of pilot control inputs to initiate load critical
maneuvers. The following regulations will be
summarized here:
pilot steering
capability
flight control
system capability
airframe capability
aero & structure
pilot
tolerance
2
3 4
1
stability criteria (PIO etc.)
structural coupling, stability
limiting
function
manoeuvre
flown
s
e
n
s
o
r

d
a
t
a
f
e
e
d
b
a
c
k

v
i
a

s
e
n
s
o
r
y

c
u
e
s
3
MIL-A-008861 A (USAF) 1971 for the US Air
Force
MIL-A-8861 B (AS) 1986 for the US Navy
DEF-STAN 00-970 1983 for the UK
AIR 2004 E 1979 for France.
The US situation at the moment is curious. (A) used to be
the main US specification for flight loads over many
years. It has been replaced for the Air Force in 1985 by
MIL-A-87221 (USAF), but this new specification is only
a frame without the essential quantitative material and as
such no great help for the designer. The US Navy on the
other hand, who traditionally used to have their own and
different specification, have now adopted the old USAF
Spec. (A) and updated and amplified it for application to
modern control system technology, including direct force
control, thrust vectoring etc. Thus (B) seems to be the
most up-to-date specification available now. Although
modern fighter tactics use combined control inputs in
several axes, for a starting basis we prefer to treat them
separately as pitching, rolling and yawing maneuvers.
2.1.2.1 Pitching manoeurves
US Air Force
Fig. 2 shows the longitudinal control inputs for a checked
maneuver required in (A) to rapidly achieve high load
factors. Table 1 gives the corresponding boundary
conditions. Case (a) requires to pull maximum positive g
by a triangular control input; if the maximum is not
achievable by this, then the pilot shall pull to the stops
and hold for such time that max. g is attained. Case (b) is
similar to (a) but control displacement and holding time t
3
shall be just sufficient to achieve max. g at the end of the
checking movement. Case (c) is similar to (b) but with
control movement not only back to zero but 1/2 of the
positive amplitude into the negative direction.
Fig. 2 Stick Inputs for pitching cases of 8861A
Limit load factor
Basic design
mass
All
masses
Max design
mass
A
i
r
c
r
a
f
t

c
l
a
s
s
Max Min
at V
H
Min at
V
L
Max Min
at V
H
t
1

[
s
e
c
]
A,F,T
1)
8.0 -3.0 -1.0 4.0 -2.0 0.2
A,F,T
2)
6.5 -3.0 -1.0 4.0 -2.0 0.2
O 6.0 -3.0 -1.0 3.0 -1.0 0.3
U 4.0 -2.0 0 2.5 -1.0 0.3
1) subsonic
2) supersonic
Table 1: Symmetrical maneuver parameters of 8861 A
These theoretical maneuvers are certainly not exactly
what pilots will do with modern fighters, but as long as
we can not use the vast amount of combat simulation
results as an all embracing envelope for flight loads, they
provide at least a design basis and they have
historically produced reasonable maneuver loads,
particularly tail loads.
US Navy:
(B) has adopted these 3 cases with slightly changed
boundary conditions, see Table 2,
Limit load factor
Basic design
mass
All
masses
Max design
mass
A
i
r
c
r
a
f
t

c
l
a
s
s
Max Min
at V
H
Min at
V
L
Max Min
at V
H
t
1

[
s
e
c
]
F, A 7.5 -3.0 -1.0 5.5 -2.0 0.2
T 7.5 -3.0 -1.0 4.0 -2.0 0.2
O 6.5 -3.0 0 3.0 -1.0 0.3
U 4.0 -2.0 0 2.5 -1.0 0.3
Table 2: Symmetrical maneuver parameters of 8861 B
(d) maximum control authority in the negative direction
shall be applied until maximum stabilizer or wing load
has been attained. This can mean more than /2 in case
(c).
4
(e) is a special case for computer control, fly -by-wire,
active control, stability augmentation, the direct lift
control, or other types of control system where the pilot
control inputs do not directly its establish control surface
position" which we shall call here generically ACT
systems. This case requires that aircraft strength shall
also be sufficient to cover modifications of cases (a) to
(c) caused by ACT systems partially failed (transients,
changed gains etc.), a requirement which is easier stated
than proven.
UK
In the UK, pitching maneuvers have traditionally been
covered by airplane response calculations after the
Czaykowski method which assumed an exponential
function for elevator movement and no checking. This
was an expedient way to obtain tail loads but the new UK
specification (C) advises that pilot control inputs should
be used now. It does not specify any details of these.
France
The French specification (D) is very similar to case (a) of
(A), with two differences: it has other load factors, see
Table 3, and it allows a slower stick return to neutral in
time t
2
; for servo controls t
1
= t
2
shall be derived from
maximum control surface rate under zero load. It does
not require checking into the negative region as (A) and
(B). (see Fig. 3)
Limit load factor Aircraft
class
Max min
T
1
[sec]
T
2
[sec]
III n
1
* -0.4 n
1
0.2 0.3
II 4.0 -1.6 0.2 0.3
I 2.5 -1.0 0.3 0.3
Table3: Symmetrical maneuver parameters of AIR 2004E
* n
1
defined in the aircraft specification
Fig. 3 Control Inputs of AIR 2004 E
2.1.2.2 Rolling maneuvers (with pitching)
US Air Force
The rolling cases of (A) assume rapid control inputs and
reversal (checked maneuvers), see Fig. 4. With 267 N
force the stick shall be moved sideways in 0.1 sec, held
until the specified bank angle is attained and then
reverted to neutral in 0.1 sec. If a roll rate greater than
270/s would result, control position may be lessened to
just achieve this value, but the roll rates shall never be
lower than those necessary to achieve the time to bank
criteria in the handling qualities specification (T
360
= 2.8
sec gives P
max
150/sec).
Fast 180 rolls are required starting from level flight with
-1 to + 1g.
Fast 360 rolls are required starting from n=1.
Rolling pull out is required to start from steady level
turns with load factors from 1 to 8 n
1
( for a typical 8 g
airplane this is 1 to 6.4 g).
By application of rapid lateral control (Fig. 4) the aircraft
shall be rolled through twice the initial bank angle. In our
typical example this would be a bank angle change of
162. Longitudinal control may be used to prevent
exceeding 0.8 n
1
during maneuver.
Fig. 4 Stick Input of rolling cases of 8861 A
US Navy
The US Navy has in (B) adopted the rolling criteria of
(A) but with significant additions: for ACT aircraft the
Pilot force is replaced by "maximum control authority".
The reference to roll performance requirements is
removed - probably because this criterion used to be less
stringent than the 270 /sec in most cases. Important is
the explicit reference to external store configurations; the
rolling cases of (A) have often been met in the clean
configuration only. But most important is the addition of
a new case for ACT aircraft. It states that the aircraft
shall be designed for maximum abrupt pilot inputs in all
three axes. But it also states that these inputs shall in no
case lead to higher rates and load factors than the
conventional cases.
This paragraph is remarkable in several respects. It
describes a control system which would digest the
wildest pilots Inputs into control outputs which are
tailored to just achieve the old load maximum. It shows
clearly the dilemma of the rule maker in the face of rapid
technical development. This is the dream of the now
5
much advertised carefree (foolproof) handling system, In
reality control systems are primarily optimized for actual
maneuver performance and not for achievement of some
theoretical load cases. On the positive side this criterion
recognizes the need to retain some reference to proven
maneuver design load practice.
Another addition in (B) is the requirement that the
structure shall also be designed to withstand the
demonstration requirements of MIL-D-87088 (AS),
which apparently is not obvious.
UK
In the UK a wider envelope of initial conditions is
required for the rolling cases, including a negative g roll
reversal: -1.5 to 7.2 g. For the maximum roll rate several
limits are given: at least 1 1/3 of the roll performance
criteria in the handling specification which amounts to
about 200 /sec; 200 /sec for ground attack and 250
/sec for aerial combat maneuvers. The control input time
history is roughly as in (A).
France
The French specification also requires negative initial
conditions for the rolling cases:
-1.6 to 6.4 g. (D) has control inputs similar to (A), but
with t
1
= 0.2 and t
3
= 0.3 or maximum servo capability.
The roll limits are more severe, i.e., a full 360 roll and
p
max
- 300 /sec. (C) and (D) may reIlect the experience
that US pilots tend to avoid negative g maneuvers in
contrast to their European colleagues:
Table 4 summarizes the rolling parameters for a typical
8 g airplane.
( A ) ( B ) ( C ) ( D )
MIL-A-8861 A MIL-A-8861-B DEF STAN 970 AIR 2004 E
180 roll 1 to +1 g
360 roll at 1g
rolling pull out
from 1 to 6.4 g,
t
1
= t
2
= 0.1 sec,
p
max
= 270/sec
Same as A plus ACS fool
proof ness with maximum
control authority plus
demonstration
requirements
Rolling pull out from
1.5 to 7.2 g,
p
max
= 1.33 p handling
-200/sec
Ground attack 200/sec
Aerial combat 250/sec
No t
1
, but maximum servo
capability
360 roll, p
max
= 360/sec
rolling pull out from 1.6
to 6.4 g
t
1
= 0.2 sec
t
2
= 0.3 sec
or max servo capability
under zero load and
t
1
= t
2
Table 4: Comparison of rolling parameters (8g airplane)
2.1.2.3 Yawing Maneuvers
Fig. 5 Rudder Inputs of 8861 A
US Air Force
Apart from the usual engine failures cases, (A) specifies
low and high speed rudder reversal.
Fig. 5a shows the rudder for maneuvers from straight and
level flight. At low speed 1334 N pedal force are
required, at high speed 800 N.
Fig. 5b shows the rudder input for the reversal case; from
maximum steady sideslip a fast recovery to zero yaw
shall be made.
US Navy
(B) has adopted these design cases and amplified them
with three new ones:
for aircraft with direct side force control, strength
shall be provided for abrupt application of control
authority up to a maximum side load factor of n
y
=
3.
for aircraft with lateral thrust vectoring capability,
all maneuvers specified in the handling and stability
criteria shall also be covered in the loads analysis.
6
it is general practice that evasive maneuvers such as
jinking, missile break etc. shall be considered in the
loads analysis.
UK
(C) requires a rudder kick with 667 N pedal force or
maximum output of the control system at all speeds. It
also requires the traditional British fishtail maneuver:
starting from straight level flight, the rudder is moved
sinusoidal for 1 1/2 periods of the Dutch Roll frequency
with an amplitude corresponding to 445 N pedal force or
2/3 of the actuator maximum.
France
(D) has a rudder reversal case very similar to Fig. 5 b and
a rudder kick without reversal, but both slightly slower
than (A) due to t
1
= 0,3 sec.
Spinning is somewhat marginal for our theme of pilot
controlled maneuvers but it deserves mentioning that it
can cause rather high loads. (B) has now increased the
yawing velocity of agile aircraft with fuselage mounted
engines from the 200 /sec in (A) to 286 /sec. This is a
severe requirement for long fuselages.
The following figures show typical load maneuvers
resulting from application of the current US Mil-Specs. to
an aircraft with moderate amount of ACT (Tornado).
Fig. 6 gives time histories of response quantities in a
rapid pitching maneuver with the control input specified
in Fig. 2, case (a). displacement o
max
and holding time are
just sufficient to achieve n
z max'
Fig. 6 Tornado rapid pitch, case(a) M=0.9, 1000ft, full
CSAS
Fig. 7 is a time history of response quantities resulting
from the control input of case c in Fig. 2 which is critical
for taileron bending moment BM.
Fig. 7 Tornado rapid pitch, case (c), M = 0.92, 22500 ft,
full CSAS
Fig. 8 corresponds to the rolling pull out maneuver with
initial load factor 0,8 n
l
. This is another critical case for
taileron loads.
Fig.8 Tornado rolling pull out M=0.92, 19100ft, full
CSAS
2.1.3 The influence of piloting technique
Having set the scene of present structural maneuver
criteria, the next step is to review how realistic they are in
a changed tactical environment with different piloting
techniques. Mohrman has given a good account of these
changes in [1], describing engagement rolls, turn reversal
with push down to increase roll rate, jinking maneuvers
etc. From the fact that these maneuvers are only weakly
correlated with the specification maneuvers one might be
tempted to conclude that the old specifications should be
abandoned altogether in favor of realistic simulation of
combat maneuvers. Before deciding upon radical cut
however, several arguments need to be considered.
7
Even for the old-fashioned aircraft without ACT the
specified control inputs were never fully representative of
actual pilot handling. They came closest for a control
system with a solid stick directly connected to tail
surfaces without sophisticated tabs, but they were only
engineering simplifications of nature - like a ( 1 - cos )
gust which does exist nowhere but is used to produce
reasonable loads.
Pilots are quite inventive in finding new techniques for
combat maneuvering - in fact this is part of the selection
process (survival of the fittest). For this reason and due to
changed tactical scenarios, most aircraft later in their
service life are used differently from the way projected at
the design stage. If a sophisticated simulated combat
maneuver is used to derive critical design loads this case
may be overtaken by evolution after a few years in
service. ACT gives the possibility of late adjustments of
the limiting functions, ideally by software changes only,
but this is equally true for an aircraft designed to the old
criteria.
Perhaps the major difference between the old criteria and
the new piloting techniques lies in the longer sequences
of combined maneuvers and not so much in the short
elementary inputs (stick to the stops, maximum pilot
force).
If so, it would be easier to adapt an aircraft designed to
the old criteria to changed operational practice than one
with sizing load cases derived from specific complex
simulated maneuvers.
An important difference to the old criteria exists in the
absolute level of maneuver loads. Improved g-suits,
increased aircraft performance and improved control
systems with load limitation - all these factors have led
pilots to pull limit loads more often and for longer
duration. There is also indication for an increased
application of negative g in jinking maneuvers. This
general tendency goes so far that high performance
aircraft are now more frequently crashed due to pilot
incapacitation (GLC).
The increased overall load level certainly necessitates
adjustment of the old fatigue strength criteria (e.g.
MIL-8866); whether it also requires expansion of the
design g-envelope, is debatable. Following the rationale
which has been the basis of our airworthiness criteria for
many years now, it would be sound engineering practice
to increase design strength if the overall load level has
statistically increased. Other people argue however, that
the load limiting capability of ACT does not only justify
staying with the old design loads, but even reducing the
factor of safety.
Whilst designers are confronted with a very real increase
in the overall level of the symmetrical load cases, the
situation is more obscure with the unsymmetrical loads.
Due to various scheduled interconnects between rudder,
taileron, aileron or spoilers, the pilot now is rarely aware
of the effect his commands have on the aircraft control
surfaces. The only real limitation of unsymmetrical
maneuvers is probably the pilot's tolerance to lateral
acceleration which is far less than in the vertical
direction. Turning to Fig. 1 again, this control function is
executed via the feedback path between boxes 3 and 1.
At this point it is well to remember that the results of any
ground based simulation are severely limited by the
absence of realistic motion cues to the pilot - nevertheless
these simulations have become an indispensable
development tool.
2.1.4 The influence of advanced control
systems
The cockpit environment has drastically changed in
recent years with the rapid development of flight control
systems. For many decades pilots had to move large
controls against inertia and air forces to keep their
machines under control. Most of the aircraft in service
now have still control movement but artificial feel to
provide some indication of the flight conditions. Now
sidestick controllers are being introduced which are very
sensitive and require almost no motion. Although man is
basically a motion sensitive animal, pilots seem to have
adapted to this type of control. But from our viewpoint of
aircraft loads, we should keep in mind that many natural
limitations which used to prevent the pilot from
commanding critical flight situations, do not exist with
ACT-aircraft. The conventional type of control is
essentially a low pass filter. With sidestick controllers
many high frequency inputs, some of them unintentional,
can make the FCS nervous.
Several loading cases in the existing criteria are based on
maximum pilot forces. The attempt in (B) to replace this
for ACT-aircraft by "maximum pilot authority" is not
convincing. What is this pilot authority? The phrase
"maximum deflection of motivators" in (C) does not
resolve the problem either. This is just another case
where we have lost an engineering yardstick which used
to work well in the past.
More important than changes at the input side are
changes in the main FCS functions. Traditionally, flight
control systems have been optimized for handling
qualities, with a few loads related functions like roll rate
limitation incorporated separately. So the problem was to
provide maximum maneuverability with sufficient flight
stability to prevent loss of control. This task requires high
authority and strong control outputs. Now ACT systems
have a new basic function, load limitation, which requires
low authority and mild control outputs. Thus FCS
optimization has become a much more demanding task to
unite two conflicting targets.
The FCS-certification effort has also increased drastically
with automatic load limitation since the FCS is now a
direct component of the proof of structural integrity.
Where it was previously efficient to show that
consecutive failures in the FCS led to degraded handling
but still preserved a minimum get-you-home capability,
the load limiting function of the FCS is directly safety
critical and must therefore satisfy severe criteria for
failure rates, redundancy etc.. To a degree this is reflected
in (B) by the requirement that the loading cases shall also
include different failure states of the FCS. The associated
problems are severe and can only be touched upon:
Sensor redundancy, -disparity, software qualification,
load distribution and a. o.
8
It is clear that proof of airworthiness of ACT aircraft
would be incomplete with consideration of the
deterministic loads cases only the ACT part needs to be
treated statistically and this can be a cumbersome journey
through the woods of failure trees. Quantitative guidance
can be taken from [2]
The overall failure rates given there are still applicable to
new designs.
Let us return now to the "carefree handling" concept
which appears to offer great possibilities for loads control
and which Air Staffs are all too ready to specify because
it would reduce pilots workload significantly and free
them for tactical tasks. In our context of maneuver loads
such a control system ideally would limit all flight loads
to the design values so that neither pilot nor designer
need to worry about exceeding the structural capability of
the airframe. This requires a large number of reliable
inputs - air data, flight path coordinates, but also
continuous compete knowledge of the aircraft mass
status, including external stores partially released (speed
limits would probably still have to be observed by the
pilot).
The central problem of such a system however, is the fact
that good handling qualities and reliable load limitation
have conflicting tendencies in the FCS optimization. So
at best, a compromise can be achieved where due to the
load limiting functions the handling envelopes are
reduced, particularly in the upper left hand corner.
Load distribution is another complicating factor for an
ACT aircraft the same flight condition can often be
achieved with a variety of aircraft configurations,
depending an foreplane position, maneuver flap
scheduling and perhaps vectored thrust. Assessment of
those cases is even more difficult because airload
distribution is already a great problem on modern agile
aircraft due to non - linearities, elastic structure, fuselage
lift, dynamic lift etc.
It appears unlikely that we shall see comprehensive
carefree handling control systems in operational use
which would also effect complete load limitation. More
realistic is the selection of a few single parameters such
as symmetric g, roll rate and perhaps sideslip which are
controlled automatically. After all, who wants a formula
1 racing car with a carefree handling control system?
One of the great benefits of ACT is its flexibility. Where
previously adjustment of the handling characteristics
during development was very limited to changes of
springs, bobweights and control surface tabs, it is now
possible to tailor handling qualities over a wide range
during flight testing without large hardware changes.
Also greater changes in operational usage can be
accommodated later on by ACT. This has consequences
for the loads; they are subject to larger changes during
the aircraft life. On the other hand development of
modern aircraft takes so long that the basic configuration
must be frozen long before the final loads situation is
known with confidence.
In consequence, the certification process needs to be
changed too. It is futile from the start trying to find
structural maneuver load criteria which cover all
eventualities. What we can do is to keep our feet an
proven ground initially, that is to use the updated
conventional criteria for the basic design. Then, for a
long period of simulation and flight testing, adjustments
are made whenever weak areas are discovered. This
requires an integrated approach by the FCS and loads
departments. The certification process must recognize
this by not aiming at the usual final operational clearance,
but over many years providing preliminary clearances
which reflect the temporary state of knowledge about
tested maneuver loads and the related build standard of
the FCS.
In summary, the maneuver loads part of aircraft design
has evolved from a relatively clean-cut, predetermined
analysis to a long iterative process which gradually
utilizes flight test information to expand the flight
envelopes; a process which is also much more demanding
because it involves the reliability of the FCS in proving
structural integrity.
2.1.5 Conclusion
Design maneuver load regulations in the NATO nations
have evolved from crude assumptions of single control
surface movement to relatively complicated series of
Pilot inputs in all three axes. These inputs need to be
standardized to permit the assessment of structural loads
with reasonable effort, but with the advent of active
control technology the hiatus between standardized
control inputs for load assessment and actual pilot
practice with agile aircraft is rapidly increasing. A
solution of this dilemma may be to design flight control
systems such that they provide "carefree handling", that
is a system which even for the wildest pilot inputs does
not lead to structural damage. But this solution has also
disadvantages:
a) structural designers lose the wealth of experience
contained in previous design practice and with it their
basis for initial dimensioning of the airframe. This affects
a large portion of the aircraft mass and later re-design
may be impossible.
b) Structural safety becomes crucially dependent an the
functioning of black boxes and their connections. As long
as we have no technically feasible direct load sensing and
controlling system, a compromise is proposed: Use the
best combination of the old criteria for initial design but
allow for a long development period flight control system
adjustments of load critical functions to fully exploit the
maneuver capability of the aircraft without structural
damage. This will require a flexible system of operational
clearances where the user can not have a complete
definition of the maneuver capabilities at the start of a
program.
We have no consistent set of airworthiness criteria which
fully covers maneuver loads of agile aircraft.
Attempts to update the existing criteria to embrace the
vast possibilities of ACT are only partially successful.
Proof of airworthiness of aircraft with ACT has become
more demanding since the load influencing functions of
the FCS are directly safety critical and must be analyzed
for failure to the same quantitative criteria as the structure
itself.
The existing criteria can and should still be used for
initial design to define the airframe. Certification needs to
9
become adaptive to reflect a long period of testing and
FCS changes .
2.1.6 References:
( A ) MIL-A-008861 A (USAF) 31.03.1971
Airplane Strength and Rigidity, Flight Loads
( B ) MIL-A-8861 D (AS) 07.02.1986
Airplane Strength and Rigidity, Flight Loads
( C ) DEF STAN 00-970 October 1985
Design and Airworthiness Requirements for
Service Aircraft, Volume 1 Airplanes,
Part 2 Structural Strength and Design for Flight
( D ) AIR 2004 E Resistance des Avion 08.03.1979
[ 1 ] Mohrman, R.:
Selecting Design Cases for Future Aircraft
AGARD-Report 730, 1986
[ 2 ] Hacklinger , M.:
Airworthiness Criteria for Operational Active
Control Systems.
Paper for DGLR panel Aeroelastics and
Structural Dynamics 1979 (translation)
2.2 Changes in USAF Structural Loads
Requirements
Daniel Sheets and Robert Gerami
Loads and Dynamic Branch
Aeronautical System Division
ASD/ENFL, Wright Patterson Air force Base OH, 45433-
6503, USA
AGARD Report 746 , May 1987
The new General Specification for Aircraft Structures,
MIL-A-87221 (USAF), does not establish the traditional,
fixed requirements, but instead it presents the current
tailored approach to establishing structural loads
requirements. In most cases the previous specifications
set arbitrary load levels and conditions to be used in
aircraft design. These requirements were based upon
historical experience, without consideration of future
potential needs or capabilities brought about by
technology advances. Instead, the new philosophy
requires that loading conditions be established rationally
for each weapon system based on anticipated usage.
Also, compliance with each condition must be verified by
analysis, model test, or full scale measurement.
2.2.1 Introduction
During the late 1970s, several conditions came together
that caused the US Air Force to develop new aircraft
structural specifications. While the USAF has always had
a policy of reviewing, revising, and upgrading existing
specifications, there were factors favoring a new
approach. The contracting and legal authorities believed
that the existing system of many layers of specifications
needed to be simplified. Also, rapidly advancing
structural technologies, coupled with new realms of
performance and control capabilities, demanded that the
structural specifications address much wider range of
conditions while using an ever widening mix of
technologies. The new military specification for aircraft
structures, MIL-A-87221 (USAF), is a major deviation
from past requirement practices. It establishes weapon
system uniquely tailored structural performance and
verification requirements for airframes based on an
in-depth consideration of operational needs and
anticipated usage. In the past, specifications set arbitrary
conditions, levels, and values to be used in the design of
broad categories of aircraft.
Various sources have alleged that design requirements
have not kept pace with current usage practices;
especially in the area of flight combat maneuvers. These
allegations ignore the new requirement philosophy and
are wrong for several reasons. The specification,
MIL-A-87221 (USAF), does not preclude the
consideration of any type of loading situation. The new
specification actually requires the consideration of any
loading condition that can be identified for either
analysis, model testing, or full scale measurement.
Therefore, if a loading condition is overlooked, the fault
is not with MIL-A-87221 since it is not a set of rigid,
pre-determined requirements.
Thus, this new approach does place a greater reliance on
the designer's insight and ability to correctly anticipate
the actual service loads. The term designer represents a
broad spectrum of individuals associated with the USAF,
System Contractor, and not just from the System Project
Office which manages system development for the
USAF. Anyone attempting to use the specification must
understand that this one document covers all types of
aircraft; from light observation, to the largest transport, to
the fastest fighters, to any of the most advanced flight
vehicles. Therefore, any application of this new
specification must be tailored to the specific type of
aircraft under design. It should also be understood that no
two aircraft designs, even of the same general type, will
have identical anticipated usage. Therefore, not only must
the detail design specification be tailored to a specific
type or category of aircraft, but it must also reflect the
specific anticipated usage of the aircraft being designed
and performance capabilities brought about by
technology improvements in aerodynamics, control
system integration, materials, and human factors.
2.2.2 Structural Loading Condition
The general organization of MIL-A-87221 is shown in
figure 1. Structural loading requirements are developed
through the application of section 3.4 of the appendix.
The verification of these requirements is established by
the use of section 4.4, also of the appendix. This
procedure when incorporated into the new specification
gives the user the best features of both a checklist
approach and total design freedom. The loading
requirement section 3.4, is divided into flight and ground
conditions as shown in figure 2. The flight and ground
10
conditions are divided into subsections as shown in
figures 2a and 2b respectively. Each of the many
subsections contain various specific load sources which
the designer can either accept or modify as appropriate.
During aircraft design, particular care must be exercised
in defining both the structural loading conditions and the
associate distributions used to design the airframe, which
in turn directly influences the performance and reliability
of the aircraft. No single section of the specification can
be addressed independently. All requirements pertaining
to all technologies must be considered as one unified
entity. Both flight and ground operating conditions must
be based on the anticipated usage, unique to a specific
aircraft design effort. These conditions reflect the
operational usage from which design loads shall evolve.
Even though this new approach gives the designer
considerable flexibility, the designer is not abandoned to
establishing all requirements without guidance or
assistance. In both the requirement and verification
sections, numerous possibilities are presented for
consideration. The applicability or non-applicability of
Bach suggested requirement or verification can be
indicated by inserting either "APP" or "N/A" in a blank
provided with Bach one. For those that are considered
applicable, either the requirement or verification
procedure is then fully defined. Additionally, unique
requirements can be added as a direct product of the
tailoring process.
2.2.3 Flight Loading Conditions
The flight conditions (subsection of 3.4) consists of
thirteen categories, from the Standard symmetrical
maneuvers, to missile evasion, to the all inclusive
"Other" category which is the one that both frees the
designer from rigid requirements and simultaneously
burdens him with the need to better define anticipated
usage. The maneuver load category suggests a minimum
of five sub-categories for consideration. There is, of
course, the usual symmetric maneuver envelope, figure 3.
However, due to current usage, various maneuvers such
as extreme yaw, jinking, or missile lock evasion are
suggested for design consideration. Any maneuver which
is possible for an anticipated aircraft and its usage, must
be considered for design purposes.
Other changes can be found in the area of turbulence
analysis. Historically, gust loading conditions have been
analyzed by a discrete approach. However, the current
procedure is to employ an exceedence distribution
calculation. In order to establish the exceedence
distribution, various parameters are needed. Fortunately,
the new specification does suggest values for these terms;
figure 4 is an example from the specification. Also,
historically, maneuver and gust loading were considered
independent and non-concurrent of each other except for
aircraft engaged in low altitude missions. However,
MIL-A-87221 actually suggests the designer rationally
consider various conditions where gust and maneuver
loads are combined because they concurrently affect the
aircraft.
A very different type of load condition occurs during
in-flight refueling. While some services use the probe and
drogue system, a few others use the flying boom
approach; a few use both types of in-flight refueling
systems. This specification provides guidance in both
these areas to establish appropriate design conditions.
Since the very beginning of aircraft pressurization,
specifications have addressed its loading effects.
However, this new specification addresses pressurization
in a more inclusive manner then in the past. Usually,
pressurization concerns have been focused an cockpits or
crew compartments. In contrast, the new specification
addresses all portions of the aircraft structure subject to a
pressure differential. The requirements to consider
pressurization even apply to such areas as fuel tanks,
avionics bays, or photographic compartments. The broad
application of this section of the specification requires
constant and capable vigilance by the designer to include
all pertinent structure.
Since this specification does not presume to directly
address all possible loading phenomena, a special
category is reserved for any unique situations. This
category is called "other" and is available so the designer
can completely define all anticipated aircraft flight
loading conditions. The important aspect of this category
is that the designer is free to include any flight loading
condition derived from operational requirements that can
be appropriately defined for analysis
2.2.4 Ground Loading Conditions
While aircraft ground operations are not as glamorous as
flight performance, they can be the source of significant
loading conditions. Unlike flight conditions, there have
been very few changes to ground operating conditions in
recent years. In some cases the loading levels have been
decreased due to improved civil engineering capabilities;
improved runways, taxiways, ramps, etc. Ground loading
conditions include all ground operations (taxi, landing,
braking, etc.) and maintenance operations (towing,
jacking, hoisting, etc.).
2.2.4.1 Ground Operations
Since the earliest days of aircraft, ground operations have
changed very little. Most of these changes have been in
the area of load magnitude, not in the type or source of
load. Before takeoff, an aircraft normally needs to taxi,
turn, pivot, and brake. Various combinations of these
operations must be considered in order to fully analyze
realistic ground operations. The resultant loads are highly
dependent on the operating conditions, which are in turn
dependent on the aircraft type and anticipated mission.
2.2.4.2 Takeoff and Landing.
Usually takeoffs and landings are performed on hard
smooth surfaces which are of more than adequate length.
However, in some situations the surface is not of
adequate length, hardness, or smoothness. Therefore,
takeoff specifications must either anticipate all possible
situations or allow the designer to establish specific
takeoff and landing requirements for each system. For
example, consideration is given to rough semi-prepared
11
and unprepared surfaces. Even rocket and catapult
assisted launch is included in the specification. However,
the designer is free to consider devices such as ski-jumps,
if they are appropriate to the aircraft and missions
involved. Since takeoffs are addressed; so too are
landings. Various surfaces, arrestment devices and
deceleration procedures are included for consideration as
possible load producing conditions. The designer and
eventual user must work together to correctly establish
landing requirements, since they can vary greatly
depending on the final usage of the aircraft.
2.2.4.3 Towing
Since the beginning of aviation, it has been necessary to
tow aircraft. While the designer is free to define his own
towing conditions and associated loads, he must also to
verify the legitimacy of these conditions. In this category
the new specification comes close to the previous Air
Force criteria specifications by providing the values
given in figures 5 and 6. One should remember that these
towing conditions are very much result of years of
empirical experience. Justifying and verifying new
towing load conditions could be a very difficult task.
2.2.4.4 Crashes
Unfortunately not all flights are successful; some end in
crashes. Different types of aircraft require various types
of design considerations for crash loads, depending an
their inherent dangers due to mission and general
configuration. For example, fighters pose crash problems
with respect to seats, fuel tanks, or cockpit equipment,
but definitely not litters or bunks. However, the design of
a transport would most assuredly involve crash load
considerations for cargo, litters, bunks, or even temporary
fuel tanks in the cargo compartment. The new
specification suggests various combinations of on-board
equipment. These suggested values, figure 7, are very
similar to the historic ones which in the past were firm
requirements. Today a designer can use factors other than
the suggested ones, as long as the alternate load factors
can be substantiated.
2.2.4.5 Maintenance
Even daily maintenance actions can impose various
loading conditions on aircraft. Many maintenance
operations require towing, jacking, or hoisting which
subject the aircraft to abnormal and unusual loading
combinations that must be considered during aircraft
design. General data is supplied for these conditions, see
figure 8. However, following the tailoring in
MIL-A-87221 (USAF)., the designer is free to define any
level of maintenance induced loading which can be
substantiated.
2.2.4.6 CONCLUSIONS
The new specification, MIL-A-87221, will allow design
requirements to be more closely tailored to the
anticipated use of the aircraft. In this way the final
product will be more efficient, with less wasted,
unneeded, and unused capabilities. This will lead in turn
to reduce costs of ownership for Air Force weapon
systems. This specification has been applied to the
definition of requirements for the Advanced Tactical
Fighter. This process is now taking place.
12
13
14
15
16
17
18
3 Maneuver Loads
Design maneuver load regulations in the NATO nations
have evolved from crude assumptions of single control
surface movement to relatively complicated series of
pilot inputs in all three axes. These inputs need to be
standardized to permit the assessment of structural loads
with reasonable effort, but with the advent of active
control technology the hiatus between standardized
control inputs for load assessment and actual pilot
practice with agile aircraft is rapidly increasing.
The flight maneuver loads are major design criteria for
agile aircraft (aerobatics, trainer, fighter aircraft), because
large portions of their airframe are sized by these loads.
They also belong traditionally to the most elusive
engineering criteria and so far engineers have never
succeeded in precisely predicting what pilots will
eventually do with their machines. One extreme solution
to this problem would be to put so much strength into the
structure that the aerodynamic and pilot tolerance
capabilities can be fully exploited by maneuvering
without failure. This is more or less the case with
aerobatics aircraft. But modern fighters would grow far
too heavy by this rule.
So the history of maneuver load criteria reflects a
continuous struggle to find a reasonable compromise
between criteria which do not unduly penalize total
aircraft performance by overweight and a tolerable
number of accidents caused by structural failure.
Several approaches are presented in the next sections
which have been used for the design of the most recent
fighter airplanes.
3.1 Classical Approach
3.1.1 Definitions
Loads External Loads on the structure
Limit Load
Military Specification (MIL-Spec.):
Maximum loads which can result from authorized flight
and ground use of the aircraft including certain
maintenance and system failures
Requirement: The cumulative effects of elastic, permanent
or thermal deformations resulting from limit loads shall
not inhibit or degrade the mechanical operations of the
airplane.
Civil Requirements (FAR,JAR):
Maximum loads to be expected in service.
Requirement: Without detrimental permanent deformation
of the structure. The deformation may not interfere with
safe operation.
Ultimate Load
Military Specification:
Limit Load multiplied by a factor of safety.
Requirement: No structural failure shall occur
Civil Requirements:
Limit Load multiplied by a factor of safety.
Requirement: No failure of the structure for at least 3
seconds.
Factor of Safety
Military Specification:
The Factor of Safety shall be 1.5.
Civil Requirements:
A Factor of Safety of 1.5 must be applied to the prescribed
Limit Load, which are considered external loads on the
structure.
General Definition:
Safety Factors are used in aircraft structural design to
prevent failures when the structure is subjected to various
indeterminate uncertainties which could not be properly
accessed by the technological means, such as:
the possible occurrences, during flight or ground
operations, of load levels higher than the limit load
uncertainties in the theoretical or experimental
determinations of stresses
scatter in the properties of structural materials, and
inaccuracies in workmanship and production
deterioration of materials during the operational life
of the aircraft.
Static Loads
Airframe static loads are considered to be those loads that
change only with flight condition: i. e. airspeed, altitude,
(angle of incidence, sideslip, rotation rates, ..) etc. with a
loads / loads-parameter oscillating below 2 Hz. These
loads can be considered to be in a steady non oscillating
state (rigid body motion).
Dynamic Loads
Dynamic loads are considered to be those loads which
arise from various oscillating elastic or aeroelastic
excitation which frequencies above 2 Hz. The loads are to
be determined by dynamic loads approaches, depending
on the sources of excitation and would include:
Atmospheric turbulence / Gusts
Buffet / Buffeting / Buzz
Stores Release and Jettison
Missile Firing
Hammershock
Ground Operations
Birdstrike
etc.
Maximum Load = Maximum external Load
(general used as classical definition)
resulting from authorized flight use (Mil.
Specification)
expected in service (FAR/ JAR Requirement)
derived by the Maximum Load Concept Approach
limited by the Flight Control System, applying
Flight Parameter Envelope Approach
derived from operational flight monitoring applying
Operational Flight Parameter Approach
19
derived from load spectra (cumulative occurrences
of loads) applying Extreme Value Distribution
Maximum Load = the structure is capable to support
(used in More Global Approach)
Maximum load case which produces the maximum
value of at least 1 failure strength criterion,
integrating Load Severity Indicators.
3.1.2 Limit Load Concept
Strength requirements are specified in terms of
Limit Loads
Military Specifications:
MIL-A-8860 (ASG),
MIL-A-008860 A (USAF),
AFGS-87221 A
is the maximum load normally authorized for
operations.
Federal Aviation Regulations:
Part 23,
Part 25
is the maximum load to be expected in service.
Ultimate Loads
is limit loads multiplied by prescribed factors of safety.
The basic premise of the Limit Load Concept is to define
that load, or set of loads, which the structure should be
capable of withstanding without permanent deformation,
interference or malfunctions of devices, degradation of
performance, or other detrimental effects.
At any load up to limit loads, the deformation may not
interfere with safe operation. The structure must be able
to support ultimate loads without failure for at least 3
seconds. The limit loads, to be used in the design of the
airframe subject to a deterministic design criteria, shall be
the most critical combination of loads which can result
from authorized ground and flight use of the aircraft.
3.1.2.1 Conventional Aircraft
A limit load or limit load factor which establishes a
strength level for design of the airplane and components
is the maximum load factor normally authorized for
operations.
The determination of the limit loads is largely specified
in the regulations (MIL, FAR, Def., etc) and is
independently of the missions / maneuvers actually
performed in operation. Worst case conditions are usually
selected as a conservative approach.
Safety factors were introduced into the design of the
structure to take care of uncertainties which could not be
properly assessed by the technological means of that
time, such as:
the possible occurrence of load levels higher
than the limit load
uncertainties in the theoretical or experimental
determination of stresses
scatter in the properties of structural materials,
and inaccuracies in workmanship and
production
deterioration of the strength of materials during
the operational life of the aircraft
3.1.2.2 Actively Controlled Aircraft
For actively controlled aircraft the limit loads are to be
determined taking into account the flight control system
(fly by wire, load alleviation) for:
normal operating conditions, without system
failures
conditions due to possible system failures
The resulting loads have to be considered for design
respectively proof of the structure.
For civil aircraft required by recent regulations (FAR,
JAR):
for normal operating systems
as limit loads, ultimate loads applying the
prescribed safety factor (1.5)
for failure conditions
the safety factor is determined by the failure
probability distinctive:
active failure ( at time of failure )
passive failure ( after failure for continuation of
flight )
The purpose for the integration of an active control
system is to enhance maneuver performance while not
eroding structural reliability, safety, and service life.
The application is described in Ref. (1)
Reasons for applying other Approaches
For conventionally controlled aircraft the regulations
gives unequivocal deterministic criteria for the
determination of the most critical combination of loads.
e.g. for flight maneuvers, the regulations (Mil-A-8861)
prescribe the time history of the control surface
deflections and numerically define several essential
maneuver load parameter for the determination of
design load level.
20
Obviously with the introduction of active control
technology, as well as care free maneuvering features,
recent specifications no longer define the control surface
deflections but rather provide the cockpit displacements
of the controls in the cockpit (Mil-A-8861).
This means that existing design load regulations and
specifications based on conventional aircraft
configurations, structural design concepts and control
systems technologies, may not be adequate to give
unequivocal criteria for the determinations of design
loads and ensure the structural integrity of future aircraft
using novel control methods.
To cope with using the limit load concept for actively
controlled aircraft several approaches have been
applied:
Maximum load concept
Background and suggested models are described in
3.2.1.
An example of application:
The flight control system for a naturally unstable
aircraft is designed with the feature to feed in
maneuver parameter boundaries ( load factors, rates,
accelerations ) in such a way that limit design loads
are not exceeded.
This approach could lead to a reduction of the safety
factor for flight maneuver loads keeping the structural
safety at least as for conventional aircraft e.g. from 1.5 to
1.4 for EFA.
The application is described in Ref. (2).
Flight Parameter Envelope Approach
The loads process is described in 3.2.5
Probabilistic determination of limit load
Operational Flight Parameter Approach
The procedure is described in 3.2.2
3.1.2.3 References
[ 1 ] H.-M. Besch, H.-G. Giesseler, J. Schuller
AGARD Report 815,
Impact of Electronic Flight Control System (EFCS)
Failure Cases on Structural Design Loads
[ 2 ] Sensburg O., Bartsch O., Bergmann H.
Journal of Aircraft, Vol.24, No.11, Nov. 1987
Reduction of the Ultimate Factor by applying a
Maximum Load Concept.
3.1.3 Safety Factors Review
3.1.3.1 History
The present - day safety factor for aircraft structures, as
applied to manned aircraft, dates back 70 years. During
the last 30 years considerable progress has been made in
the fields of structural materials, semi finished products
and testing methods. Furthermore advances in
aerodynamic and aeroelasticity, combined with
developments in electronic data processing, facilitate a
more precise prediction of structural loads and structural
analysis.
A reappraisal of the safety factor would therefore seem to
be in order, not with the intention of lowering the level of
safety, but with the aim for examining the various safety
requirements in the light of present knowledge. This,
together with the fact that there exists a lack of a rational
basis for the factors of safety concept presently applied
to the design of air vehicles, brought up a discussion of
changing the structural safety concept and the factors
involved within AGARD-SMP in 1977. The Structural
and Materials Panel formed an ad hoc Group to conduct
this discussion. Three pilot papers contained in Ref.(1)
addressed the different aspects to be envisaged, and show
up inconsistencies of the present concept as well as
means and methods for permissible changes.
The result of the discussion following the presentations
before the Sub - Committee was, that it would not be
appropriate at the present time to change the concept, but
it was found worthwhile to have a collection and
evaluation of all those factors concerning structural safety
including the philosophies which back up the application
of these factors.
The Sub - Committee found it most suitable to collect all
pertinent data and back up information by means of a
questionnaire, which was drafted by two coordinators
(one for North America, one for Europe) and reviewed by
the members of the Sub - Committee.
This questionnaire was distributed to the addressed
Airworthiness Authorities of the NATO - Nations with a
request for cooperation. The replies of the questionnaire
were summarized and evaluated by the coordinators and
presented before the Sub - Committee. The answers
given, including the results of personal discussions
between coordinators and nominated representatives of
the authorities, are condensed published in Ref.(2).
From the evaluation it may be concluded that there exists
a considerable amount of agreement with respect to the
Factors of Safety and their application. On the other
hand, some disagreements and interpretations have
resulted. Thus, this report forms a basis for discussing the
disagreements in order to achieve a higher degree of
conformity between the authorities of NATO - Countries
with a regard to structural safety and reliability.
At that time the present concept and the Factors of Safety
were in general regarded as satisfactory with the intention
to review the Safety Concept till such time as more
knowledge and experience in application of new
technologies are available;
e.g.
Improvement of knowledge about flight and ground
loads occurring in service (operational loads) to
know the margin between the design conditions and
the operational conditions.
21
Introduction of new technologies, which are not
included in the scope of the existing design
requirements
active control
behavior of new materials ( composites )
3.1.3.2 REFERENCES
[1] AGARD - Report No. 661
Factors of Safety , Historical Development , State of the
Art and Future Outlook.
[2] AGARD - Report No. 667
Factors of Safety , Related to Structural Integrity .
A Review of Data from Military Airworthiness
Authorities.
3.1.3.3 Possible Methods for Splitting of Safety
Factors
In the mean time significant progress and experiences in
load determination for conventional aircraft and for
actively controlled aircraft have been made as well as
determinations of load conditions have been applied for
cases which are not covered by the several existing
airworthiness regulations; e.g. as special conditions.
Therefore it is time to take up the review of the Safety
Factor Concept. Factors of safety can be rationalized by
splitting into Loads (FSl) and structural / material
uncertainties (FSs).
The present - day safety factor covers the uncertainties as
a global factor mainly applied for
possible exeedances of loads in relation to the
design loads
uncertainties in structural analysis
without realizing the particular uncertainties of loads and
structural analysis separately i.e. the global factor is
applied as the same value for both. This application of the
same factor of safety for loads determination and for
structural analysis can lead to an apparent margin of
safety which is higher or lower than the global factor is
intended to cover.
By splitting the factor into two parts, as suggested by the
Study Group Structures of AECMA (see chapter 3.2.1.1)
for loads and for structural analysis, a clear relation of the
safety margin is determined.
FSl for loads uncertainties
FSs for structure uncertainties
The product of both factors is known, keeping the
approved total factor of 1.5 .
FS = FSl x FSs = 1.25 x 1.20 = 1.50
Another suggestion from US ( D. Gibson) is to divide the
Factor of Safety into three terms
o U1 uncertainty related to loads computation
o U2 to operational environment
o U3 to structural analysis
In this proposal U1 and U3 are the same as FSl and FSs.
U2 for predicting the actual operational environment
might be applied using deterministic criteria. The
proposed values for all terms are 1.15.
e.g. U1 x U2 x U3 = 1.15 x 1.15 x 1.15 = 1.52
For aircraft which apparently will not be able to exceed
design loads during operations e.g.
applying operational maneuver models for deriving
or updating of design loads (see chapter 3.2.4)
applying flight parameters envelope approach for
limiting specified response parameters (see chapter
3.2.5 )
The value of U2 might be 1.0 resulting in a final Factor
of Safety
FS = 1.15 x 1.15 = 1.32
3.2 Non Classical Approach
3.2.1 Maximum Load Concept
3.2.1.1 Background
The Airworthiness Committee of the international Civil
Aviation Organization (ICAO) discussed, among other
things, the subject of maximum load concept in the
period from1957 to 1970. It was decided in Montreal in
late 1970 not to pursue this concept for the time being as
a possible basis for airworthiness regulations. Several
proposals however, were made to improve structural
safety. This subject was also discussed by the Study
Group Structures of the AECMA (Association
Europienne des Constructeurs de Material Aerospatial) in
the context of the Joint Airworthiness Requirement
(JAR). These deliberations led to the suggestion to split
the proven safety factor of 1.5 into two parts, in a rational
fashion, one for uncertainties in the loading
(determination of loads), the other for uncertainties in
strength analysis including scatter of material properties
and inaccuracies in construction.
22
Allowable loads are defined as those load values that will
only be exceeded by expected loads with a prescribed
small probability. These loads are then referred to as
maximum loads.
Gust or landing loads are strongly influenced by random
physical or human characteristics. But also in these cases
safety could be much better defined by extrapolation of
loads from statistical data, rather than the application of a
safety factor of 1.5 for all cases. Furthermore, loads that
are limited naturally by the ability of the aircraft to
produce them, or by internal aircraft systems, (load
alleviation, flight control systems) could be regarded as
maximum loads to which a safety factor need not be
applied. The determination of maximum loads with a
small probability of being exceeded is entirely possible
for modern fighters which are limited in their maneuvers,
or for control configured vehicles (CCV) which are in
any case equipped with an active flight control system
(flybywire). As a principle the prescribed design
boundaries and the corresponding safety factor should not
be applied separately, i.e. the entire design philosophy
should be considered. Therefore a mixed application of
various regulations to a single project is not advisable.
Up to now the safety factor has been reduced in only a
few cases. Within the pertinent regulations only the case
of the American MIL-A-8860 (ASG) issue is known,
where no safety margin is required for the undercarriage
and its supporting structure.
It may be supposed that with the consent of the
appropriate authorities the safety factor or the load level
could be reduced in the following cases:
in emergencies, such as emergency landings into an
arresting net or cable
for transient phenomena (hammer shock pressure in
aircraft inlets)
where actuators are power-limited and large loads
cannot be produced
3.2.1.2 Suggested Models
The following models are proposed for the application of
the Maximum Load Concept.
Semi-statistical / semi deterministic
In the past operational loads were predominantly checked
by measurement of the main load parameters, in the form
of cumulative frequencies or load - parameter - spectra
(Ref. 1).
They are:
the normal load factor, in flight and on the ground
the angle of sideslip and/or the transverse load factor
the rolling velocity in flight
the bank angle during landing
On the basis of these load - parameter - spectra a
probability of occurrence of the main load parameters is
defined for each type of mission and maneuver, and the
maximum value of the main load parameter can be
determined from this.
If, for instance, an aircraft is designed for air-to-air
combat, a maximum load factor of 9.0 may be derived
from the statistical cumulative frequency distribution for
every tenth aircraft after 4000 flight hours. This value is
taken to be maximum main load parameter. For this load
parameter the loads produced by the maneuvers specified
in the pertinent regulations are determined by means of a
deterministic calculation such that the maximum value of
the main load parameter is just attained, but not
exceeded. An example is the loads as a function of time
produced by the actuation of cockpit controls according
to MIL-A-008861.
A recent approach for active controlled aircraft has been
applied to the European Fighter (EFA) for the
determination of the design loads, called Flight Parameter
Envelope Approach. ( Description see 3.2.5 )
Semi-statistical / semi empirical
It has been known for years that VG and VGH
measurements do not suffice for the definition of criteria
for structural design.
In order to obtain statistically supported design criteria, a
special NACA Sub-Committee on Aircraft Loads
recommended (1954) to expand statistical load programs
to the extend that they included measurements of time
histories of eight parameters, three linear accelerations (x,
y, z,), three angular accelerations (p, q, r,), airspeed (V)
and altitude (H).
The first measurement of this kind where made with the
F 105 D Fighter with the aim to develop a maneuver load
concept which was to predict design loads (Ref. 2). All
data were processed to calculate time histories of loads,
with peaks called observed loads. The data
oscillogramms were examined in order to define 23
recognizable types of maneuver. Assuming that for every
type of maneuver the same sequence of aircraft motion
occurs with the exception of differences in amplitude and
duration, the measured parameters were normalized with
respect to amplitude and time.
Finally, to determine the loads, the normalized
parameters were denormalized in order to get the load
peak distribution for the wing, the fuselage, and the
empennage. The good agreement between the observed
and predicted load peak distribution demonstrated the
feasibility of the maneuver model technique for the F-105
D aircraft. The F-106 Fighter was selected to demonstrate
this model, thereby determining the models usefulness
on another aircraft. The detailed results of 3770 flight test
hours made it possible to apply the maneuver model
technique i.e. the empirical calculation of component
loads as compared to F-106 design loads (Ref. 3).
The results in the form of cumulative occurrence of the
loads for wing, elevon, and vertical tail made it possible
to determine the design load for a given cumulative
occurrence.
23
A recent approach has been elaborated in the Working
Group 27 of AGARD-SMP called Operational Maneuver
Model. The demonstration of the feasibility is reported in
AGARD Advisory Report 340 Evaluation of Loads from
operational Flight Maneuvers (Ref. 4).
(Description see 3.2.2 Operational Flight Parameter
Approach)
Statistical: Extreme Value Distribution
As a rule, load spectra are produced with the objective of
determining magnitude and frequency of operational
loads. These, in turn, are used in fatigue tests to
determine the corresponding fatigue life of structure.
Loads spectra like these are derived from relatively short
time records, compared to the actual operational life time;
they do not contain those maximum values that might be
expected to occur during the entire operational life of the
structure, i.e. a knowledge of which is necessary for the
design.
Determination of Extreme Value Distribution
In cases where the range, the maximum value, and scatter
of the spectrum may be safely assumed, an extreme -
value distribution can be established, describing extreme
values of loads / load parameters by its magnitude and
related probability of exceedences (suggested by Prof. O.
Buxbaum, ( Ref. 5 )). By means of extreme load
distributions the derivation of extreme loads is feasible
for determinate probabilities of exceedences, and thereby
the design load can be determined.
Examples of applications
Maximum rolling moments on horizontal tail
derived from in - flight measurement with C160
Military Transport Aircraft, AGARD Report No.
661, page 9
Fig. 1 shows the extreme value distribution
Maximum loads on vertical tail derived from in -
flight measurements with F-106 Fighter Aircraft
AIAA - Paper No. 70-948, page 8
Fig. 2 shows the cumulative occurrences
3.2.1.3 References
[1] J. Taylor, Manual of Aircraft Loads,
AGARDograph 83 (1965)
[2] Larry E. Clay and Heber L. Short,
Statistical predicting Maneuver Loads from eight-channel
Flight Data
Report No. TL 166-68-1 (1/1968) NASA CR-100152
[3] James D. Jost and Guin S. Johnson,
Structural Design Loads for Strength Fatigue computed
with a multi-variable Load Environment Model
AIAA - Paper No. 70 - 948
[4] AGARD ADVISORY REPORT 340
Evaluation of Loads from Operational Flight Maneuvers
Final Working Group Report of Structures and Materials
Panel Working Group 27
[5] O. Buxbaum,
Verfahren zur Ermittlung von Bemessungslasten
schwingbruchgefhrdeter Bauteile aus Extremwerten von
Hufigkeitsverteilungen
LBF - Bericht Nr. FB - 75 (1967)
24
FIG. 1 EXTREME VALUE DISTRIBUTION
25
FIG. 2 CUMULATIVE OCCURRANCES OF VERTCAL STABILIZER LOADS
26
3.2.2 Operational Flight Parameter
Approach
3.2.2.1 Introduction
The determination of the design maneuver loads is
largely specified in regulations independently of the
maneuvers or missions actually performed in operation.
For conventionally controlled aircraft the regulations give
the time history of the control surface deflections and
numerically define several essential maneuver load
parameters for the determination of the design load level.
Obviously with the introduction of the fly-by-wire and/or
active control technology, as well as care free
maneuvering features, recent specifications no longer
define the control surface deflections but rather provide
the cockpit displacements of the controls in the cockpit.
This means that existing design load regulations and
specifications based on conventional aircraft
configurations, structural design concepts and control
system technologies, may not be adequate to ensure the
structural integrity of future military aircraft
configurations using novel control methods, structural
concepts and combat tactics.
In service, maneuvers, especially combat maneuvers, are
flown in accordance with practiced rules that lead to
specified motions of the aircraft in the sky. An evaluation
of operational flight maneuvers has been made for
several aircraft types flown by the USAF, CF and GAF
with the aim of deriving operational loads by applying
parameters measured in operational flights.
This approach is based on the assumption that maneuvers
trained and flown by the NATO Air Forces can be
standardized.
The standardized maneuver time history is the
replacement as a quasi unit maneuver, for all operational
maneuvers of the same type.
The Standardized Maneuver is obtained by normalization
of parameter amplitudes and maneuver time to make the
parameters independent of mass configurations, intensity
of the maneuver, flight condition, flight control system,
and of the aircraft type.
The goal is to find a standardized time history for each
type of maneuver, which is independent of the extreme
values of the relevant parameters and aircraft type.
One promising approach is to derive design loads from a
careful analysis of operational maneuvers by current
fighters to extract critical parameters and their range of
values. To investigate this approach, Working Group 27
Evaluation of Loads from Operational Flight Maneuver
was formed, AGARD involvement was particularly
relevant since it allowed the expansion of the types of
aircraft and the control systems considered in the study.
The Working Group formulated a set of activities that
addressed the fundamental premises of a method to
generate operational loads from flight parameters by
determination of Standard Maneuvers independent of the
aircraft type and the control system.
The flow chart in Figure 1 presents the general data flow
and indicates the major phases of the procedure.
These operational loads can be statistically evaluated for
use in static design and for fracture assessment.
In the first part of the procedure the verification of the
Operational Maneuver Parameter Time Histories is
described in boxes with black frames, Fig 3.2.3.
The steps of the verification are:
Recording and Evaluation of Operational Parameters
Identification of the Maneuver Types
Normalization of the Parameters
Determination of the Standard Maneuver Types
In the second part the Derivation of Operational Flight
Loads is described in boxes with red frames in 3.2.4
applying the Maneuver Model in the steps:
Selection of the Standard Maneuver Type to be
considered
Definition of the Boundary Condition as design
criteria
Calculation of the Control Deflections necessary to
perform the Operational Maneuver
Response Calculation and Verification of the
parameter time history
Determination of Structural Loads
The evaluation of this procedure done by the Working
Group (WG 27) has demonstrated the feasibility of
determining loads from operational flight maneuvers
(Ref. 1)
This Operational Flight Maneuver Approach can be used
for:
The judgment of the operational load level for
aircraft already designed with regard to the design
level (static and fatigue) as specified in the
regulations.
That means the margin between design loads and the
extreme operational loads is known.
The determination of the load level for static and
fatigue design due to operation for new aircraft to be
developed.
3.2.2.2 References
(1) AGARD ADVISORY REPORT 340
Structures and Materials Panel, Working Group 27
on Evaluation of Loads from Operational Flight
Maneuvers.
(2) AGARD REPORT 815
Loads and Requirements for Military Aircraft, Page
3 1, and Page 4 1
27
Fig. 1: Procedure Overview
3.2.3 Determination and Verification of
Operational Maneuver Parameters and
Time Histories
3.2.3.1 Verification Performed
Based on the hypothesis that all operational maneuvers
performed in service can be verified as standard
maneuvers ( normalized parameter time histories for each
independent maneuver type ) the determination of
operational loads is feasible applying the Operational
Flight Parameter Approach. The verification of this
approach to generate operational loads from flight
parameters by determination of a set Standard Maneuvers
consisting of normalized operational parameter time
histories is described.
The Standard Maneuver procedure is shown in figure 2 as
a flow chart.
For each type of Standard Maneuvers the normalized
motion parameters are to be validated independent of
aircraft type, mass configuration and flight control
system.
For the evaluation of operational parameters, the
following data were made available and have been
judged as applicable.
Flight test data by GAF Test Center for specific
operational maneuvers on three aircraft ( Alpha Jet,
F 4 F, Tornado)
Data from simulations by GAF for specific
operational maneuvers recorded on Dual Flight
Simulator for two aircraft ( F 4, JF 90 )
Service data by USAF recorded on the F-16
(selected subset from over 300 sorties from 97
aircraft )
Service data by CF recorded on the CF-18 fleet
monitoring) (selected subset of CF-18 fleet
monitoring )
Taking all data available, which have been found to be
suitable for separation into maneuver types, the data base
is about 13 maneuver types.
For two maneuver types, High - g turn and Barrel roll,
more than 60 maneuvers for each maneuver type have
been considered as applicable for evaluation.
Recorded Operational Parameters
C
B
Operational Parameters
Time Histories
Standard Maneuver Type A
C
B
Normalization
Process
Maneuver Type A
C

B
Boundary
Conditions
Maneuver Type A
Aircraft
Basic Data
M A N E U V ER M O D E L
Structural Loads
Static Design and / or Fatigue
Maneuver
Identification
Flight-Test-Data Service-Data Simulation-Data
28
The actively controlled aircraft ( Tornado, F-16, CF-18 )
fit in the same scatter band as the conventional controlled
aircraft. This means the hypothesis that the operational
maneuvers are performed in the same way, i.e.
performing the same normalized parameter time history,
can be considered as confirmed.
The result is, that the Operational Standard Maneuver
independent of the aircraft type is applicable as unit input
for calculation of the movement of a specific aircraft by
reconstitution of the real aircraft configuration and flight
condition.
3.2.3.2 OPERATIONAL PARAMETERS
The number of parameters defining the aircraft motion
should be chosen in such a way that recording and
evaluation cause minimal expense. This can be achieved
by using parameters available from existing systems of
the aircraft. Each aircraft motion must be represented by
a data set of relevant parameter time histories.
The following operational parameters are necessary:
Ma Mach-number
Alt Altitude
n(x) Longitudinal Load Factor
n(y) Lateral Load Factor
n(z) Normal Load Factor
p Roll Rate
q Pitch Rate
r Yaw Rate
t Maneuver Time
the Eulerian Angles, if available:
Bank Angle
Pitch Altitude
Heading
and additional parameters only for the verification
process:
(alpha) Angle of Attack
(beta) Angle of Sideslip
(xi) Aileron / Flaperon Deflection
(eta) Elevator Deflection
(zeta) Rudder Deflection
3.2.3.3 STANDARD MANEUVER PROCEDURE
Provided the operational parameter time histories of the
basic parameter are available in correct units, this
procedure includes several steps:
(1) Maneuver type identification
(2) Normalization of relevant parameter time histories
for a number of identified maneuvers of the same
maneuver type for comparison
(3) Determination of the mean values for each relevant
parameter time history of the same maneuver type
(4) Idealization and tuning of the parameter time
histories
(5) Determination of the standard maneuver time
histories
The result of this procedure is a data set of standardized
parameter time histories. The parameters are roll rate,
pitch rate and yaw rate of the selected maneuver type.
See Figure 2.
29
FIG 2: Standard Maneuver Procedure
30
3.2.3.4 MANEUVER IDENTIFICATION
The goal of the maneuver identification is to select the
relevant maneuver segments from the recorded
operational data base. A maneuver is identified by
comparing the observed data with the predefined
maneuver characteristics as described in the Maneuver
Type Description of selected maneuvers:
Turn
N(z) 2, p t 20/ sec, 40 90
Roll steady to bank angle, pull, the bank angle is held as
long as desired, opposite roll back to level
Roll rates of opposite sign before and after g peak.
High g Turn
N(z) > 2
Turn Maneuver
Break
N(z) > 3
High g Turn Maneuver with g peak during initial
maneuver time.
Scissors
A series of High g Turn Maneuvers
Roll Reversal
N(z) >2, p >t20/sec, 20 90
Roll steady to bank angle, directly opposite roll back to
level.
High g Rolls / Barrel Rolls
N(z) > 1.5, p > t 20/sec, (max) 360
Roll steady in one direction
Barrel Roll over top rise to a peak value . Barrel roll
underneath descend to a negative peak value.
Pull sym.
N(z) > 1.5 < 10
From 1g to 1g
The maneuver identification parameters are mainly load
factor n(z), roll rate p and bank angle .
First:
The data are checked for completeness and suitability for
separating them into missions and maneuver types.
Second:
The start and end time of each maneuver type are
identified when the roll rate is near zero and the g is
approximately 1.
The bank angle also indicates the type of maneuver, i. e.
full roll 360, half roll 180, turn < 90
Figure3 :Identified Time Histories of Correlated Operational Parameters
31
FIG 4: Unified Roll Directions
FIG 5:Normalizsation of Parameters
32
Figure 3 shows as an example for the identification of a
High g Turn Maneuver. In this case the roll rate trace
primarily defines the maneuver length.
The pilot first rolls the aircraft in the direction of the turn
and finally rolls it back to the wings level position. In
parallel, the g rises to a peak value. The peak is held as
long as desired. The g drops down from its peak as the
aircraft is rolled back to the wings level.
The start and the end of the maneuver are determined as
follows: the maneuver starts when the first negative /
positive deflection of the roll rate trace starts and the
maneuver finishes after recovering i.e. the opposite
deflection of the trace, decreased to zero.
The Eulerian angles , , ,give the aircraft orientation
with respect to the earths coordinate system.
The bank angle values indicate the type of maneuver as
defined in Maneuver Type Description.
All recorded parameters are time related.
3.2.3.5 NORMALIZATION
Normalization is necessary because several maneuvers of
the same type are different in roll direction , amplitude of
motion and in maneuver time. For the calculation of
loads from operational maneuvers it not important to
separate the maneuver types into different roll directions.
Therefore, maneuvers of the same type are transformed
into a unified roll direction. See Figure 4.
For a requisite comparison, a two dimensional
normalization is necessary.
Figure 5 illustrates the basic procedure of normalization.
The ordinate presents one of the parameters of motion :
y= n(y), n(z), p, ........for several maneuvers of the same
type : y(1), y(2), ........y(n).
These parameters are normalized by relating them to the
maximum values (absolute derivation from zero) which
have occurred. This means the maximum value of each
normalized parameter becomes in this case:
Y= y(1)max = y(2)max = + 1.0
The time is presented by the abscissa t , where by the
maneuver executing time is marked by t(1), t(2), .......t(n)
for several maneuvers.
The normalization is accomplished in that way that:
firstly, the maneuver time is chosen as the value 1.0 i. e.
t(1)= t(2) = T = 1.0
secondly, the extreme values of the relevant parameters
is chosen at the same normalized time.
The time scale normalization factor for all correlated
parameters: n(y),n(z),p, q, r, , , , within, fore
example, a High g Turn was derived from the roll rate
trace. See Figure 6
FIG 6: Correlated Parameters
33
FIG 7: Normalized Roll Rate Trace
FIG 8: Time Ratio
34
Fig.9 Shifted Roll Rate Traces
Fig. 10 Comparison of Normalized Rate Traces
35
In the normalized time scale, T=0 corresponds to the time
when the roll rate trace first goes negative or positive
(start of the maneuver ), and T=1 corresponds to the time
when the roll rate trace is back to zero after the opposite
roll rate peak (finish of the maneuver). Figure 7 shows
the normalized roll rate trace (positive roll direction).
This normalization procedure is dependent on the
accurate maneuver start value. (p0)
In several cases the start values of the available time
slices are very poor. One reason is the low sample rate of
e.g. 1 or 2/sec. Recordings from flight tests are sampled
24 times per second.
An other reason is the selected parameter threshold
values of the data reduction and maneuver identification
process, combined with a low sample rate.
For these cases an upgraded normalization procedure,
derived from the basic procedure, is used.
The estimated time of a High g Turn t(m) had a very
high correlation with the difference between the time of
the first and the second roll rate peak. See Figure 8. This
time ratio is very important for the normalization
procedure
The time transformation from real time into normalized
time requires several steps:
1. Determination of time ratio. The time ratio is defined
by t`(1)= dt/t(m)
2. Harmonization For the comparison of the parameter
traces, a harmonization of the maneuver time ratio is
necessary.
sf
n
= scale factor
3. Shifting A new interpolation of a similar number of
time steps for each of the correlated parameters for all
maneuver of the same type is necessary Then the roll rate
traces were shifted in a way, that all selected first peaks
coincided at the same time step.
All correlated parameters are shifted parallel in the
similar way.
Figure 9 presents the comparison of the shifted roll rate
traces versus normalized time for the selected High g
Turn maneuvers.
The amplitudes of the traced are normalized individually.
Each value of the trace is divided by its absolute
deviation value from zero, therefore, all normalized
amplitudes will fall between t1.0.
Figure 10 shows the result of the peak to peak
normalization procedure.
The application of the two-dimensional normalization
procedure is very helpful for the comparison of maneuver
time histories. In this normalized form, all parameter time
histories are independent of the aircraft type.
3.2.3.6 MEAN VALUES
After normalization of the maneuver time, for all selected
maneuvers of the same type, the typical values of the
relevant parameters in this case the peaks of the roll
rate coincide at the same normalized time. Each
parameter time history contains the similar number of
time steps, independent of is individual maneuver length.
This is the basis for calculating the arithmetic mean
values for each of the time steps.
Figure 9 presents the comparison of the non- normalized
roll rate traces versus normalized time for the selected
High g Turn maneuvers. The roll rate is a good example
for all relevant parameters.
Note: The amplitudes for the mean value calculation are
not normalized.
The mean value is defined by:
y j
y j
n
m
i
i
n
( )
( )

1
n = number of maneuver of the same
type
j = time step
y
i
(j) = relevant parameter
y
m
(j) = mean value
The mean values of all parameters have been formed in
combination by smoothing of the time history.
For plot comparison, a normalization of the amplitude is
necessary.
3.2.3.7 IDEALIZATION
The mean value traces represent a good estimation of the
relationship between the selected parameters during a
maneuver (e. g . High g Turn ).
For the compensation of any minor errors by the mean
value calculation and for reasons of compatibility, the
mean values have to be idealized and tuned.
The interpretation of idealized and tuned as follows:
To cover the most extreme peaks of the control surface
deflections possible, the most extreme accelerations in
roll (p), pitch (q), and yaw (r ) are used.
These values are obtained by linearization of the
acceleration time history in a way that the same response
of the aircraft is obtained.
For the idealization, the calculation is performed in three
steps.
n n sf t sf t sf t sf t ..... 3 3 2 2 1 1
36
y
y
x

In the first step, the following parameters were

calculated:
The three angular accelerations p, q and r by
differentiating the three angular rates p (roll), q (pitch)
and r (yaw) with respect to maneuver time.
The differentiation was given by
In the second step, the acceleration traces p, q, r, were
replaced by linearized traces
With respect to the zeros of the traces and extreme
values of p, q, r and the
corresponding extreme values of roll -, pitch- and yaw
rate.
Figure 11 presents the comparison of derived roll
acceleration trace and idealized trace versus maneuver
time for a High g Turn Maneuver.
In the third step, the three angular rates (roll, pitch,
yaw) were recalculated
By integrating the idealized values of the three angular
accelerations (p, q, r).
FIG 11 : Idealization Traces
FIG 12: Standard Maneuver
For the reasons of compatibility, the idealized data have
to be tuned, that means the relation between the three
Eulerian angles , and the angular rates p, q, r is
verified with the equations:
The result is the standardized maneuver.
Figure 12 presents the idealized and tuned standardized
traces of the three angular rates for a High g Turn
maneuver.
For each type of standardized maneuver the normalized
motion parameters are independent of aircraft type, mass
configuration and flight control system.
3.2.4 Flight Loads derived from Operational
Maneuvers
The determination of operational loads is considered as
feasible applying an Operational Maneuver Model. The
essential input for the Operational Maneuver Model is a
set of Operational Standard Maneuvers consisting of
normalized operational parameter time histories, as
determined in 3.2.3.
The operational loads can be determined by introducing
aircraft basic data, flight condition and boundary
conditions for the maneuver to be considered.
3.2.4.1 Application of the Operational Maneuver
Model
The application of the Operational Maneuver Model is
feasible for the determination of loads in general.
for Extreme Operational Loads / Limit Loads taking
into account the boundary conditions for design
for Fatigue Loads by building a usage spectrum
made up of reconstituted Operational Standard
Maneuvers
for Loads related to recorded parameters taking into
account the recorded parameters directly without
application of standardization procedure
(normalization, mean values, tuning, idealization)
and without tailoring by boundary conditions
Aircraft Basic Data
Aircraft basic data is the main input for the Operational
Maneuver Model and is required to perform the
reconstitution from the standardized maneuvers.
q +

cos

sin cos
p

sin
r +

sin

cos cos
37
For calculation of the control deflections necessary to
generate the operational parameter time history, the
following data are needed:
Aircraft configuration
geometric data
operational mass
inertia properties
Aerodynamic data set for the aircraft Cl, Cm= f(),
Cy, Cl, Cn = f(,)
Flight Control System data
for conventionally controlled aircraft: mechanical
gearing / limits
for active controlled aircraft: flight control law
(EFCS)
Engine data- thrust
Flight Condition- airspeed, Ma altitude
3.2.4.2 For calculation of structural loads on
aircraft components the following data are
needed:
- aerodynamic data set for the components to be
considered (wing. tailplane)
- mass data for the components to be considered
3.2.4.3 Boundary Conditions as Design Criteria
Boundary Conditions have to be considered as the main
input for defining the load level.
This is necessary for the determination of the extreme
operational maneuvers and consequently for the
verification of design loads.
The boundary parameters to be defined for an operational
maneuver are:
Design Maneuvers
o the shortest maneuver time (Tman = minimum)
o realizable by the control system and the
aerodynamic limits
o the maximum vertical load factor ( nz )
o the maximum lateral load factor ( ny )
o the maximum bank angle () for the maneuver
to be considered
These boundary condition parameters can be derived
from spectra of main load parameters by applying
extreme value distributions, an example is shown in
Figure 13.
If no spectra are available the main load parameters
stated in the Design Requirements ( MIL Spec. ) can be
applied.
Fatigue Maneuvers
All the main load parameters can be taken from related
spectra available.
The procedure of Operational Maneuver Model is
shown in Figure 14 as a flow chart.
Using the Standardized Operational Parameters the
reconstitution into real time is performed.
For these operational parameters time histories in real
time the control deflections necessary to generate the
operational maneuver can be determined as follows:
roll control by applying roll and yaw equations
pitch control using the pitch equation, taking into
account the symmetrical aileron deflections
yaw control by applying sideslip and yaw equations
Using these control deflections the response calculation is
done for real time conditions, but for the purpose of
checking the results with respect to the standardized
maneuvers, the response parameters are normalized.
In a comparison of the parameters between input and
output, the standardization is checked. In case of
confirmation of the conformity of the main response
parameters with the standardized parameters, the output
parameters are considered to be verified. These verified
data represent the model parameters for load calculation.
The calculation of the Operational Loads is performed in
the conventional manner applying the verified model
parameters in particular the control deflections
determined for the Operational Maneuver to be
considered.
38
FIG 13 : Boundary Conditions for Design Maneuvers
39
FIG 14 : Procedure of the Operational Maneuver Model
40
3.2.5 Flight Parameter Envelopes Approach
Abstract
This part of the manual will explain in detail the Flight
Parameter Envelope Approach:
A new method to determine the critical flight design loads
for a modern control configured fighter aircraft. The way
from the initial design phase up to the Final Operational
Clearance (FOC) will be examined.
The Flight Parameter Envelope Approach has to be seen
in conjunction with the new design tools (i.e. Loads
Model) and the modern digital Flight Control Systems
with carefree handling and load limiting procedures. The
definition of Flight Parameter Envelopes will then be
useful and feasible if computer tools are available to do
extensive load investigations for the total aircraft under
balanced aircraft conditions and if the FCS will limit the
aircraft responses (carefree handling) and with it the
aircraft loads (load limiting system).
The definition of Flight Parameter Envelopes may be a
problem for new aeroplanes where in the beginning of
the aircraft development only limited information about
the aircraft responses from previous or similar aircraft is
available. New techniques, such as thrust vectoring for
high angle of attack maneuvering in combination with
higher dynamic pressures may cause new problems. But
the poststall flight conditions up to now known are only
loads critical locally because the dynamic pressures in
the flown poststall regime is low.
However for aircraft like the Eurofighter generation the
definition of Flight Parameter Envelopes is a useful and
feasible approach to determine the critical flight design
loads and to overcome the additional problem that
Military Specifications became more and more obsolete
for aircraft design.
List of Symbols
A/C Aircraft
ALE Allowable Loads Envelope
CFC Carbon Fibre Composites
DOF Degree of Freedom
FCS Flight Control System
FOC Final Operational Clearance
HISSS Aerodynamic Program - Higher
Order Panel Sub- and Supersonic
Singularity Method
IFTC Initial Flight Training Clearance
MAST Major Airframe Static Test
MAFT Major Airframe Fatigue Test
MLA Maneuver Load Alleviation
RF Structural Reserve Factor
f
limit
Limit Load Factor
f
ult.
Ultimate Load Factor
F
x
, F
y
, F
z
Forces
M
x
, M
y
, M
z
Moments
c. g. center of gravity
q
dyn
dynamic pressure
n
x
, n
y
, n
z
load factors
p roll velocity
q pitch velocity
r yaw velocity
p
dot
roll acceleration
q
dot
pitch acceleration
r
dot
yaw acceleration
angle of attack
sideslip angle
q
dyn
product of sideslip angle and
dynamic pressure

F/P
foreplane deflection angle

T/E
trailing edge deflection angle

R
rudder deflection angle
3.2.5.1 Introduction
When starting with feasibility studies for a new fighter
aircraft in the beginning of the eighties indications from
an aircraft designed in the early seventies were confirmed
that a change of the applications of Military
Specifications for the aircraft design would be necessary.
This was also valid for the evaluation of aircraft design
loads (e.g. MIL-A-08861A).
The increase in new technologies e.g.
increase of computer capacity
digital flight control systems (FCS)
new materials e.g. Carbon Fibre Composites (CFC)
better and more efficient design tools e.g. Structural
Optimization Tool, Loads Model, etc.
led to a change of the design and performance
requirements for a new fighter generation.
The high work load of the pilots should be reduced in
contrast to the increase of the tasks of the aircraft such as
performance, agility, etc.. The consequence was to design
an aerodynamic unstable aircraft - increase of agility
with a digital Flight Control System (FCS)
The requirement to reduce the workload of the pilot could
be fulfilled by a carefree handling and automatic load
limiting procedure in the FCS control laws. With it the
control function of the pilot for the instrument panel in
the cockpit is reduced to a minimum and eyes out of the
cockpit whilst maneuvering is possible.
To overcome the new situation for calculation of critical
design loads for modern fighter aircraft the so called
Flight Parameter Envelope Approach was developed and
will be described here for an aerodynamically unstable
aircraft with foreplanes (see Fig. 1) featuring:
41
artificial longitudinal stability
extensive control augmentation throughout the flight
envelope
carefree maneuver capability with automatic load
protection achieved by careful control of maneuver
response parameters
Fig. 1 - Demonstrator Aircraft for Flight Parameter
Envelope Approach
The main problem is to realize an agile and carefree load
limiting FCS. Therefore a robust structural design of the
airframe is necessary including an appropriate growth
potential for possible changes of the FCS control laws
covering aircraft role changes which may influence the
design loads and with it the aircraft structure. To make
sure that the airframe and the FCS are harmonized:
aircraft structure and FCS control laws have to be
developed concurrently.
In comparison to earlier aircraft like Tornado the design
loads for the new FCS controlled fighter aircraft have to
be defined without a detailed knowledge of the final
standard of the FCS because
a very limited understanding of the FCS- control laws
is available in the initial design phase.
This problem can be solved by the definition of new
Structural Design Criteria where among other design
conditions the principal flight maneuver requirements for
the aircraft have to be defined. In this case the FCS
dependent loads critical Flight Parameter Envelopes (s.
Fig. 2) are defined by:
translatory accelerations (n
y
, n
z
)
rotational velocities (p, r)
rotational accelerations (p
dot
, q
dot
, r
dot
)
sideslip conditions (q
dyn
)
etc.
To take into consideration all requirements of the
different aircraft design disciplines the Flight Parameter
Envelopes have to be defined in not only considering
FCS but also
Flightmechanics
Aerodynamics
Structural Dynamics
Loads
Fig. 2 Loads Critical Flight Parameter Envelopes for the Loads Model Interdependence between the Flight Parameter
Envelopes and Critical Design Load Cases for Main A/C- Components
42
The calculation of aircraft design loads will be done with
a modern computer tool the so called Loads Model and
the Flight Parameter Envelopes are a part of this tool.
3.2.5.2 The Flight Parameter Envelope Approach
and the Loads Model
Both the FCS dependent Flight Parameter Envelopes
(Fig. 2) and the Loads Model (Fig. 3) result in a highly
efficient computer tool for aircraft design load
calculations:
- the maneuver requirements of the aircraft controlled
by the FCS are indirectly defined by the Flight
Parameter Envelopes and the Loads Model contains
all the important aircraft mass and aerodynamic
informations which have to be known to calculate
the critical design loads for the aircraft
3.2.5.3 Description of the Loads Model
The todays computer capacities allow extensive load
investigations considering:
- all mass informations (masses, c.g.s, moments-of-
inertia, mass distributions) for the total aircraft and
specific aircraft components
- the corresponding aerodynamic information
(aerodynamic pressures, aerodynamic coefficients/
derivatives) for the total aircraft and the defined
aircraft components for different Mach numbers
- the static aeroelastic input (flexibility factors and
increments for total aircraft and aircraft
components) to correct the rigid aerodynamics
(aerodynamic pressures, aerodynamic coefficients/
derivatives) for defined Mach numbers.
The mass- and aerodynamic data have to be stated for
different loads critical aircraft configurations.
The idea of the Loads Model is to calculate the critical
aircraft component design loads (aircraft component
loads envelopes) to get balanced load cases for the total
aircraft. That means the total sum of the aircraft
component forces and moments is zero (equilibrium) for
each load case:
F
x,y,z
= 0 M
x,y,z
= 0
These balanced load cases (Fig. 4) are the basis for the
calculation of nodal point loads for the total aircraft
Finite Element Model (FE-Model) and for the stress
analysis.
Simplified the Loads Model is a combination of big input
and output data files and a number of computer programs
(Fig. 3). The input data sets contain all information which
is necessary for load calculations while the output data
sets contain the results of the load calculations as load
case conditions, forces, moments, aircraft component
load envelopes, etc..
The computer programs of the Loads Model can be
classified into two different groups
- programs to establish and to handle the required data
sets
- programs to compute the critical aircraft component
loads (balanced load cases, loads envelopes)
Fig. 3 Loads Model - Overall View
43
Fig. 4 Total Aircraft Balanced Load Case
To use the Loads Model efficiently the structural design
rules including the flight maneuver requirements have to
be defined for the new aircraft. This will be done in the
SDC.
3.2.5.4 Structural Design Criteria (SDC)
Because more and more the Military Specifications (e.g.
MIL-A-08861A) are getting obsolete for the design of
modern fighter aircraft it becomes important to define the
new structural design rules in the Structural Design
Criteria.
The following conditions have to be defined in the
Structural Design Criteria:
Design Flight Envelope- Mach/altitude
n
z-max./min.
vs. Mach
f
limit
, f
ult.
- limit/ultimate load factor
Loads critical aircraft configurations with and
without stores key configurations
Aircraft design masses:
Basic Flight Design Mass, Maximum Design Mass,
Minimum Flying Mass, Landing Design Mass, etc.
Gust conditions:
gust design speeds in combination with aircraft
speeds, gust lengths
Temperatures:
maximum recovery temperature
maximum stagnation temperature
Ground Loads Criteria:
sink rate, crosswind, arresting, repaired runway, etc.
Departure and Spin
Hammershock conditions
Bird strike conditions
Static aeroelastic requirements
Flutter/divergence requirements
Fatigue conditions:
safe life or fail save philosophy
g-spectrum, scatter factor, aircraft service life, etc.
etc.
Additional to the above described design conditions also
the principal flight manoeuvre requirements for the
aircraft
have to be defined.
3.2.5.5 Flight Parameter Envelopes for Structural
Design
The application of the single axis pitch, roll or yaw
maneuvers (MIL-A-08861A) is no longer sufficient for
the definition of design loads (Fig. 5 and Fig. 6).
0.0 0.5 1.0 1.5 2.0 2.5
-15
-10
-5
0
5
10
15
Time (s)

Y

T
i
t
l
e
Vertical Load Factor
Angle-of-Attack
Pitch Rate
Taileron
Pilot Input
Fig. 5 MIL - Pull-Push Maneuver
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0
-6
-4
-2
0
2
4
6
8
Time (s)

Vertical Load Factor
Angle-of-Sideslip
Lateral Load Factor
Taileron
Roll Rate
Roll Acceleration
Yaw Rate
Pilot Input
Fig. 6 MIL - Rolling Pull Out Maneuver
The carefree maneuver capability with automatic load
protection allows the superposition of combined pilot
control inputs in roll, pitch and yaw and with it numerous
different operational maneuvers which have to be taken
under consideration to find the critical design loads.
Some typical pilot stick inputs for flight clearance
maneuvers are shown in Fig. 7.
44
Fig. 7 Typical Pilot Stick Input
The following Flight Parameter Envelopes have to be
defined (s. Fig. 2):
n
z
= f(q
dot
)
n
y
= f(r
dot
)
n
z
= f(p, p
dot
, r, r
dot
, ny, *q
dyn
)
p, r vs. p
dot
, r
dot
As it can be seen mainly the inertia dominated parameters
as the translatory accelerations (n
z
, n
y
) and the rotational
velocities (p, r) and rotational accelerations (p
dot
, q
dot
, r
dot
)
have to be defined while only one aerodynamic
parameter is q
dyn
(sideslip angle dynamic pressure).
The sideslip angle is well controllable by the FCS and
with it the product q
dyn
. q
dyn
can be defined under
consideration of the gust requirements for the aircraft.
Important for the definition of the Flight Parameter
Envelopes for the structural design of an aircraft are also
the possible tolerances of the flight parameters (s. Fig. 8).
These have to be defined
Fig. 8 Flight Control System Design -
Tolerance of Flight Parameter
- For example:
to define n
zmax./min.
for the most important Flight
Parameter Envelopes
n
z
= f(q
dot
)
n
z
= f(p, p
dot
, r, r
dot
, ny, *q
dyn
)
it should be known how exact the FCS controls the
vertical load factor n
z
(s. Fig. 8):
n
z
= n
z max./min.
t n
z
If in this case the defined tolerances are to small an
increase of the n
z
overswing (tn
z
) may cause
problems, because the load limiting procedure of
the FCS can become uncertain therefore or on the
other hand an increase of the critical aircraft loads
has to be accepted for which the aircraft structure
has to be checked for.
These Flight Parameter Envelopes will be used now to
determine the design load and the load envelopes for the
aircraft main components see Para. 3.2.5.8.
The interdependence between the Flight Parameter
Envelopes and critical design load cases for the different
aircraft components can be seen on Fig. 2.
3.2.5.6 Total Aircraft and Component
Aerodynamics
To get balanced load cases the total aircraft
aerodynamic as well as the corresponding component
aerodynamic is integrated in the Loads Model regarding
all loads critical aerodynamic influences. The result must
fulfil the condition:
- sum of component aerodynamics = total aircraft
aerodynamics
The following aerodynamic data sets are part of the
Loads Model:
- aerodynamic pressures of the total aircraft for all
aerodynamic influences (, , control surface
deflections, p, q, r, etc.) for different Mach numbers
- the corresponding aerodynamic
coefficients/derivatives of the aircraft components -
result of aerodynamic pressure integration for all
defined monitor stations (Fig. 9)
- the corresponding aerodynamic
coefficients/derivatives of the total aircraft sum of
component coefficient/ derivatives
- the static aeroelastic corrections of the aerodynamic
pressures for all aerodynamic influences as
, , control deflections, p, q, r, etc.
and the aerodynamic pressures of aeroelastic inertia
effects and the corresponding integration results
(coefficients/derivatives) for
n
z
, n
y
, p
dot
, q
dot
, r
dot
together with the correction factors and increments
for the aerodynamic coefficients/derivatives for the
aircraft components and the total aircraft
- the corrected flexible aerodynamic pressures
including the corresponding flexible total aircraft
aerodynamics and the flexible aircraft component
aerodynamics
45
The main programs for establishing the required
aerodynamic data sets and for data set handling are:
- a theoretical aerodynamic program (e.g. the Dasa
HISSS program higher order panel method) to
calculate the rigid aerodynamic pressures for the
above described loads relevant aerodynamic
influences.
In Fig. 10 it is shown how starting from a CATIA
model the HISS panel model will be derived.
- a correlation and integration program to compare
and correct the theoretical total aircraft aerodynamic
results up to first total aircraft wind tunnel
measurements and with it to correct the aerodynamic
Fig. 10 HISSS Panel Model of Demonstrator
Aircraft Calculation of Aerodynamic Pressures
for Total Aircraft
- pressures and the aerodynamic coefficients
/derivatives for the aircraft and the aircraft
components
- a static aeroelastic program to calculate the
aeroelastic pressure increments for the correction of
the rigid pressure distributions and to calculate the
correction factors and increments for the
aerodynamic coefficients/derivatives for the aircraft
components and the total aircraft to establish the
flexible aerodynamic data set.
an aerodynamic pressure summation program to
summarize the aerodynamic pressures due to
, , control deflections, p, q, r, etc.
for the selected critical load cases to calculate the
aerodynamic nodal point loads for the FE- Model.
3.2.5.7 Total Aircraft- and Component Masses
For the calculation of balanced load cases the mass
conditions for the defined design masses (Basic Flight
Design Mass, Maximum Design Mass, Minimum Flying
Mass, Landing Design Mass, etc.) for the total aircraft as
aircraft mass
aircraft c.g.
aircraft moments of inertia
as well as the corresponding component mass conditions
have to be integrated into the Loads Model.
- Sum of component masses = total aircraft mass
Fig. 9 - Load Monitor Stations for Demonstrator Aircraft and Corresponding Main Loads Components
46
The following mass data sets are part of the Loads
Model:
- the aircraft component masses, component c.g.s and
moments of inertia including the corresponding
internal fuel states and external stores (Fig. 9 A/C
Monitor Stations)
- the total aircraft mass, c.g., moments of inertia
including the internal fuel states and external stores
as sum of the above described aircraft component
masses
3.2.5.8 Aircraft Loads Monitoring
The calculation of critical design load cases (loads
monitoring) for the aircraft components (monitor
stations) can be started when the required input data sets
for the Loads Model are established. The outcome of the
aircraft loads monitoring are Loads Envelopes (Fig. 11)
for the defined monitor stations.
The computer program which will be used for the
calculation of critical load cases under consideration of
the defined Flight Parameter Envelopes is the so called
Balance Program. The loads analysis for the monitor
stations (Fig. 9) will be performed by means of user
defined dynamic equilibrium points (time steps of a time
dependent flight simulation):
- The user has to define for each load case the
following flight parameters
Mach number, altitude, n
z
, n
y
, p, p
dot
, q, q
dot
, r, r
dot
respecting the Flight Parameter Envelopes (Fig. 2)
and as a special case for this demonstrator aircraft
the foreplane deflection (
F/P
) and trailing edge
deflection (
T/E-sym.
)
under consideration of the foreplane schedule
- The Balance Program will define the remaining
ones:
, , -
T/E-sym.
or -
F/P
, -
T/E-unsym.
, -
R
and nx and the thrust level
if required. In a second step the corresponding air-,
inertia- and net- loads for all monitor stations are
computed for the selection of critical design loads
to establish the loads envelopes for the defined
aircraft components
To be sure that the defined requirements will be fulfilled
the program also checks
- the derived control surface deflection angles
compared to the max. deflection angles
- the derived hinge moments for the control surfaces
compared to the max. defined hinge moments if
necessary
- the user defined flight parameters compared to the
Flight Parameter Envelopes
It seems to be useful to establish a program for loads
calculations which can be used for different degrees of
freedom (DOF):
- 6 DOF balance of F
x
, F
y
, F
z
, M
x
, M
y
, M
z
- 5 DOF - without F
x
balance (tangential force)
- 3 DOF balance of F
x
, F
z
, M
y
for pure symmetric
conditions
- 2 DOF balance of F
z
, M
y
for pure symmetric
conditions without F
x
balance
It should also be possible later on in the aircraft clearance
phase when the carefree handling and load limiting FCS
is available to use a flight simulation program to do time
dependent loads critical flight simulations and to
calculate the corresponding flight load time histories (air-
, inertia-, net- loads for all time steps) for the aircraft
monitor stations with the Loads Model.
Fig. 11 Example of Loads Envelopes for Monitor Stations Design Load Cases
47
To fulfil the above described additional program check
functions the following margins have to be defined:
- max. deflection angles for control surfaces versus
Mach number
- max. allowable hinge moments for the control
surfaces respective max. normal forces if necessary -
as result of structural optimization of wing, fin and
foreplane
- engine thrust conditions if necessary
- Maneuver Load Alleviation (MLA) concept if the
FCS will have a MLA procedure to reduce the
wing bending moment respective the other in Para.
3.2.5.13 described load reducing FCS rules
- as a special case for this demonstrator aircraft the
foreplane trim schedule including possible
tolerances because the foreplane and the trailing
edge flaps will be used for symmetric flight control
3.2.5.9 Loads Process, Aircraft Design and
Clearance Phases
After the feasibility studies respective definition phase
the normal development process of an aircraft structure
has three phases:
- Design Phase
- Check Stress Phase
- Structural Clearance Phase
For these three development phases the accuracy of the
input data (aircraft masses, aerodynamic, etc.) for the
Loads Model differs and with it the accuracy of the load
calculations. But as explained before the standard of the
input data for the Loads Model is relatively high even at
the beginning of the aircraft development due to modern
computer tools (i.e. theoretical aerodynamic programs)
and the possible crossreading to other similar aircraft.
But more important is that with the Flight Parameter
Envelopes the principal flight maneuver requirements for
the aircraft can be defined very early and with it the
interaction of FCS and the aircraft loads. During the
development of the aircraft structure the Flight Parameter
Envelopes have to be checked in line with the FCS
development.
3.2.5.10 Design Phase
Before starting loads calculations with the 1
st
flexible
Loads Model in the Design Phase the in Para. 3.2.3.8
described prerequisites have to be settled additional to the
Flight Parameter Envelopes to be sure that the loads are
the critical ones and are not maximized:
- A structural optimization has to be done and with it
an optimization of the control surface efficiencies
under consideration of aeroelastic influences, failure
conditions and deflection rates (Fig. 12). Based on
these optimization studies the critical hinge
moments respective normal forces for the control
surfaces can be defined. The result of optimization is
configuration freeze.
- The max. deflection angles versus Mach number and
the maneuver conditions for the control surfaces
have to be defined for example the foreplane trim
schedule.
- A maneuver load alleviation (MLA) concept should
be defined if necessary under consideration of
the required reduction of wing root bending
moment for high g conditions
the trailing edge split flap schedule as function
of g respective
the foreplane trim schedule.
Fig. 12 Flexible Loads Model -
Static Aeroelastic Influences
If all these prerequisites are defined and integrated in the
Loads Model the load investigation can start.
During the Design Phase the Loads Model consists of
theoretical linear aerodynamics compared with first
windtunnel test results and corrected if necessary. The
flexible aerodynamic data set includes all important static
aeroelastic corrections for selected Mach/altitude points
(Fig. 13).
Fig. 13 Flight Envelope Mach-Altitude Points for
Flexible Loads Model Flexible Aerodynamic Data Set
The main benefit to do the load investigations with the
first flexible Loads Model is
- the loads for the aircraft components can be
calculated for total aircraft balanced conditions for
different aerodynamic configurations (with and
without stores) and different aircraft masses (fuel,
external stores) under consideration of the FCS
requirements (Flight Parameter Envelopes).
48
3.2.5.11 Check Stress Phase
The Check Stress Phase is the second development phase.
The design loads have to be checked and updated with
the updated Loads Model for the design of the production
aircraft structure:
- the panel model for the theoretical aerodynamic
calculations has to be updated (configuration
changes, external stores, etc.)
- the new theoretical linear aerodynamic has to be
updated by comparing and correcting it to the latest
windtunnel tests (configuration changes, additional
store configurations, mass flow, etc.)
- first windtunnel based store aerodynamic increments
can be available (store balances) and can be
included in the Loads Model
- the static aeroelastic corrections have to be updated
by using the updated structure (FE- Model) and the
updated aerodynamic pressures
- the aircraft masses have to be updated for
production aircraft standard
- the foreplane trim schedule and the tolerances for
the trim schedule have to be updated
- the MLA concept has to be checked and updated if
necessary
- the max. hinge moments for the control surfaces
have to be checked and updated if necessary
- if required additional monitor stations have to be
included in the Loads Model
- the Flight Parameter Envelopes have to be checked
and updated in line with the FCS development. That
means in detail that the flight control laws have to
be reviewed during all design phases to check their
function as a load limiting system. For example the
defined tolerances of the Flight Parameter
Envelopes have to be checked, e.g. the n
z
tolerances:
n
z max./min.
t n
z
as explained in Para. 3.2.5.5.
As for the Design Phase the load calculations have to be
done by using the Balance Program and the updated
Flight Parameter Envelopes. The up to now available
FCS has only a check function because the carefree
handling and load limiting procedures are not finally
agreed (preliminary carefree handling). The load
investigation should be expanded and additional
Mach/altitude points should be considered.
The revised aircraft component design load cases
(balanced load cases, load envelopes) from the Check
Stress Phase are the basis for the stress analysis for the
production aircraft and with it for the structural clearance
activities in the Clearance Phase.
3.2.5.12 Structural Clearance Phase
The aircraft clearance will be done in different steps from
the first flight clearance for the prototypes up to the
Initial Flight Training Clearance (IFTC) and the Final
Operational Clearance (FOC - 100% load level) for the
production aircraft.
The aircraft structure has to be cleared for the conditions
defined in the Structural Design Criteria as there are:
design flight envelope (Ma/altitude)
critical aircraft configurations
limit/ultimate load factor
aircraft design masses
n
z-max./min.
vs. Mach
etc.
Fig. 14 Allowable Load Envelope for Aircraft
Clearance Phases Structural Reserve Factors < 1.0 are
considered
For the clearance of the aircraft structure so called
Allowable Loads Envelopes (ALE) will be used. The
ALEs (Fig. 14) contain the structural information of the
prototypes respective of the production aircraft. The
ALEs have to be defined by the stress office based on
the design load envelopes of the aircraft components and
under consideration of the results from the stress analysis
and structural tests. To be on the severe side during the
clearance activities (flight test) only structural Reserve
Factors (RF) < 1.0 have to be considered in the ALEs.
The prerequisites to increase the clearance level are :
- Major Airframe Static Test (MAST) to limit,
ultimate, failure load condition and other aircraft
component tests - to check the aircraft structure
- FCS updates from preliminary carefree handling to
full carefree handling to check the load limiting
procedure of the FCS
- Validation of the Loads Model via the Flight Load
Survey to update the data basis for loads monitoring
and to proof also the load limiting procedure of the
FCS
The first Loads Model for the structural clearance of the
aircraft consists of non-linear aerodynamic data based on
wind tunnel pressure plotting measurements. The
validation of this non-linear Loads Model will be done by
the Flight Load Survey. The Flight Load Survey will be
performed for selected primary aircraft configurations
(clean aircraft and external store configurations). During
the Flight Load Survey aerodynamic pressures of the
surfaces (wing, foreplane, fin) and the fuselage will be
49
measured (Fig. 15). The integrated pressures
(aerodynamic coefficients for the total aircraft and for
aircraft components) will be correlated against the load
predictions from the non-linear Loads Model. The Loads
Model will be than corrected where significant
discrepancies exist. Finally the flight validated Loads
Model for the primary aircraft configurations is available
and should be used for the Final Operational Clearance
(FOC) 100 % load level and production FCS.
During the Structural Clearance Phase at all clearance
levels the confidence that the load level will not be
exceeded has to be shown by the loads monitoring of
loads critical flight simulations using the current FCS and
the validated Loads Model. Some typical pilot stick
inputs for the flight simulations (flight clearance
maneuvers) are shown on Fig. 7.
The loads from the simulated flight maneuvers have to be
compared to the Allowable Loads Envelopes for each
monitor station. If the loads monitoring shows that the
loads are inside the ALEs the clearance step is fulfilled.
If not:
- the areas have to be defined where control law
changes are required to maintain acceptable loads
or
- modifications may be necessary to improve the
aircraft structure for higher loads
3.2.5.13 Load Optimized Maneuvers
In the past the aircraft were optimized mainly to
aerodynamic performance conditions (drag, etc.) and the
design loads were the result of the aerodynamic
configuration, the aircraft mass conditions and the
application of single axis pitch, roll or yaw maneuvers
(e.g. MIL-A-08861A).
A new possibility for the latest high performance fighter
aircraft generation like Eurofighter are load optimized
maneuvers because the FCS can be used in some cases
for load reduction under the consideration that the aircraft
performance is not prejudiced.
Three examples for load optimized maneuvers controlled
by the FCS are given below:
1. Load optimized foreplane/trailing edge deflection
schedule as a special case for the demonstrator
aircraft described in this paper:
a) reduction of front fuselage loads
The front fuselage loads are normally
dominated by the inertia loads. To reduce the
front fuselage loads (F
z
-normal force and M
y
-
vertical bending moment) the foreplane has to
be deflected in that way that the aerodynamic
foreplane loads are acting against the front
fuselage inertia loads (s. Fig.16 ). In this case
the aircraft has to be controlled by the trailing
edge flaps.
b) reduction of trailing edge flap loads - e.g.
hinge moments.
For low g conditions (1g) where the maximum
roll performance of the aircraft is required the
trailing edge flaps can be zero loaded for the
aircraft trim conditions by trimming the aircraft
only with the foreplane. The trailing edge flap
itself has to be deflected in that way that the
influence on the flap will be compensated:

T/E-symm(nz=1.0)
= f(, Mach, A/C-cg)
With it the flap hinge moments can be reduced
and the roll efficiency of the aircraft can be
increased in some cases.
Fig. 15 Flight Load Survey - Pressure Transducers at the Prototype of Demonstrator Aircraft
50
Fig. 16 Front Fuselage - Load Reduction Load
Optimized Foreplane/Trailing Edge Schedule
Procedure a) may be used only for the front fuselage
loads critical flight conditions as high gs turns at low
aircraft masses (minimum flying mass) where the normal
aerodynamic discharge for the front fuselage is a
minimum and with it the net load is a maximum. In this
case the trailing edge flap loading is relatively low
compared to the maximum aircraft rolling conditions and
can be used therefore for exclusive aircraft control in the
pitch axis. In all other cases the aircraft performance will
be more important.
Procedure b) is a possible solution for hinge moment
reduction if the control surface loads are increasing and
the size of the flap actuators cannot be changed.
2. Maneuver Load Alleviation - MLA (differential
trailing edge flap deflection of i/b- o/b- flap):
the shift of the aerodynamic center of pressure
towards the wing root reduces the wing root
bending moment and with it the wing
attachment load conditions
In this case the i/b- flap has to be deflected
downwards to increase the wing lift in the
inboard wing area while the o/b- flap has to be
deflected upwards to reduce the lift in the
outboard wing area under the condition that the
total wing lift has not to be changed (s. Fig. 17).
This differential trailing edge flap deflection
has to be superimposed to the full span trailing
edge flap trim condition. The small effect on
the aircraft trim conditions by using the MLA-
system has to be corrected by a full span
trailing edge deflection itself or by the
foreplane.
Fig. 17 Maneuver Load Alleviation (MLA) Change of
Wing Lift Distribution and Shift of Center of Pressure
The MLA- system could be important at high
gs and high dynamic pressure in the lower -
region (elliptical wing lift distribution, linear
aerodynamics).
At higher there may be a natural shift of the
center of pressure to the wing root because the
wing lift distribution becomes more and more a
triangle due to non linear aerodynamics. (s. Fig.
18).
Fig. 18 Spanwise Normal Force Distribution Natural
Shift of Center of Pressure to the Wing Root
The MLA- system can be important for the critical wing
up bending conditions at max. gs for the static design
respective the most critical gs (mean proportional gs)
for fatigue design because the aerodynamic design often
didnt allow to increase the lever arm of the wing root
attachment to carry over the wing bending moment by a
couple of forces (s. Fig. 19).
3. Prevention of overswing of control surfaces
(deflection angles):
to prevent load peaks on the control surfaces
during rapid aircraft maneuvers (e.g. rapid
rolling) an overswing of the control surfaces
should be avoided. An example for the trailing
edge flap is shown on Fig. 20. In this case the
overswing of the flap is optimized by a small
change of the T90 condition and with it the flap
loads (hinge moments) are reduced
significantly.
Fig. 19 Wing Root Carry Over of Wing Bending
Moment
The above described maneuvers can be defined for the
critical static design loads as well as for fatigue loads
which becomes more and more important for the
structural design of the aircraft.
51
In all these cases it must decided whether the load
optimized maneuvers sacrifice aircraft performance or
whether the benefit (i.e. mass saving) is big enough to
compensate the loss of performance!
Fig. 20 Dynamic Overswing of Trailing Edge Flaps
Change of T-90 Conditions
One way to assess this question is to evaluate required
operational maneuvers with respect to extreme or fatigue
maneuvers as evaluated by the former AGARD-WG 27
(AGARD AR 340). For further information see Chapter
3.2.2 Operational Flight Parameter Approach.
On the other hand the q
dyn
requirement defined in the
flight parameter envelopes (s. Fig. 2) is also a load
limiting condition controlled by the FCS as explained in
Para. 3.2.5.5. With it the Fin loads and the side force and
side bending moment of the rear and front fuselage can
be limited.
3.2.5.14 Ultimate Load Factors
Historically a reduction of the ultimate load factor f
ult.
was done several times down to f
ult.
=1.5 now which was
for a long time seen as the lowest possible limit.
The situation was changed for FCS controlled aircraft
with carefree handling and load limiting procedures.
Based on the assumption that the aerodynamic and inertia
flight loads for the aircraft are limited by the FCS by
controlling the important flight parameters
, p and n
z
respective
directly the ultimate load factor can be reduced for
example from
f
ult.
=1.5 to f
ult.
=1.4
(as agreed with the British-, German-, Italian- and
Spanish- authorities for the Eurofighter)
But as explained in Para. 3.2.5.12 an extensive Flight
Load Survey has to be done to verify the load limiting
procedure of the FCS and to proof the reduction of the
ultimate load factor.
For FCS independent loads (e.g. landing gear loads,
Hammershock pressures, etc.) the ultimate load factor
will still be 1.5.
For further information about the ultimate load factor see
Chapter 3.1.3 Safety Factor Review.
3.2.5.15 Conclusion
The calculation of aircraft loads under consideration of
Flight Parameter Envelopes is useful and practicable for
modern high performance fighter aircraft with a carefree
handling and load limiting FCS.
As demonstrated for the Eurofighter:
- the integrated design of FCS and aircraft structure is
possible
- the carefree handling and load limiting procedure of
the FCS is working
- the defined design loads by using the Flight
Parameter Envelopes are acceptable and leading to a
robust but not to conservative design of the aircraft
structure - compared to the loads evaluated with the
FCS (time dependent flight load simulations) later
on in the A/C- Clearance Phase the design loads are
well
- the reduction of the ultimate load factor from f-
ult
=
1.5 to f-
ult
= 1.4 based on the FCS- load limiting
function is useful and leads to a lighter aircraft
structure
On the other hand the enormous increase in system
complexity for a modern high performance fighter
aircraft with a carefree handling and load limiting FCS
leads to extensive investigations:
- the flight control laws have to be reviewed during all
design phases to check their function as a load
limiting system
- the necessary careful and accurate load
investigations during all design phases are very
extensive
- an extensive Flight Load Survey has to be done for
Loads Model validation and with it to proof the load
limiting procedure of the FCS and additional if
necessary to proof the reduction of the ultimate load
factor
- the ALE concept has to be verified by detailed stress
analysis, static test and possible restrengthening of
the aircraft structure
As explained above the permanent monitoring of the
structural design parameters as Flight Parameter
Envelopes, ALEs, etc., is indispensable to minimize the
risk of a non optimal structural design of the aircraft.
Therefore it should be emphasized once more that
various disciplines as Loads, Aeroelastics,
Flightmechanics, Flight Control, Stress, Aerodynamics,
Flight Test have to cooperate in a very close manner, the
so called concurrent aircraft engineering.
52
3.3 Dynamic Loads
3.3.1 Introduction
The intention of this chapter is to discuss the prediction
of unsteady loads arising as a result of pilot actions (as
opposed to atmospheric turbulence, say). Gusts and
ground loads are treated in separate chapters. Loads due
to buffet and buffeting, hammershock, gunfire and store
ejection/release loads are mentioned. The aim is met by
briefly describing the background, prediction processes
and calculation methods, and certification issues.
Consideration of the latter is essential, even at the design
stage. In addition, the likely way forward for this
technology is noted. A table is provided as a guide for
consideration of dynamic loading sources and their
effects on an airframe.
In addition, examples of dynamic load analyses and
testing for validation purposes are given in section 3.4,
whilst birdstrike is discussed in 3.5. The latter does not
strictly come within the terms of this chapter, but is
classified under threats. However, it is such a
significant source of aircraft in-service incidents, and
hence a driver of future designs, that it is included here.
In the course of the item, reference is made to some
specific papers and work known to the author. However,
it should be noted that hundreds of technical papers
relating to the overall subject are available world-wide.
Since there are several approaches documented, this
chapter does not make prescriptive statements regarding
the correct approach. Rather, readers are encouraged
to adopt information and data applicable and appropriate
to their own specific technical challenges. The aim is to
raise awareness, not define methods in detail.
The airframe static load can be thought of as one that
changes only with flight condition e.g. airspeed, angle of
incidence, altitude etc. For the purposes of this report,
the airframe dynamic load component can be considered
to be the oscillating part of the load which has a
frequency in the range 2 - 100Hz. This is not a hard and
fast rule. However, loads oscillating below 2Hz can be
considered to be due to 'rigid body' motion. Above
100Hz, the load is unlikely to be adversely affecting a
major structural item, more likely to be a localized effect
e.g. an acoustic, stores or equipment environmental
effect.
There are many sources of dynamic loads on a military
combat aircraft. Traditionally, combat aircraft were not
designed and optimized to the degree that is expected
today. Dynamic effects were therefore included in the
early design phases of an aircraft project by applying a
factor to the static design loads (which were usually
maneuver defined for combat aircraft). The pessimism
that this introduced could be tolerated and covered the
majority of dynamic loading effects. It was only when
structural or equipment problems emerged during project
development, or even in-service, that dynamic loads were
considered in more detail. This situation was
compounded by an absence of advanced unsteady
response prediction tools.
The performance of modern military combat aircraft has
increased, taking the airframe into situations where the
airflow over the structure becomes separated and
oscillatory. The unsteady environment to which a
modern airframe is subjected has therefore become
increasingly harsh. At the same time, a requirement
exists to reduce the factors applied to the design loads to
drive down structural mass. The need to predict the
unsteady load component more accurately, to ensure
safety, has therefore become correspondingly more
important. To that end, modern military combat aircraft
are designed to withstand the worst static and dynamic
load cases which they are likely to encounter in-service.
This has led to some regions of modern combat aircraft
structures being designed by dynamic load cases.
3.3.2 Types of Dynamically Acting Loads
3.3.2.1 Buzz
Buzz is a single degree of freedom flutter whereby
limited amplitude oscillations of surface panels or control
surfaces occur due to a loss in aerodynamic damping and
may involve the local resonance of such surfaces. This
loss is attributed to boundary layer and shock wave
induced instabilities in the surrounding flow field.
Examples of such instabilities include oscillations of
shock waves over a control surface and separated flow
caused by an upstream shock wave.
Although the limited amplitudes of oscillation associated
with buzz phenomena do not cause catastrophic structural
failure, as can happen with a two (or more) degree of
freedom flutter, structural fatigue can arise. Common
solutions to reduce the adverse effects of buzz
phenomena include manipulation of the flow field (e.g.
using vortex generators) to reduce instabilities and
stiffening of the control surface hinges to reduce freeplay.
3.3.2.2 Buffet and Buffeting
Buffet is an excitation caused by the separation of air
flow over a surface. This can be separation in an
unsteady manner causing excitation of the surface from
which it is separating, or separation from upstream
components such that the resulting unsteady flow
impinges upon a downstream surface. This is worse at
high angles of attack. Buffeting is the associated
airframe structural response. Buffet and buffeting are
phenomena that are unavoidable in highly maneuverable
combat aircraft.
For many years fighter aircraft have had to penetrate into
the buffeting region of the flight envelope in order to gain
maximum turn performance. With conventional control
systems, the buffet onset was in many ways a useful
feature because it provided the pilot with a clear warning
that he was approaching the limits of aircraft
controllability. Increasing buffet penetration, for instance
by increasing angle of attack, is also accompanied by
related characteristics such as wing-rock and nose slice.
With the advent of complex, active flight control
systems, modern aircraft can remain controllable well
beyond traditional boundaries, and even into post-stall
conditions. This has implications upon structural design
due to the potentially greater time spent in unsteady flow
53
conditions (fatigue implications) and the large magnitude
of these unsteady loading actions (strength).
Consequently, the ability to predict these flows has
assumed a far greater importance in aircraft design.
Another consequence of active flight control systems is
the potential for affecting the structural response under
unsteady loading conditions. If the system interprets
structural response as aircraft response and tries to
correct it by driving the controls, then there is a potential
for increasing the loads on the structure. This area of
expertise is known as Aero-servo-elasticity (ASE) or
Structural Coupling. A well-designed flight control
system (FCS) will not exhibit such adverse
characteristics. It is not a design driver when assessing
loads, but an awareness of the total system (aircraft +
FCS) characteristics is required for flight clearance work.
Ways of using active control for reducing structural
response to unsteady loading, like buffet, are under
consideration. A view of this is given in reference 1.
The above is applicable to combat aircraft. However,
buffet also occurs due to impingement of vortical and
wake flow on downstream surfaces, separated flow over
control surfaces, and flow interaction between adjacent
stores (or engines), their pylons and other airframe
structure, to name a few generic examples. These are not
restricted to highly maneuverable aircraft. Indeed,
straight and level flight at transonic conditions, on any
class of aircraft, can lead to complex shock-boundary
layer interactions, which induce separated flow and hence
buffet, i.e. a forced response.
Further buffet inducers include excrescence and
cavities. Examples of the former include blade aerials,
chaff/flare dispensers, auxiliary cooling system intakes
and exhausts. Flow separation occurs from these unless
they are carefully designed, and faired-in specifically to
avoid this phenomenon. The result is unsteady pressure
fluctuations on surrounding, external paneling and
surfaces. The risk here is that surface panel modal
frequencies can be excited which can lead to rapid
fatiguing of the affected structure.
Flow spillage from cavities can have similar effects. The
cavities can be those occurring when the landing gear is
deployed, or when internally carried weapons are
released. The latter is likely to be much more of a
problem due to the wider range of flight conditions at
which it may occur.
Further, there is much potential for adversely affecting
the internal and back-up structure of the weapons bay due
to acoustic effects. Similarly, stores and equipment
installed in the bay will have difficult environmental
clearance issues to overcome. Control of such acoustic
environments is a major study area.
3.3.2.3 Hammershock
Hammershock (H/S) is an event whereby an aircraft
engine surges, sending a pressure pulse upstream,
opposing the direction of airflow that would exist during
normal engine operation. This results in a loss of engine
performance, the possibility of a flame-out and/or
permanent engine damage.
H/S events can occur anywhere within a combat aircraft
flight envelope but are more significant at the envelope
extremities. They have many causes. These include:
over-fuelling;
bird strike;
foreign object ingestion and
disturbed intake airflow (e.g. wake ingestion).
A single surge may occur or a series of pressure pulses
may be generated if the surge becomes 'locked-in' i.e.
conditions are such that repeated surges occur.
The pressure pulse created impinges on the engine intake
and on the forward fuselage. Both of these items must
have sufficient strength to withstand a H/S event. This is
particularly critical for aircraft which have foreplanes
located in the path of the pulse. The concern here is that
a locked-in surge may occur with a pulse frequency close
to a fundamental foreplane vibration mode. If an item of
structure is excited at a frequency near one of its natural
vibration modes (i.e. a resonant frequency), the resulting
amplitudes of vibration and hence load are large.
Realistic prediction of the excitation can be achieved by
deliberately surging an engine on the ground and
measuring the resultant pressure pulse amplitudes in the
intake duct, splitter plate/lip regions and forward of the
intake. Account can then be taken of airspeed, altitude
etc. to derive excitation throughout the desired flight
envelope. Wind tunnel testing is an alternative approach,
but scale effects are significant, and can lead to major
over-prediction if not accounted for adequately.
H/S was considered during the development of EAP
(shown in Figure 1). This resulted in the foreplanes being
modified to prevent them 'tuning' with the predicted pulse
H/S frequency. This proved to be overly cautious. The
actual pressure pulses dissipated more quickly than was
anticipated or had been measured in the wind-tunnel.
This experience, of course, can be used on future aircraft
projects.
54
Figure 1 : EAP Technology Demonstrator
3.3.2.3.1 Influence on inlet duct design
Examples of load cases on the inlet duct include
maneuver g-loads, steady state pressures and
hydrostatic pressures of neighboring fuel tanks. However,
the pressure loads acting on the inlet duct caused by the
propagation of the high velocity pressure wave(s)
associated with surge phenomena is the predominant
design factor for combat aircraft.
The majority of modern combat aircraft utilize
rectangular, or other non-circular, shaped inlets with a
gradual longitudinal change into a circular shape duct in
order to merge effectively with the engine face. The H/S
loads become critical for such variable duct geometry due
to complex load paths in the throat region and stress
distributions around the corners of, say, a rectangular
inlet. The H/S loads associated with the circular duct
sections produce hoop tension and are less critical.
From reference 2, two aspects of H/S phenomena which
are of importance to the dynamic response of the intake
duct structure are (i) magnitude of the pressure wave and
(ii) the rise time to positive and negative peaks. It should
be noted that the negative peak is caused by the reflected
H/S pressure wave at the forward intake. Figure 2 shows
a typical example of a H/S excitation time history in
which the vertical axis represents the ratio of incremental
H/S pressure to maximum incremental H/S pressure and
the horizontal axis corresponds to the H/S pulse duration
().
The characteristics of H/S loading as described above
leads to the consideration of dynamic magnification of
loads during duct design, especially when taking into
account of locked in surges. This is due to the potential
of a pulse sequence having repetition frequencies which
could coincide with the natural frequencies of the duct
paneling.
Conventional approaches of designing ducts to cope with
H/S loads include increasing duct skin thickness and
employing additional ring stiffeners around the duct in
between the frames. Furthermore, special attention is
made to the local design of frames and stiffeners in the
rectangular sections of the duct as well as axial fastener
and bond peel strengths which could result in localized
structural strengthening. Approaches such as these serve
to increase duct weight: an undesirable trend.
Rectangular inlet Circular duct at engine face
55
Figure 2 Characteristics of Hammershock loading
Another aspect of duct design in relation to H/S
phenomena is the attenuation of pressure waves.
Attenuation is key to the reduction of pressure loads
acting throughout the duct, particularly in critical areas
such as frontal inlet region. Two processes (detailed
discussion provided in reference 3) which can relieve
pressures are (i) airflow bleed through a bypass exit
which reduces diffuser volume and (ii) ramp edge
leakage to the plenum allowing pressure transmissions at
sonic velocity. However, trade-off studies must be
conducted to determine the feasibility of duct weight
reduction due to the alleviation of pressure loads, against
the losses in intake efficiency during operation of the
bleed / leakage processes, and the weight increases due to
implementation of the more complex mechanisms
involved.
3.3.2.4 Gunfire
This is an obvious source of high energy, short duration
dynamic loading. Attention is traditionally given to
designing structure to absorb recoil forces transmitted to
it, whether from an internal or pod-mounted installation.
Conventional metallic structure, with its joints and
fastenings, tends to absorb energy (via damping and
friction) better than extensively bonded designs. Hence,
transmission of loading is limited. With bonded
structures the recoil effects can affect a much larger part
of the airframe. This gives the potential for tuning with
modal frequencies, and hence loading problems.
Muzzle/exhaust blast could increase this effect if
transmitted through a significant part of the airframe. It
could be possible for some parts to be loaded by both the
recoil forces and the blast effects. Even if this is not the
case, the blast effects on localized external structure
should be assessed. Again, tuning with panel modal
frequencies is a possibility given the current range of
gunfire rates. From the blast impingement point of view,
pod mounted guns are usually better. Almost by
definition, they are mounted such that the gun muzzle
will be further away from the aircraft. This would be
expected to allow some dissipation of the blast energy
before hitting the nearest parts of the airframe.
3.3.2.5 Store Release / Jettison / Missile Firing
Stores release can vary from jettison of fuel tanks to
missile firing activities. Stores release design cases are
few and far between, but the possibility must be
considered. The effects of store release during extreme
maneuvers must be assessed.
Excitation of the airframe arises from the 'kick' provided
by the loss of mass during release, this effect being
directly in line with the mass of the store, and also from
the ejector release units which push the store away from
the aircraft. Unlike buffet, gunblast and H/S excitation,
the point of application of a release impulse to the
structure is more localized. However, the effect can be
just as global if significant transmission through the
airframe is possible, as discussed in the previous section
on gunfire.
Special design consideration must be given to 'ripple'
store releases i.e. multiple stores released in rapid
succession. This may be required to give a wide
munitions coverage of the target or as part of an
emergency stores jettison sequence. As with H/S events,
the proximity of release 'pulses' could have an excitation
frequency close to a major airframe vibration mode. The
result would be large structural oscillations. This implies
large structural loads but would also affect 'dumb' store
delivery accuracy.
P
HS
/ P
MAX HS
Time
1.00
0.00
-0.40

56
3.3.3 Prediction Process & Methods
3.3.3.1 Loads Prediction and Simulation
The main emphasis here is about primary lifting surfaces
undergoing general bending and torsional responses due
to a dynamic loading action, eg. buffet excitation.
Localized loads use similar principles, but may not need
a full aero-structural simulation. This depends upon the
needs of the technical problem being addressed.
There are 2 major approaches. The first is empirical, and
assumes that the new design is similar in general nature
to a previous project for which there exists an adequate
database of information.
The second approach

can be classed as the theoretical
approach although it does not yield an exact solution; the
accuracy being dependant upon the quality of the input
data, and the inherent assumptions regarding linearity of
characteristics.
3.3.3.1.1 Empirical Approach
An example of a successful use of an empirical approach
is that of designing EAP to account for fin buffeting.
Figure 3 illustrates how an initial prediction of structural
response can be carried out. From Tornado measured
characteristics, an estimate of EAP fin response was
made. It assumes that the dominant parameters affecting
the fin response are wing sweep angle, incidence, and
dynamic pressure.
Incidence (AoA)
EMPIRICISM
Fin vibration characteristics
TORNADO
( =45
0
)
TORNADO
( =25
0
)
EAP
( =57
0
)
FIGURE 3. Fin Vibration Characteristics
Actual numbers on the axes are removed to preserve the
unclassified nature of this document. However, use of
the original plot will lead to the response on the EAP fin
for a given flight condition. Assuming a detailed
knowledge of the fin structural characteristics, then the
internal structural loads can be derived. This was
successful because of the large amount of information
generated, and hence available, in the course of studying
fin buffeting on Tornado.
As stated before, there is a large amount of publicly
available information which could allow derivation of
empirical methods for other projects. The example given
would not, of course, be applicable to twin fin designs, or
if the new fin structure (and, hence, modal response) was
radically different.
3.3.3.1.2 Theoretical Approach
This approach requires a numerical model of the structure
(inertia, damping and stiffness), numerical representation
of the oscillatory aerodynamics (damping and stiffness)
and numerical representation of the forcing function (eg.
buffet excitation).
The mathematical equation to be solved is of the
following form
Ax V Bx V Cx Dx Ex F t
E E
( ) + + + +
2
where
A = generalized inertia matrix
B = generalized aerodynamic damping matrix
C = generalized aerodynamic stiffness matrix
D = generalized structural damping matrix
E = generalized structural stiffness matrix
V
E
= equivalent airspeed
x = generalized co-ordinates
= relative air density
F(t)= generalized forcing function
Post-processing of the output from the response solution
leads to derivation of loads at defined points on the
structure. The process is shown diagrammatically in
figure 4.
57
NASTRAN, or In-Company developed alternative, is
used as the analytical tool for the calculation technique
shown above.
There are several points to note. In current practice, the
unsteady aerodynamics and structural models are linear
approximations. Development of improved, advanced
aerodynamic methods is discussed later. For early design
information there is unlikely to be detailed structural and
mass data available. In addition, the excitation function
may well be derived from existing databases pending
availability of wind tunnel test data.
For the detailed design and clearance phases of a project
the response model is likely to be the same as that used
for Flutter assessments. During the clearance phases of a
project, it should be possible to include a structural model
matched to reflect GVT data. The excitation data will
probably be based on wind tunnel testing of the finalized
project lines. However, it will still be subject to scaling
from wind-tunnel to full scale, as well as normal wind
tunnel accuracies. This is for a rigid wind tunnel model
and is illustrated in figure 5.
An interesting, but less used variation of the above, is to
create a dynamically scaled, flexible wind tunnel model.
This involves scaling the full size structural
characteristics to the model, but does mean that the
surface forces and moments can be measured directly.
There is still the problem of then re-scaling to full size in
order to derive the full scale loads.
The first approach is likely to be used earlier in the
design cycle. Unless the new aircraft is a development of
an existing type, detailed structural information will not
be available for manufacture of the flexible wind tunnel
model. The latter is also likely to be more expensive
because, in addition to increased model manufacturing
costs, a dedicated set of test runs will be required. The
rigid data can possibly be acquired on a ride-along basis
with other testing.
Alternative
Aerodynamic
Theoretical
Methods
Buffet Exction
Figure 4 : Buffeting Response Calculation Process
(Generalise from fig. 16 of Ref. 3)
Finite Element
Model
Modal Vib.
Characteristics
Doublet-Lattice
Aerodynamics [Nastran
based or in-house]
Dynamic Response
Loads
Derivation
[Forces &
Moments]
Figure 5 : Development of Buffet Excitation
Tunnel Test of Rigid
Model
Pressure Measurement Buffet Excitation W/T to Full- Size
Scaling
58
A useful guide to the state-of-the-art for numerical
aeroelastic simulation techniques is reference 4.
3.3.3.1.3 Hybrid W/T - CFD Techniques
Reference 5 is experimentally based and gives a good
summary of the aerostructural buffet problem. As it
points out, testing is expensive. Ideally, given the
advances in computing power in recent years, increasing
maturity of steady CFD techniques and accelerating
interest in unsteady CFD, then it should be possible to
replace some of the wind tunnel testing essential to
reference 5 and generally improve accuracy of the
aerodynamic predictions.
Researchers are now beginning to develop these
approaches. Until unsteady CFD techniques are more
mature, a pragmatic approach is needed to allow the
engineer (as opposed to the researcher) a means of
addressing buffet and buffeting early in the design
process. Hence, a combination of steady CFD analysis
with unsteady pressure measurements from wind tunnel
testing is a realistic approach. There are still some
problems, most notably prediction of aerodynamic
damping levels during buffeting at higher incidences.
3.3.3.1.4 Superposition of Steady and Unsteady
Loading
The above treatment relates to derivation of the unsteady
excitation. However, it is the total response, and hence
loading, that we are interested in from the structural
design and clearance point of view.
An aircraft operating on the ground or in flight
encounters two distinct types of loading - static and
dynamic. Of course, the airframe structure itself cannot
distinguish between the two loads. It is subject to the
combination of them, the total load.
Design activities are affected by available prediction
tools and techniques. It is common practice, for the
purposes of aircraft design and clearance activities, that
the two types of loads are calculated discretely. These
are then combined to give total predicted load. Figure 6
shows the principle diagrammatically.
It is important to ensure a coherent approach. There are
different ways of achieving the same result by assuming
that the principle of superposition holds (see table
below).
-3.5
-2.5
-1.5
-0.5
0.5
1.5
2.5
3.5
4.5
5.5
6.5
0
0
.
2
0
.
3
0
.
5
0
.
7
0
.
9
1
.
1
1
.
3
1
.
5
1
.
7
1
.
9
TIME
L
O
A
D

(
k
N
)
-3.5
-2.5
-1.5
-0.5
0.5
1.5
2.5
3.5
4.5
5.5
6.5
0
0
.
2
0
.
3
0
.
5
0
.
7
0
.
9
1
.
1
1
.
3
1
.
5
1
.
7
1
.
9
TIME
L
O
A
D

(
k
N
)
-3.5
-2.5
-1.5
-0.5
0.5
1.5
2.5
3.5
4.5
5.5
6.5
0
0
.
2
0
.
3
0
.
5
0
.
7
0
.
9
1
.
1
1
.
3
1
.
5
1
.
7
1
.
9
TIME
L
O
A
D

(
k
N
)
DYNAMIC LOADS - A DEFINITION
Nominal frequency range 2 -100 Hz
Figure 6: Superposition of Steady and Unsteady Loads
Quasi-Steady Loads Simulation Methods Dynamics Simulation Methods
1. Time varying throughout manoeuvre ie.
rigid body steady manoeuvre loads
Incremental loads due to unsteady effects on a
flexible structure
2. Constant loads from starting point of
manoeuvre
Incremental loads due to time varying rigid
body motion + Incremental loads due to
unsteady effects on a flexible structure
3. - Total loads due to time varying rigid body
motion + loads due to unsteady effects on a
flexible structure + FCS
Steady + Dynamic = Total
Nominal frequency range 2 - 100 Hz
59
These approaches are driven by pragmatic applications of
available methods and tools. It is a recognition that not
all organizations have the latest available technology and
computing power. Indeed, the third approach above is
only recently becoming more common as tool sets and
design processes become more integrated. For instance,
formerly it might have been necessary to have separate
methods for development and analysis of structural,
aerodynamic and FCS models. If consideration of other
disciplines was necessary, each would probably model
the others in its own home environment. This led to a
number of notionally similar numerical models being
developed - each needing extensive quality assurance and
checking, and none of them fully compatible.
As stated before, there is no definitive method. Readers
must judge the appropriate way forward for their own
particular projects. However, it should be noted that
some aspects of 1 and 2 above are favourable because the
quasi-steady loads can be based upon more mature,
speedier, theoretical methods (CFD) than unsteady
loading. In addition, for similar reasons there are likely
to be more extensive wind tunnel test data available.
3.3.4 Design Assumptions, Criteria and
Certification
Reference 6, gives a very brief overview of important
dynamic loading phenomena that should be considered
during the design of combat aircraft. It notes, however,
that specific design and certification criteria/guidelines
are few.
This can lead to lengthy discussions with Customers and
Certification Authorities about what should be addressed
in design and certification of a given aircraft project.
Experience has shown that an open-minded approach at
the design stage, which can include work that positively
eliminates a phenomenon from consideration, will ensure
a smoother progression, later in the project cycle, to flight
clearance and qualification. In short, at present there are
no hard rules governing consideration of dynamic loading
in structural design, other than that it should be taken into
account!
As engineers, we are bound to consider these loading
actions because they can be significant. This is
illustrated by the technical papers covering fin and tail
buffeting on F-18, and similar aircraft, which are
numerous (e.g. references 5, 7, 8, 9 and 10 picked nearly
at random from a wide choice). Wing buffeting is a well
known phenomenon, and also well documented. It is
clear that buffeting must be examined in the early stages
of design for aircraft with significant maneuver
capability. The problem for other areas is deciding what
is an acceptably low risk for a given set of circumstances.
Often, there are little data available which can be
analyzed effectively.
It is stressed that the reader must decide what is
appropriate for his particular work. It must be clear what
the latest design criteria are, and what is applicable to a
given project. If standards change through the life of an
aircraft project, this can lead to a very complex
documentation trail!
USE OF UNSTEADY CFD IN EXCITATION
PREDICTION
Databases
Experimental
W / T
Flight
TRADITIONAL
Steady CFD
W / T Unsteady
Pressure
Measurements
PRESENT
Unsteady CFD For
Magnitude and
Frequency Content
( Excitation Response
Structural Interaction)
FUTURE
Figure 7: Use of Unsteady CFD in Excitation Prediction
3.3.5 Developments
The above figure illustrates the changing approach to the
use of CFD in the prediction and simulation of dynamic
loading phenomena. The overall thrust has been to be
able to use CFD to replace/supplement wind tunnel
measurements for prediction of buffet, and other,
unsteady excitation. In addition, use of CFD for
improved response aerodynamics (particularly damping )
increasingly allows assessment of aerodynamically non-
linear effects. Key to this capability on the response side
is the unsteady CFD/structural modeling interfacing
methods. This is available at research and academic
60
levels, but is not yet sufficiently robust or rapid for
production application.
Reference 11 gives an outline of some work done in the
UK to address the shorter term requirements of engineers.
It reports on the combination of an extensive set of wind
tunnel tests with the aim of providing insight into the
aerodynamic phenomena associated with novel wing
planforms. These planforms impact both steady and
unsteady aerodynamics.
The wind tunnel tests have produced steady pressure
distributions, overall forces and moments, surface oil
flow patterns and unsteady surface pressure frequency
spectra. The steady flow results have been compared
with output from converged Reynolds Averaged Navier-
Stokes (RANS) CFD solutions.
The work has enabled a design tool to be proposed for
use early in the design process. For an arbitrary wing
planform, at maneuvering conditions, steady CFD can be
used to establish mean flow topology, including tracking
of vortex shear layers. Empirical representations of the
characteristic buffet frequencies can then identify the
dominant frequencies of the dynamic loads. When
coupled with relatively simple finite element models,
predictions of buffeting response are expected to be
sufficiently accurate to enable meaningful evaluation and
comparison of different wing planforms.
3.3.6 Summary
The above discussions are aimed at raising awareness of
dynamic loading effects, and their prediction, which is
advisable to consider at the design stage of an aircraft
project. Historically, this has not been so prevalent, but
is necessary now due to the requirements to more
effectively optimize structures, from both a strength and
fatigue point of view. Indeed, active control of structural
response (due to buffeting, say) is under very energetic
research and must now also be considered as a possible
option at the design stage of an aircraft project.
Because of the immense breadth of the subject, there are
no definitive statements here. Readers are required to
formulate their own approach to their own particular
technical challenges.
It is apparent that wind tunnel and CFD methods are vital
to future prediction techniques, particularly of non-linear
aerodynamic effects. However, examination of non-
linear structural effects (e.g. control surface backlash
characteristics) as part of the overall aero-structural
system are dependant upon more robust and rapid
techniques for coupling CFD with a FEM than are
available at present.
The table below is intended as an aide memoir. It
summarizes different types of dynamic loading and
which parts of an aircraft they affect. It includes gusts
and ground operations for completeness, although these
are described in different chapters.
61
SOURCE OF
LOADING
COMPONENTS AFFECTED TYPES OF AIRCRAFT /
COMMENTS
ATMOSPHERIC
TURBULENCE / GUSTS
WING
FORE / TAIL PLANE
FIN
FUSELAGE
CREW
EQUIPMENT
STORES & PYLONS
SENSORS & PROBES
HIGH SPEED AIRCRAFT WITH
RELATIVELY LOW WING
LOADING
BUFFET / BUFFETING /
BUZZ
WING
FORE / TAIL PLANE
FIN
STORES & PYLONS
LOCALISED EFFECTS
eg. Excrescences
Panels
Sensors & Probes
Airbrake
ALL TYPES, BUT
PARTICULARLY THOSE WITH
SIGNIFICANT AoA AND
MANOEUVRING CAPABILITY
Bluff shaped excrescences mounted
on large panels
STORES RELEASE &
JETTISON
WING
FUSELAGE
PYLONS
ATTACHMENTS &
BACK-UP STRUCTURE
ALL TYPES
MISSILE FIRING As above +
PLUME EFFECTS on
Local panels
Control surfaces
Tailplane etc.
ALL TYPES
HAMMERSHOCK INTAKE & DUCT
FOREPLANES
FRONT FUSELAGE
SENSORS & PROBES
CANARD CONFIGURATIONS
WITH CHIN INTAKES AFT OF
FOREPLANES
GROUND
OPERATIONS
WING
FORE / TAIL PLANE
FIN
FUSELAGE
CREW
EQUIPMENT
STORES & PYLONS
SENSORS & PROBES
ALL TYPES BUT WORSE FOR
CARRIER-BORNE & VSTOL
Any extreme action that can be
achieved by the pilot
BIRDSTRIKE NOSE CONE
COCKPIT / TRANSPARENCY
FOREPLANE
WING LEADING EDGE
INLET FACE
Plus any other forward facing sections
of the airframe
ALL TYPES
Other hazards include airborne and
ground debris
3.3.7 Acknowledgements
Thanks are due for the assistance of Mr. S Samarasekera,
BAE SYSTEMS Aerodynamic Technology , and to Mr.
C Bingham, BAE SYSTEMS Structural Technology.
3.3.8 References
1. PAPER PRESENTED AT RTO CONFERENCE
OTTAWA OCT 1999
NASA LANGLEY RESEARCH CENTERS
Contributions to international active buffet
alleviation programs,
R. W. MOSES, OCTOBER 1999
62
2. AGARD-R-815
THE IMPACT OF DYNAMIC LOADS ON THE
DESIGN OF MILITARY AIRCRAFT
Papers presented at 83
rd
Meeting of the AGARD
Structures and Materials Panel, held in Florence,
Italy, 4-5 September 1996 Published February 1997
3. REVIEW OF HAMMERSHOCK PRESSURES IN
AIRCRAFT INLETS
L C YOUNG and W D BEAULIEU
Rockwell International, Los Angles, California
JANUARY 1975
4. AGARD-R-822
Numerical Unsteady Aerodynamic and Aeroelastic
Simulation
Papers presented at Workshop in Aalborg, Denmark,
OCTOBER 1997 Published March 1998
5. AGARD-CP-483 paper 11
PREDICTIONS OF F-111 TACT AIRCRAFT
BUFFET RESPONSE
AM CUNNINGHAM jr., CF COEAPRIL 1990
6. AGARD-R-815 paper 9
DYNAMIC LOADING CONSIDERATIONS IN
DESIGN OF MODERN COMBAT AIRCRAFT
R CHAPMAN SEPTEMBER 1996
7. AGARD-CP-483 paper 2
A UNIFIED APPROACH TO BUFFET
RESPONSE OF FIGHTER AIRCRAFT
EMPENNAGE
MA FERMAN et al APRIL 1990
8. AIAA paper 91-1049
SOME BUFFET RESPONSE
CHARACTERISTICS OF A TWIN-VERTICAL-
TAIL CONFIGURATION
SW MOSS et al APRIL 1991
9. AIAA paper 92-2127
BUFFET LOAD MEASUREMENTS ON AN F/A-
18 VERTICAL FIN AT HIGH-ANGLE-OF-
ATTACK
BHK LEE, FC TANG JANUARY 1992
10. AGARD-R-815 paper 6
A COMPARISON OF PRESSURE
MEASUREMENTS BETWEEN A FULL-SCALE
AND 1/6 SCALE F/A-18 TWIN TAIL DURING
BUFFET
RW MOSES, E PENDLETON
SEPTEMBER 1996
11. BATH UNIVERSITY, UK Ph.D. Thesis
AN INVESTIGATION OF BUFFET OVER LOW
OBSERVABLE PLANFORMS
M I WOODS 1999
3.4 Managing the Technical Risk Dynamic
Loads in-flight Monitoring
The principle adopted throughout design and clearance of
combat aircraft with respect to dynamic loads is one of
caution, due to the known deficiencies in prediction
techniques. Each design could be over-engineered and
every clearance might be unduly restrictive if the
approximations remain un-quantified. To try to minimize
this risk, dynamic loading predictions are validated
against flight test measurements during envelope
expansion flying within the development phase of the
project.
The flight test envelope expansion process for modern
combat aircraft is a rapid one. To be able to keep pace
with this programme whilst ensuring that in-flight
dynamic loads are on the safe side of predictions, a high
level of visibility of aircraft response amplitudes and
trends is required. In addition, for really rapid turn-
around and test-conduct these data need to be presented
to the monitoring engineer in real time. In this way,
should response trends appear to be worse or response
amplitudes greater than predictions, the testing can be
halted, or modified, before safety is compromised.
Further, due to the data visibility, in-depth evaluation of
any discrepancies can then be carried out post-flight more
effectively.
Real-time unsteady response monitoring is achieved at
BAE Szstems, Warton, via the 'Dynamic Loads
Monitoring System'. The low cost system described here,
commissioned at BAE Systems, Warton, has been used
for the EF2000 Project. It is currently undergoing
modernization.
3.4.1 Dynamic Loads Monitoring System
The Dynamic Loads Monitoring System comprises a
series of pen recorders which display up to 24 real-time
acceleration time-histories for various defined locations
on the aircraft. Figure 1 shows a typical instrumentation
layout for vibration monitoring on a military aircraft
(EAP). In addition, a VAX-based, in-house developed
software package displays the following in real-time:
fin acceleration/dynamic pressure at a defined
fin location vs. incidence angle. These data are
compared with a predicted fin buffet trend
which takes into account, if required, airbrake
operation;
fin acceleration at a defined location vs.
incidence angle. These data can be compared
with a user-defined maximum allowable
acceleration;
wing accelerations for up to 3 defined wing
locations. These data are compared with user-
defined maximum allowable accelerations;
wing acceleration/dynamic pressure at a
defined wing location vs. incidence angle.
These data are compared with a predicted wing
buffeting trend.
A typical example of the software output is shown in
figure 2.
It is worth noting at this stage that airframe loads are
monitored, by implication, via acceleration levels i.e. it is
assumed that, if unsteady acceleration predictions are
consistent with measurements, then the airframe dynamic
loads will also match predictions. Two outputs are
therefore required from the load prediction models
mentioned earlier. The first, for design and clearance
purposes, is actual loading information. The second, for
loads monitoring purposes, is acceleration response data.
63
Strain-gauges could be used to measure load 'directly'.
There are, however, a number of problems associated
with their use, namely:
suitable calibrations being available to convert
gauge signal to load;
reliability of the gauges and the signals that
they produce;
strain gauge signals vary with temperature;
the gauge is measuring structural load in a
highly localized area, making prediction more
difficult to do accurately. Measured
accelerations give a more global picture of
structural response.
3.4.2 Dynamic Loading Phenomena
Monitored
In an ideal world, the dynamic loads engineer would be
able to monitor all regions of an aircraft for all types of
unsteady phenomenon. This would, of course, bring with
it the problem of how to display such a volume of data in
a usable form. Unfortunately (or fortunately), there is a
limit to the amount of instrumentation which can be fitted
to a given test aircraft. Priorities must be decided as to
which dynamic loading effects are to be monitored, but
never to the detriment of flight safety. This decision may
be made easier if loading predictions for a given effect
are small compared to available structural strength and
can therefore be safely disregarded.
The monitoring system at Warton is used to assess the
dynamic response induced by:
gust loading and flutter test induced dynamic
loads via acceleration time-histories displayed
on the pen recorders;
fin and wing buffet loads via acceleration
amplitudes and trends with incidence angle,
displayed using the VAX-based monitoring
software.
3.4.3 Dynamic Loads Monitoring System
Implementation
Figure 3 shows how the Dynamic Loads Monitoring
System is implemented at Warton.
Accelerometer data from various locations on the
airframe is transmitted to the Monitoring System (via a
Ground Station) at a rate of 512 samples per second.
Using the Nyquist Theorem, this allows the monitoring
engineer to observe vibration response having a
maximum theoretical frequency of 256Hz. This
frequency range is sufficient for the dynamic phenomena
being monitored, as defined earlier. In addition, a
selection of aircraft data (Mach no., incidence angle,
dynamic pressure and time) are transmitted to the system
at 32 samples per second.
The (digital) accelerometer data to be displayed using the
pen recorders is converted to an analogue signal and is
plotted throughout the flight. This provides a useful data
quality check in addition to displaying response
amplitudes. The pens used for this have a transfer
function such that signals with frequencies up to around
80Hz are not attenuated.
The VAX-based software component of the monitoring
system is only used for certain flight test points - those
where significant wing and/or fin buffet is likely to occur
e.g. wind-up turn maneuvers. The fin and wing buffet
accelerometer data are conditioned as follows:
high and low-pass filtered to remove any DC
signal component and to include only the
response frequencies of interest. This is
limited to only those frequencies associated
with the first few fundamental aircraft vibration
modes (the modes most likely to cause
structural damage in the case of buffet
monitoring).
data 'drop-outs' are checked for and any data
'spikes' are suppressed.
Buffet analysis is initiated and terminated by the
monitoring engineer. Conditioned data is captured by the
system over one second and the requisite analysis
performed to obtain zero-to-peak acceleration levels and
zero-to-peak acceleration levels normalized by dynamic
pressure. These data are then plotted to the monitor
screen (vs. incidence angle where applicable) using the
lower rate aircraft data. This process is repeated until the
system is commanded to stop. The plot presented to the
user is therefore continually updated as a given maneuver
progresses. This process is summarized in figure 4.
The data acquired during monitoring are saved to disk for
post-flight analysis, if required.
Figure 5 shows an example of the wing buffet data
available to a monitoring engineer during a wind-up-turn
(WUT) maneuver. The acceleration time-history for a
wing parameter is shown (W3). It can be seen that as the
WUT progresses, the vibration amplitude increases and
then attenuates as the turn is completed and straight and
level flight resumed. Peak acceleration amplitudes for
this and two other accelerometers (W1, W2 and W3) are
plotted for comparison with user-defined maximum
allowable vibration levels at 1 second intervals. In
addition, the trend of peak g/dynamic pressure is plotted
against incidence angle for comparison with the predicted
trend.
Figure 5 shows that whilst an acceleration time-history is
useful as a data quality check, the software based
monitoring system provides a quick way of verifying that
the dynamic loading on the aircraft is within prescribed
limits. Simplification of the loads monitoring task is
welcome in the high-pressure flight test environment.
Figure 5 shows that, for this test point at least:
wing buffet trend predictions are well matched
by flight measurements and
amplitudes of vibration at the wing
accelerometer locations are well within
allowable limits.
As such, with respect to buffeting response, this test has
been flown safely. It should be noted that these results
are for a single test point. To form any sensible
conclusions about the predictive techniques used, a more
extensive survey of results would have to be performed.
64
FIGURE 1 - Typical Accelerometer Layout on Military Aircraft (EAP)
Accelerometer Locations
65
FIGURE 2 - Monitoring System Exampl e Data Plots
P E A K G ( % )
P E A K G ( % ) P E A K G / D Y N A M I C P R E S S U R E
P E A K G / D Y N A M I C P R E S S U R E
W 1
W 2
W 3
F
I
N

B
U
F
F
E
T
W
I
N
G

B
U
F
F
E
T
P
r
e
d
i
c
t
e
d

T
r
e
n
d
P
r
e
d
i
c
t
e
d

T
r
e
n
d
M

0
.
5
M

0
.
9
N
C
I
D
E
N
C
E

A
N
G
L
E

(
d
e
g
s
)
N
C
I
D
E
N
C
E

A
N
G
L
E

(
d
e
g
s
)
N
C
I
D
E
N
C
E

A
N
G
L
E

(
d
e
g
s
)
66
FIGURE 3 - Dynamic Loads Monitoring System General Layout
D
a
t
a

T
r
a
n
s
m
i
s
s
i
o
n
F
l
i
g
h
t

T
e
s
t
G
r
o
u
n
d
S
t
a
t
i
o
n
A
C
C
E
L
E
R
O
M
E
T
E
R
D
A
T
A

(
5
1
2

S
/
S
)
A
c
c
e
l
e
r
o
m
e
t
e
r
s
F
L
I
G
H
T

D
A
T
A
(
3
2

S
/
S
)
I
n
c
i
d
e
n
c
e
M
a
c
h

N
o
.
D
y
n
a
m
i
c

P
r
e
s
s
u
r
e
T
i
m
e
P
E
N

R
E
C
O
R
D
E
R
S
S
I
G
N
A
L

C
O
N
D
I
T
I
O
N
I
N
G
F
i
l
t
e
r
i
n
g
S
p
i
k
e

S
u
p
p
r
e
s
s
i
o
n
D
r
o
p
-
O
u
t

C
h
e
c
k
s
D
I
G
I
T
A
L

T
O

A
N
A
L
O
G
U
E
C
O
N
V
E
R
S
I
O
N
D
Y
N
A
M
I
C

L
O
A
D
S

M
O
N
I
T
O
R
I
N
G

S
Y
S
T
E
M
V
A
X
S
A
V
E
67
FIGURE 4 - Calculation of Trends With Aircraft Incidence
A
N
A
L
Y
S
I
S
A
N
A
L
Y
S
I
S
A
N
A
L
Y
S
I
S
A
N
A
L
Y
S
I
S
A
N
A
L
Y
S
I
S
A
N
A
L
Y
S
I
S
A
N
A
L
Y
S
I
S
A
N
A
L
Y
S
I
S
A
N
A
L
Y
S
I
S
cel. (g)
cidence,
(degs)
ynamic
ressure,
(kPa)
USER DEFINED
ACQUISITION
START
TIME
g
peak
g
peak
g
peak

mean

mean

mean
q
mean
q
mean
q
mean
PLOT
g
peak
/q
mean
vs
mean
&
g
peak
vs
mean
&
g
peak
vs wing transducer
UPDATE PLOT UPDATE PLOT
1 Second
68
FIGURE 5 - Monitoring System Example Wing Buffeting Output
W
3

T
I
M
E

H
I
S
T
O
R
Y
P
r
e
d
i
c
t
e
d

T
r
e
n
d
M

0
.
5
M

0
.
9
G P E A K ( % )
G P E A K / D Y N A M I C P R E S S U R E
I
N
C
I
D
E
N
C
E

A
N
G
L
E

(
d
e
g
s
)
W 1
W 2
W 3
A
l
l
o
w
a
b
l
e

V
i
b
r
a
t
i
o
n

A
m
p
l
i
t
u
d
e

L
i
m
i
t
s
3.5 Airframe Certification Against Birdstrike
Threats
The phenomena of birdstrikes requires serious
assessment during the design stages of an aircraft. Over
the last decade there has been an increase in fatal
accidents due to birdstrikes on military aircraft.
Furthermore, it is the single greatest cause of military
aircraft loss in peace time.
To certify the airframe against birdstrikes, resistance to
representative impulse loads acting on all leading edge
and forward facing sections of the airframe must be
considered early in the design phase. The design work
would involve predictions of stress levels associated with
such loads in both the skin and substructure of the
frontal airframe region. To prevent stress levels
exceeding the allowable limit, high strain rate
performance, yield strength and fracture toughness may
be critical factors in determining material selection.
Furthermore, past testing has revealed that structural
components with sharp leading edges (i.e. leading edge
radius less than bird diameter) leads to a significant
increase in the impact velocity required to cause
structural damage, due to higher local stiffness levels
inherent in smaller radii. Therefore, design specifications
of leading edges for forward facing regions of the
airframe can be influenced by birdstrike phenomena, in
addition to aerodynamic, structural and manufacturing
aspects.
3.5.1 Certification via Empirical Testing
Chapter 209 of Ref. 1 specifies the minimum
requirements for the resistance of airframes to damage
caused by birdstrike ;
A 1kg bird with an impact velocity of 480 knots
must not penetrate the structure.
69
A 1kg bird with an impact velocity of 366 knots
must not cause structural damage.
The latter specification reflects the need to reduce the
cost of repair after lower kinematics energy impacts.
Currently, meeting this specification is an expensive and
time consuming procedure, primarily due to model
manufacture and test set up costs.
The standard approach is to fire real (dead) birds using
compressed air in a gas gun. The birds are fired at
varying projectile velocities (up to high subsonic Mach
No.s) onto the frontal area of the airframe, i.e. nose
cone, transparency, intake lips, foreplane, wing leading
edges etc. Testing considers birdstrikes head on to the
airframe and angles up to 15 - 17 degree azimuth from
the nose direction. Maximum deflections of the structure
are recorded and the impacted structure is inspected for
damage and evidence of penetration. This data may be
supported by strain gauge information, high speed
photography and deflection time history data from laser
measuring devices. Due to the difficulties involved in
firing real birds, the inherent variability in the bird
structure, the difficulty in controlling the centre of gravity
location and the bird orientation, tests are notoriously
prone to high levels of variability.
Empirical design rules are available for metallic
structures however equivalent methods are not available
for composites making the potential role of analysis more
important. A single test that fails the structure may not
provide much information for a successful redesign to be
produced, particularly in the light of other design
considerations that may apply.
Current Developments
In an attempt to alleviate costs involved with standard
birdstrike testing, one approach that has been accepted in
the civil aerospace industry is to certify aircraft against
birdstrikes using generic analysis (Ref. 2). However, it
may be some time yet before military aircraft would be
allowed to be certified in this way.
The idea behind the generic analysis approach is that if
you have designed and tested a similar component before,
and if the analytical method has proven accuracy,
clearance of a new generic component can be achieved
by analysis alone. Generic analysis requires
comprehensive understanding of mechanical properties
and failure modes of the airframe structure and bird
behaviour under impact. Bird impacts above a certain
velocity threshold has been shown to be essentially
fluidic. The modelling of an event which incorporates
both fluidic and structural behaviour, with strong
interaction, presents significant challenges to the
available codes and analysis techniques.
Coupled Euler-Lagrange and smooth particle
hydrodynamic codes are now being developed that will
significantly improve the modelling capability in the
future. Current analytical techniques attempt to represent
the bird behaviour in the best possible manor in a
Lagrangian approach.
The failure behaviour of structures under high velocity
impact and the representation of these events in the codes
is also subject to on going research and development.
This is particularly significant in the area of composite
materials where there are many complex failure modes
and particular problems in including these effects into the
codes.
To address these issues and improve the analytical
capability several working groups and research activities
have been set up in industry. These include programs that
have established bird biometrics and flocking behaviour,
investigated the use of more consistent artificial birds,
investigated the high rate failure behaviour of composites
and assessed the on-going developments in the available
codes.
The results of one (FE based) birdstrike prediction tool is
shown in Figure 6 below. The figure shows a strain map
of a leading edge after impact and allows direct
comparisons with strains measured from experiment.
Upon extensive validation of birdstrike FE prediction
tools, some form of certification of airframes against
birdstrikes by analysis could become feasible, although it
is envisaged that empirical testing will never be fully
eradicated from a combat aircrafts developmental
programme.
3.5.2 References
1. DEFENCE STANDARD 00-970 MINISTRY OF
DEFENCE
DESIGN AND AIRWORTHINESS
REQUIREMENTS FOR SERVICE AIRCRAFT
VOLUME 1 AEROPLANES, BOOK 1
2. S351 IMechE SEMINAR PAPER 1
DEVELOPMENT OF A BIRDSTRIKE
CLEARANCE PHILOSOPHY
C H EDGE
Published in Foreign Object Impact and Energy
Absorbing Structure MARCH 1998
70
Figure 6: Birdstrike FE-Prediction
4 Gust loads
4.1 Introduction
Aircraft are often subjected to abrupt movements of air in
the form of turbulence or gusts. These gusts can impose
considerable loads on aircraft. Gusts may come from all
directions. Vertical gusts load the wing, fuselage and
horizontal tail. In the case of horizontal gusts we
distinguish lateral or side gusts, loading the fuselage,
vertical tail and pylons and longitudinal or head-on
gusts which may cause important loads on flap structure.
For transport type aircraft, gust load cases are the most
critical for strength design, and gust loads are the main
fatigue loading source for the major part of the structure.
Combat type aircraft structures are generally manoeuvre
load critical, but for specific parts of the structure like
thin outer wing sections and pylons, gusts may determine
critical design load cases
1
. Since the recognition that
turbulence produced significant loads (around 1915) gust
design criteria have been formulated, which have evolved
over the years and are still under development
2,3
.
All major current Airworthiness Codes include two sets
of gust criteria, based on a Discrete Gust concept and a
Continuous Gust concept. In the following, the main
aspects of these two concepts will be briefly explained.
4.1.1 Discrete Gusts
The basic loading mechanism of gusts is schematically
illustrated in fig 4.1. An aircraft flying with speed V
entering an upward gust with velocity w experiences a
sudden change in angle of attack =w/V. This gives
rise to an additional air load
V
w
SC V
2
1
L
L
2

=
It will be clear, however, that the abrupt or sharp-edged
gust indicated in figure 4.1 is physically impossible; it
implies an instantaneous change in lift and a real gust
must have some distance over which its effect builds up.
Additionally, due to so-called aerodynamic inertia, a
sudden change in angle of attack does not immediately
result in a proportional change in lift. Hence, the load felt
by the structure is modified by this effect. The resulting
load depends upon the size and shape of the gust and the
response characteristics of the aircraft. Different
Discrete Gust shapes have been assumed in gust
criteria, ranging from the simple sharp-edged shape
shown in figure 4.1 (in the early twenties), through the
ramp type gust used in e.g. the former BCAR
Requirements to the 1-cos gust shape included in
almost all current airworthiness codes.
71
Essentially, the Discrete Gust Criterion consists of a
design gust of specified shape and magnitude U
ds
(which is a function of altitude). The design value Y
des
of
any load quantity y is to be found by calculating the time
response y(t) to the gust, and taking the maximum of y(t)
as Y
des
. For many years, the main Airworthiness Codes
included simplifying assumptions with regard to the
length of the gust ( e.g. a (1-cos)-gust of 25 wing cords)
and allowed the assumption of an aircraft response in
plunge only (in the absence of a more rational
investigation), resulting in very simple gust-response
expressions as given e.g. by the well known Pratt
Formula
3
.
With the growing size and increasing flexibility of
aircraft these assumptions became more and more
unacceptable. Hence, the major Airworthiness Codes
currently demand for a full dynamic response calculation,
including all rigid and all relevant elastic modes. As the
length of the gust has a direct effect on the structural
response, a range of gust lengths has to be considered.
The one giving the highest design load (the Tuned
Discrete Gust) must be assumed, up to a defined level of
severity e.g. the minimum gust distance is specified.
4.1.2 Continuous Gusts
The discrete gust concept assumes an atmosphere where
separate and independent gust bumps occur that may
hit the aircraft. Measurements in gusty conditions,
however, revealed a pattern more resembling a process of
continuous turbulence. This notion led in the early
sixties to the development of a completely new gust
concept and a set of additional Design Criteria, known as
the Continuous Turbulence Concept and the PSD
(Power Spectral Density) Gust Design Criteria.
In this concept, the loading action is described as a
continuous process of random turbulence. Over shorter
periods of time this process may be considered as
stationary with Gaussian properties and standard
deviation
w
. In the longer term, the standard deviation
or gust intensity is not a constant, but varies randomly
with a given probability function. The turbulence is
characterized by the von Karman type Power Spectral
Density function, describing how the energy in the
process is distributed with frequency.
On the basis of this turbulence concept, two design
methods were developed referred to as the Mission
Analysis and the Design Envelope Concepts
5
. The
Mission Analysis Concept, which is of a purely
statistical nature, has the virtue of elegance. It is,
however, difficult to apply and may lead to
unconservative predictions if the actual Mission Profile
of an aircraft changes and starts to deviate from design
assumptions. Hence, the criterion is seldom applied and it
is expected that in the near future it will be deleted from
the Airworthiness Codes.
The Design Envelope criterion shows a resemblance to
the Discrete Gust Criterion in that it also specifies a
design gust strength U

as a function of altitude the

design value Y
des
of any load quantity y is found from

U * A Y r
des
The response parameter r A , which is actually the
ratio of the standard deviations of the load output y and
gust input w in stationary Gaussian turbulence, may be
considered as defining an average weighted response;
r A is calculated by integrating the product of load
transfer function squared and the turbulence PSD
function over all gust frequencies. Thus, r A defines
essentially an average response, taking into account for
which frequencies the load is sensitive (as defined by the
transfer function) and also which gust frequencies (or
gust lengths) occur in the atmosphere.
Comparing now the PSD- gust criterion and the Discrete
gust criterion, we notice the difference and the reason
why both criteria are included in our design procedures.
The PSD criterion is based on a rational and consistent
model of the atmospheric turbulence; it defines design
loads that are based on an average response, considering
all possible gust lengths that prevail in random
turbulence. The Discrete Gust Criterion is typically a
worst case criterion; the highest load resulting from a
discrete bump with most adverse length must be taken.
The Discrete Gust cases are included and maintained in
airworthiness codes to safeguard against sudden more or
less stand alone gust outbursts that have been observed
to occur in practice.
4.1.3 Gust Load Requirements
Gust load requirements have been, and are subject to, a
process of continuous change due to the experience
gained from previous aircraft, changes in aircraft design
philosophy and advances in analysis techniques. Section
4.2 gives an overview of the gust requirements in the
principal civil and military requirements prevailing today.
The military requirements tend to lag behind compared to
FAR/JAR 25, due to a lack of available flight data as well
as the lower criticality of gust loads for military aircraft.
In FAR 25 and JAR 25, major changes have been
included over the last few years with regard to the
discrete gust cases and a major change of the continuous
gust criteria is in preparation. A relevant part of the
associated NPRM (Notice on Proposed Rule Making) is
included in Paragraph 4.2.
These developments have prepared by the ARAC Loads
and Dynamic Handling Working Group, supported by the
Committee of International Gust Specialists.
Airworthiness Requirements tend to be put in rather
general legal terms, which may be subject to different
interpretation. Additional documents, describing
acceptable means and methods to comply with the
requirements may be very helpful. Such information
may be contained in ACJs (Acceptable means of
Compliance to JAR) in the case of JAR requirements, or
in Advisory Circulars in the case of FAR requirements.
Traditionally, the calculation of aircraft response has
been made assuming linearity. With the advent of
nonlinear active control systems, aircraft are becoming
increasingly nonlinear and the assumption of linearity is
becoming more and more unacceptable for accurate load
prediction. The calculation of the response to a discrete
gust for a nonlinear aircraft may be time-consuming but
offers no fundamental problem. Three deterministic type
methods are considered here: Matched Filter Theory, the
72
Noback (or IDPSD) method and the Spectral Gust
(Brink-Spalink ) method.
The existing PSD gust design criteria, however, are
fundamentally based on linear response behaviour.
Current Airworthiness Codes do not contain explicit rules
how to determine PSD-gust loads for non-linear aircraft,
but the NPRM presented in paragraph 4.2 foresees in this
shortcoming. In case of significant non-linearities, one
approach towards determining the PSD design loads is to
calculate the aircraft response in the time domain of the
aircraft to a patch of stationary random turbulence with
an rms. value equal to 0.4 times the design gust velocity
U

. This procedure is known as the Stochastic

Simulation method, is physically well founded,
straightforward and relatively easy to apply but very
computer time consuming and hence expensive. The
alternative Probability Exceedence Criterion (PEC)
method is also considered. A further approach is the
Statistical Discrete Gust method, which attempts to
combine both discrete and stochastic methodologies. Full
details about the methods can be found in Appendix
A4.1. There is a need to assess, validate and compare
these methods before they can be accepted for
Certification purposes.
Section 4.3 presents a comparison of the above methods
for design load calculations using various aircraft models
with different nonlinearities. Two different institutes
carried out these calculations and comparative results are
given. Concluding remarks are presented in section 4.4.
4.2 Overview of Gust Requirements
4.2.1 Draft NPRM on Continuous
Turbulence.
The Discrete Gust Criteria in FAR25 and JAR25 have
been changed a few years ago, but the Continuous Gust
Requirements in these codes have not been changed since
the late sixties.
A Draft NPRM (Notice on Proposed Rule Making) has
been prepared recently, proposing changes in the FAR25.
It is expected that these proposed changes will also be
adopted in the JAR 25 Code.
The proposed requirement includes a revision to the gust
intensity model used in the design envelope method to
continuous turbulence on the basis of more recent
statistical data (including CAADRP data). The mission
analysis method will be eliminated and a new
requirement included for considering combined vertical
and lateral turbulence. Provisions for treating non-
linearities will also be included.
A summary of the most relevant changes that are
proposed for paragraph 25.341 are:
(b) Continuous Turbulence criteria:
(3) The limit turbulence intensities U, in feet
per second true airspeed required for compliance with
paragraph are
(i) At speed from VB to VC:
U

=U
ref
F
g
Where
U
ref
is the turbulence intensity that varies
linearly with the altitude from 90 fps (TAS at
sea level to 79 fps (TAS) at 24000 feet and is
then constant at 79 fps (TAS) up to an altitude
of 50000 feet.
F
g
is the flight profile alleviation factor defined
in paragraph (a)(6) of this section;
(ii) At speed V
D
: U

is equal to the
values obtained under subparagraph
(3)(i) of this paragraph.
(iii) At speeds between V
C
and V
D
: U

is
equal to a value obtained by linear
interpolation.
(iv) At all speeds both positive and
negative continuous turbulence must
be considered.
(4) When an automatic system affecting the
dynamic response of the airplane is
included in the analysis, the effects of
system non-linearities on loads must be
taken into account in a realistic or
conservative manner.
(5) If necessary for the assessment of loads
on airplanes with significant non-
linearities, it must be assumed that the
turbulence field has a root-mean square
velocity equal to 0.4 times the U

values
specified in subparagraph (3). The value
of limit load is that load with the same
probability of exceedence in the
turbulence field as a velocity of U
.
(6) The resultant combined stresses from
both the vertical and lateral components
of turbulence must be considered when
significant. The stresses must be
determined on the assumption that the
vertical and lateral components are
uncorrelated.
4.3 Comparison of Methods to calculated
Continuous Turbulence Design Loads for
Non-Linear Aircraft
This section presents results of comparative studies to
evaluate methods for the calculation of design loads. The
simulations were carried out by the National Aerospace
Laboratory NLR N and the University of Manchester
UK, using the same aircraft models. A number of
different methods were considered:
Stochastic Methods
Stochastic Simulation (SS)
Probability of Exceedence Criterion (PEC)
Power Spectral Density (PSD) [only for the linear
cases]
Deterministic Methods
Matched Filter Based method (MFB), both 1-
dimensional and Multidimensional
Indirect Deterministic Power Spectral Density
Method (IDPSD)
Spectral Gust procedure (SG)
Stochastic-Deterministic Methods
Statistical Discrete Gust (SDG)
73
A brief description of these methods is given in Appendix
A4.1.
The following nonlinear aircraft models were used:
- Noback model: 2 DOF large transport aircraft with
load alleviation through ailerons.
- F100 model: medium-sized transport with "Fokker-
100-like" characteristics with load alleviation
through ailerons.
- A310 model: an A310 model with load alleviation
through ailerons and spoilers.
A description of these models is given in Appendix A4.2.
Nonlinearity is introduced in these models by limits on
the control surface deflections. The A310 model control
surfaces can only deflect upward (max. 10 deg.) in the
nonlinear version, so that a non-symmetrical nonlinearity
is introduced. Analysis could be performed using either
the linear or non-linear versions of these models.
4.3.1 Analyses made by NLR
The NLR investigation
4
compared the three Deterministic
methods with the Stochastic Simulation methods and the
PSD technique for the linear cases. For linear aircraft
models, these Deterministic PSD methods and Stochastic
Simulation result in design and correlated load values y
d
and z
c
that are equal to the "standard" PSD loads:
.
U
A
=
z

U
A
= y
z yz
c
y d


For nonlinear aircraft models, the standard PSD method
cannot be applied, because the model transfer functions
are then dependent on the input signal. The Stochastic
Simulation method has been proposed for the definition
of design and correlated loads in nonlinear cases. This
method is based on the probability of exceedence of load
levels. The Deterministic methods aim to comply with
this Stochastic Simulation procedure in nonlinear
calculations.
By showing results of calculations for these three aircraft
models it was demonstrated that the Deterministic and
the Stochastic Simulation procedures effectively lead to
correct PSD loads in linear cases. The results for three
nonlinear aircraft models obtained with the Deterministic
methods are presented, and the degree of compliance of
the Deterministic methods with Stochastic Simulation
was investigated.
In Appendix A4.1 it is explained that the Deterministic
methods follow a more or less similar scheme. An
essential part in the procedures is the so-called gust filter.
The Power Spectral Density of the gust filter response to
a pulse input should have the von Karman power
spectrum shape. The impulse response power spectrum
can be calculated directly from the frequency-domain
representation of the gust filter G(jf):
( )
( ) ( )
T
jf G jf G
f

=
where T = length of impulse response.
The gust filter impulse response for the IDPSD filter
gives by definition exactly the von Karman Spectrum.
Comparing the original MFB gust filter ("NASA"), and a
new MFB gust filter that has been taken from Hoblit
5
, it
appears that the Hoblit filter clearly approaches the von
Karman PSD better than the original NASA filter. The
Hoblit gust filter has therefore been implemented in the
present MFB procedure, which resulted in correct PSD
loads in linear cases, contrary to MFB with the original
NASA gust filter, where slight deviations from AU

were
found.
The bar-charts in figures 4.2 - 4.7 show the results of the
calculations for the three aircraft models and five
calculation methods. The notation in the axis labels of
these figures is as follows:
y,des = design load value of load quantity y.
y,cor z = correlated value of y if z has its
design value.
nonlin = closed loop system, nonlinear
(limited) load alleviation.
nolim = closed loop system, linear
(unlimited) load alleviation.
nocon = open loop system (linear).
Stoch. Simul. = Stochastic Simulation result.
PSD = standard PSD result.
POS = "positive" design load case (A310
model only).
NEG = "negative" design load case (A310
model only).
Note that correlated load values in some cases are given
with opposite sign, indicated by a minus sign in the
legend. The results for the linear and nonlinear versions
of the A310 model are given in separate figures, because
there is a difference between "positive" and "negative"
nonlinear design load cases, due to the fact that ailerons
and spoilers can only deflect upward in the nonlinear
version of this model.
These bar charts demonstrate that the three Deterministic
methods comply with the standard PSD results in linear
cases, so it may be concluded that all Deterministic
procedures lead to correct results for linear aircraft
models. Figure 4.2 for the linear A310 model shows
standard PSD results and Deterministic PSD results
together with Stochastic Simulation results. It can be seen
that the Stochastic Simulation procedure gives design
loads close to the standard PSD values, and correlated
loads may deviate a few percent (of the design load
value) from the theoretical value, see for instance the
correlated bending for the uncontrolled A310 model.
In nonlinear conditions, where controller actions are
limited, the Stochastic and Deterministic methods lead to
different results. MFB and IDPSD do not differ much,
but the correlated load values are different in some cases.
A second optimization loop could have been added to
MFB/IDPSD, calculating outputs at e.g. four more k/K
eq
values around the optimum found, and find a higher
maximum output with somewhat different correlated load
values. An even more rigorous search routine, the "multi-
dimensional search", might also be applied. As it is
believed, on the basis of NASA investigations, that such
a routine would change the design conditions by a very
small amount in respect to the one-dimensional search,
such calculations were not performed.
74
MFB and IDPSD both approach the Stochastic
Simulation results reasonably in figure 4.3; only the
correlated value of n for the nonlinear F100 model is
really very incorrect (wrong sign) for both methods, see
figure 4.4. The corresponding MFB/IDPSD design levels
of the bending moment in figure 4.5 differ more than
10 % from the Stochastic Simulation value. The SG
procedure design loads and correlated loads can both
deviate appreciably from Stochastic Simulation results.
Similar findings were obtained for the Noback model,
figures 4.6-4.7, where the major differences occur in the
correlated y values.
The ailerons and spoilers of the A310 model can only
deflect upward in the nonlinear version, so that different
gust design loads will occur in positive and negative
directions. In the IDPSD and MFB procedures, negative
gust cases are created by reversing the sign of the gust
inputs to the "first system". In the SG procedure the sign
of a design load is determined, by calculating the sign of:
dt y y
0

where y is the load quantity response to an SG input.

It can be seen in figure 4.3 that the positive and negative
design load cases of wing bending do not differ
significantly, but the negative torsion design load is
considerably lower than the positive design load in the
results of Stochastic Simulation, MFB, and IDPSD. It is a
good point for MFB and IDPSD that they appear to
represent this effect in the same way as the Stochastic
Simulation method.
With regard to the required computational times the
following observations could be made. The SG method is
very fast, because only four time responses are
calculated. The IDPSD method takes some more
calculation time than MFB, because the "first system"
response in IDPSD is twice as long as in MFB. Stochastic
Simulation takes much more time than the other methods
(14 times the MFB time), mainly due to the counting
procedures for finding design levels and correlated loads.
The following conclusions can be drawn from this
comparison of Deterministic methods with the Stochastic
Simulation and "standard" PSD methods:
- With the Hoblit gust filter, MFB is equivalent to
IDPSD and "standard" PSD in linear cases.
- The results of MFB and IDPSD are reasonably
similar in nonlinear cases; correlated loads may
deviate somewhat.
- MFB and IDPSD reasonably approach Stochastic
Simulation results in nonlinear cases, but this is not
enough for design load calculations.
- The SG method deviates significantly from the other
methods in nonlinear cases.
- Stochastic Simulation takes much more calculation
time than the Deterministic methods.
4.3.2 Analyses made by the University of
Manchester
The following methods were investigated at the
University of Manchester:
- IDPSD: Indirect Deterministic Power Spectral
Density
- MFB 1-D: Matched Filter Based 1-Dimensional
- MFB Multi-D: Matched Filter Based Multi-
Dimensional
- PEC: Probability of Exceedence Criteria
- SS: Stochastic Simulation
- SDG: Statistical Discrete Gust
The description of the methods can be found in Appendix
4.1. The methods were applied to the simple 2-dof and
A310 aircraft. Since the absolutely correct design load
cannot be obtained for a nonlinear system, one of the
methods was to be used as a benchmark. In this case, the
benchmark was chosen to be the Matched Filter Based 1-
Dimensional search method. This choice was dictated by
the relative simplicity of the method and by the fact that
it is less computationally expensive than the other
methods. However, the term "benchmark" does not imply
that the design loads predicted by the MDB 1-D method
are taken to be the best estimates.
The graphical comparisons between the methods
presented in this section are based on the following
figures (unless otherwise stated).
- Figures 4.8 and 4.9 show a direct comparison of
maximum and correlated loads obtained by the
methods for the Noback aircraft model.
- Figures 4.10 and 4.11 show a direct comparison of
maximum and correlated loads obtained by the
methods for the A310 aircraft model.
- Figures 4.12 and 4.13 Load variation with time and
critical gust shape for Noback aircraft load 2 and
A310 load 3
4.3.2.1 Stochastic Simulation Method
The figures show a very good agreement between results
using the SS method and those from the two deterministic
methods. Figure 4.12 shows the load variation with time
and the critical gust shape for the Noback aircraft as
predicted by the MFB, SS and IDPSD methods. It can be
seen that, even thought there is some differences between
the three gust shapes, the load variations are in very good
agreement with each other. This phenomenon highlights
the main difficulty in predicting gust loads and worst-
case gusts for nonlinear aircraft i.e. that there is not one
single solution.
The good agreement between the two deterministic
methods and the SSB however, heavily depends on the
choice of the value of the turbulence intensity,
g
. The
authors of reference 6 suggest that, in order to compare
the two methods, the value of the turbulence intensity
used with the MFB scheme should be

U
g

where

U is the design gust velocity. For the SSB

method, the suggested value is
3 /

U
g

75
The turbulence intensity used during the course of this
work was
5 . 2 /

U
g

This value was preferred
4
to 3 /

U because it agrees
more closely with the representative,
wr
, value at
normal civil aircraft cruising altitudes.
4.3.2.2 PEC method
The design and correlated loads obtained by the PEC
method are in considerable agreement with those
obtained by the SSB method, which is logical since both
methods are stochastic approaches applied to the same
simulated patches of turbulence.
The comments made in the previous paragraph about
turbulence intensity also apply to the PEC approach.
4.3.2.3 SDG method
The SDG method is the approach that yields loads which
are in least agreement with those obtained from the other
techniques. For the Noback aircraft, the SDG yields the
most conservative design load for load 1 and the least
conservative one for load 2. For the A310, the SDG
estimate for load 3 is in good agreement with those
obtained from the DPSD procedures but, for load 4 the
SDG again provides the least conservative design loads.
This discrepancy is caused by the fact that the SDG
methodology, being based on a search through families of
discrete gusts, is significantly different to the other four
methodologies (see Appendix 4.1).
4.3.2.4 IDPSD method
The agreement between the IDPSD and the MFB 1-D
methods is, generally, very good. For the particular case
of the worst-case gust for Load2 of the Noback aircraft
(figure 4.12), the agreement breaks down to a certain
extent. The figure shows that the gust shape estimated
using the IDPSD lies between the SSB and MFB 1-D
gusts. Nevertheless the resulting maximum loads are still
comparable.
Since both the Noback and MFB 1-D methods are
deterministic methods, estimating worst-case gusts there
is no problem with scaling the turbulence intensity value
in order to get agreement between the two methods.
4.3.2.5 MFB Multi-Dimensional Search
Table 4.1 shows a comparison of results from the 1-
dimensional and the multi-dimensional MFB searches,
obtained from the Noback and A310 models. The table
confirms previous findings
7,8
that the 1-dimensional
search provides a very good estimate of the design load.
The design loads for the Noback model have been
improved upon by the MFB M-D method by up to 6.8%.
However, for the A310 model, the improvement is almost
negligible. The fact that the multi-dimensional search is
much more computationally expensive but only delivers a
small improvement in the final result suggests that the 1-
dimensional search is more suitable, especially in the
case of the gust-load prediction for a full aircraft, where
the design loads need to be predicted at a very large
number of stations over the whole aircraft.
% Improvement
6.8
6.7
0.1
0.2
Load MFB 1-D MFB M-D
Noback Load 1 10.73 m/s
2
11.46 m/s
2
Noback Load 1 6.55 m/s
2
7.02 m/s
2
A310 load 2 2.8242x10
6
lb.ft 2.8261x10
6
lb.ft
A310 load 3 2.3736x10
5
lb.ft 2.3793x10
5
lb.ft
Table 4.1:Comparison of design loads by the MFB M-D
and MFB 1-D methods for the Noback and A310 models
4.3.2.6 Comparative Results
The IDPSD method tends to predict slightly more
conservative results than the MFB 1-D method. In the
case of the Noback model the IDPSD results are closest
to those obtained from the MFB M-D method. Since the
SSB and PEC are stochastic, their design load predictions
change slightly every time the calculations are
performed. Consequently, there is no definitive way of
determining whether these predictions are generally more
or less conservative than the results obtained with the
other two methods.
Another important conclusion is that the design load
predictions of the methods agree more closely with each
other than the correlated load predictions. In reference 4
this phenomenon is also noted. Additionally, Vink
4
shows the cause of the phenomenon to be that the
theoretical standard deviation of the design load will
generally be smaller than the theoretical standard
deviation of the correlated loads.
In many cases the methods predict very different worst-
case gust shapes but quite similar design loads. Table 4.1
shows the worst-case gusts and resulting load variations
calculated from the SSB, MFB and IDPSD methods for
the A310 wing torsion load. It can be clearly seen that
three considerably different worst-case gust shapes yield
very similar load variations and, hence, maximum loads.
Again, this phenomenon is caused by the nonlinearity of
the aircraft under investigation.
Table 4.2 compares the computational expense of the
SSB, MFB 1-D, PEC and IDPSD methods. Neither the
CPU time nor the number of floating point operations
(flops) figures are absolute. CPU time depends on the
computer used, the software installed. The number of
flops performed depends on the programming and on the
routine that counts the flops. Nevertheless there is a clear
pattern to the results in the tables. The least
computationally expensive method is the MFB 1-D and
the most computationally expensive one is the SSB, with
76
the IDPSD and PEC methods lying somewhere in
between. The CPU time and number of flops for the
multi-dimensional MFB and SDG methods are labelled
"variable" in the table since the method relies on a
directed random search. Hence, the duration of the
calculations is different every time the procedure is
applied, but always much longer than the duration of any
of the other methods.
Method CPU time
IDPSD 24.45
MFB 1-D 18.73
MFB M-D Variable*
PEC 100.93
SDG Variable*
SSB 274.85
Table 4.2:Comparison of computational expense of the
methods (applied to the A310 model) * Variable times
are caused by optimization procedures
4.4 Conclusions & Recommendations
This report has provided a brief historical background
and an overview of the current state of the airworthiness
regulations as regards to gust loadings. In the future,
certification regarding the effects of non-linearities on the
gust loading of aircraft will become increasingly
important. A number of the most promising gust load
prediction methods, including both stochastic and
deterministic techniques, have been described and
compared analytically.
The nature of non-linear systems means that the principle
of superposition does not hold and large amount of
computation is required to determine the design gust
loads. Even then, there is no guarantee that a maximum
has been achieved. The computation can be performed
either via a stochastic approach that considers a large
amount of turbulent data, or a deterministic procedure
whereby some type of search is undertaken to find the
maximum loads.
Two comparative studies were carried out using three
different non-linear aircraft models. Gust loads obtained
using the different methods were compared. It was found
that most of the analysis techniques gave similar
estimates, although some variation in results was found
using the version of the Statistical Discrete Gust method
employed for this work, and also the Spectral Gust
method. There is not enough evidence however to
categorically say one method is better, or worse, than the
others. The deterministic methods require less
computation.
There is a requirement for the research community to
develop new analysis methods that are able to predict
design gust loads without resorting to large amounts of
computation. The test cases used in this study should be
employed as benchmark test cases for future comparative
work.
4.5 References
1 Various authors: Loads and Requirements for
Military Aircraft. Papers presented at the 83rd
Meeting of the AGARD SMP, Florence,
September 1996. AGARD Report 815,
February 1997.
2 Flomenhoft, H.I., Brief History of Gust Models
for Aircraft Design J. Aircraft v31 n5 pp1225
1227 1994.
3 Fuller. J.R., Evolution of Airplane Gust Loads
Design Requirements J Aircraft v32 n2 pp 235
246. 1995.
4 Vink,W.J.; A stochastic simulation procedure
compared to deterministic methods for PSD
gust design loads. NLR TP 98240, 1998.
5 Hoblit,F.M.; Gust loads on aircraft: concepts
and applications, AIAA,Inc.,1988
6 R.C. Scott, A.S. Pototzky, and B. Perry III,
Matched-Filter and Stochastic-Simulation-
Based methods of gust loads prediction,
Journal of Aircraft, 32(5):1047--1055, 1995
7 P.J. Goggin, Comparison of stochastic and
deterministic nonlinear gust analysis methods
to meet continuous turbulence criteria. Report
798, AGARD, May 1994.
8 R.C. Scott, A.S. Pototzky, and B. Perry III,
Computation of maximized gust loads for
nonlinear aircraft using Matched-Filter-Based
schemes. J.Aircraft,30(5):763--768, 1993.
9 R.Noback, S.D.G., P.S.D. and the nonlinear
airplane, TP 88018 U, NLR, National
Aerospace Laboratory, Holland, 1988
10 J.G. Jones, Statistical-Discrete-Gust Method
for predicting aircraft loads and dynamic
response, Journal of Aircraft, 26(4):382-392,
1989.
11 D.L. Hull, Design limit loads based upon
statistical discrete gust methodology, Report
798, AGARD, May 1994.
12 G.W. Foster & J.G. Jones, Analysis of
atmospheric turbulence measurements by
spectral and discrete-gust methods,
Aeronautical Journal}, pp 162-176, 1989.
13 E.Aarst & J.Korst, Simulated annealing and
Boltzman machines, John Wiley & Sons, 1989.
14 R.C. Scott, A.S. Pototzky, & B.Perry III,
Similarity between methods based on matched
filter theory and on stochastic simulation,
AIAA-92-2369-CP, 1992.
15 A.S. Pototzky & T.A. Zeiler, Calculating time-
correlated gust loads using matched filter and
random process theories. Journal of Aircraft,
28(5):346-352, 1991.
16 R.C. Scott, A.S. Pototzky, & B.Perry III,
Computation of maximized gust loads for
nonlinear aircraft using Matched-Filter-Based
schemes, Journal of Aircraft, 30(5):763-768,
1993.
17 J.E. Cooper & G.Dimitriadis, Prediction of
maximum loads due to turbulent gusts using
nonlinear system identification, In Proceedings
of the CEAS International Forum on
Aeroelasticity and Structural Dynamics,
Volume II, pages 71-78, Rome, Italy, June
1997
77
18 J.G. Jones, Formulation of Design Envelope
criterion in terms of Deterministic Spectral
Procedure, J. Aircraft, 30(1):137-139, 1993.
19 G.Rosenberg, D.A.Cowling, & M.Hockenhull,
The deterministic spectral procedure for gust
response analysis of nonlinear aircraft models.
Intl Forum on Aeroelasticity and Structural
Dynamics. pp 339 358. 1993
20 R.C. Scott, A.S. Pototzky, and B. Perry III,
Maximized gust loads for a nonlinear airplane
using matched filter theory and constrained
optimization. NASA TM 104138, 1991.
21 R.Noback, The Deterministic Power-Spectral-
Density method for nonlinear systems, TP
92342 U, NLR, National Aerospace
Laboratory, Holland, 1992.
22 R.Noback. The Deterministic Power-Spectral-
Density method for linear systems. TP 92062
U, NLR, National Aerospace Laboratory,
Holland, 1992.
V
V
w

= w/V
L=1/2 V
2
S Cl

Figure 4.1. Basic Gust Loading Mechanism
78
Figure 4.2 Bending and Torsion Loads. Linear A310.
79
Figure 4.3 Bending and Torsion Loads. Non-Linear A310.
80
Figure 4.4 F-100 Design and Correlated Loads
Figure 4.5: F-100 Design and Correlated Loads
81
Figure 4.6 Noback Aircraft c/g Acceleration
Figure 4.7 Noback Model c/g Acceleration by Aileron
82
1 2
0
2
4
6
8
10
12
Design Load -Correlated Load
C
e
n
t
r
e

o
f

G
r
a
v
i
t
y

A
c
c
e
l
e
r
a
t
i
o
n

(
m
/
s
2
)
IDPSD
MFB 1-D
MFB M-D
PEC
SDG
SSB
Figure 4.8: Results for Noback model, centre of gravity acceleration
1 2
0
1
2
3
4
5
6
7
8
Design Load -Correlated Load
C
o
G

A
c
c
e
l
e
r
a
t
i
o
n

C
a
u
s
e
d

b
y

A
i
l
e
r
o
n

O
n
l
y

(
m
/
s
2
)
IDPSD
MFB 1-D
MFB M-D
PEC
SDG
SSB
Figure 4.9: Results for Noback model, centre of gravity acceleration caused by aileron only
83
1 2
0
0.5
1
1.5
2
2.5
3
x 10
6
Design Load -Correlated Load
W
i
n
g

B
e
n
d
i
n
g

(
l
b
.
f
t
)
IDPSD
MFB 1-D
MFB M-D
PEC
SDG
SSB
Figure 4.10: Results for A310 model, wing bending
Figure 4.11: Results for A310 model, wing torsion
1 2
0
0. 5
1
1. 5
2
2. 5
x 10
5
Desi gn Load - Cor r el at ed Load
W
i
n
g

T
o
r
s
i
o
n

(
l
b
.
f
t
)
I DPSD
MFB 1 - D
MFB M- D
P E C
S DG
S S B
84
Figure 4.12: Comparison between SSB, MFB 1-D and IDPSD (labeled nob)
for Noback a/c load 2 (design load and gust shape)
Figure 4.13: Comparison between SSB, MFB 1-D and IDPSD (labeled nob) for A310 wing torsion
(design load and gust shape)
85
4.6 APPENDIX A4.1
Methods for design gust load prediction for nonlinear
aircraft
This appendix gives a brief description of the methods
considered in this chapter. They have been categorized as
either Stochastic or Deterministic methods, although
arguably the Statistical Discrete Gust methods could be
in their own section. Further details can be found in the
references.
4.6.1 Stochastic Methods
4.6.1.1 Probability of Exceedence Criteria
The Probability of Exceedence Criteria (PEC) method
9
is
an extension of the Power Spectral Density method
(PSD) for nonlinear aircraft. The PEC is stochastic and
attempts to calculate design loads. The procedure is as
follows
7,9
:
1. The flight conditions at which the design loads are
to be evaluated are prescribed and values of U

and
b
2
are determined from the airworthiness
requirements. b
2
is a coefficient used in the
expression for the probability that the load will
exceed the design load - its variation with altitude
can be found in reference 5.
2. A representative value of the rms gust intensity,
wr
, is computed using
( )
2
/ 4 1 1
2
2
2
b U
b
wr

+ +

3. An input white noise signal with

wr
is generated,
passed through a gust pre-filter and fed into the
nonlinear aeroelastic model. The resulting load time
history for load y is used to calculate the probability
that the design load will be exceeded in a turbulent
flow-field of intensity
wr
using
( )
wr
y
2
wr
2
2
wr
wr d
dyd
A 2
y
exp A
2
1
, y y P
d

,
_

>

where
wr y
A /
4. The design load is defined as the value of the load
for which

,
_

>
wr
wr d
U
y y P

2
erfc
2
1
) , (
where erfc is the error function complement.
Instead of calculating the probability distribution of load
y, it is possible to obtain the design load by estimating the
number of exceedences, N, of this load given by
4
) , y y ( P
dt
T
N
wr d
tot
>
where T
tot
is the total length of the simulation (in
seconds) and dt is the time step. Then, the array
containing the load response y is sorted from higher to
lower values and the design load is the Nth element of the
sorted array. If N is not an integer, linear interpolation
can be used to obtain the design load.

,
_

,
_

w
2
2
2
w
2
U
erfc
b 2
exp
This procedure only gives an estimate of the nonlinear
design load which may be substantially different to the
real value
9
. The estimate can be improved by repeating
the procedure for two values of
w
at which the value
of the following quantity is the same
Then, the design loads obtained for these two values of
gust intensity can be combined with the initial estimate
such that
) ( y 25 . 0 ) ( y 25 . 0 ) ( y 5 . 0 y
2 w d 1 w d wr d d
+ +
It has been suggested
4
that, instead of three simulations
with three different values of
w
, only one simulation
with 5 . 2 /

U
w
can be performed. The results
will be adequate in the altitude range of 22,000ft-35,000ft
since, in this range, the value of
wr
is very close to
5 . 2 /

U . This latter approach is also adopted in the

present work since it is suggested that increasing the total
simulation length at one value of
w
improves the
quality of the design load predictions by a larger amount
than increasing the number of simulations at different
values of
w
.
The correlated loads can be obtained using
P(z > z
c
|y y
d
) = 0.5
i.e. the probability of load z to be higher than the
correlated load, z
c
, when load y assumes its design value
is 0.5. This is implemented by extracting the value of z at
all the time instances were y=y
d
. The probability
distribution of these values is then calculated and the
correlated load is obtained as the load whose probability
is 0.5. As with the design load, the correlated loads can
be obtained using the number of exceedences instead of
the probability distribution.
Since the PEC input to an aeroelastic model is stochastic
turbulence, modelled as white noise, in order for the
method to work accurately, long simulation times are
needed so that the variance of the input is as close as
86
possible to
w
and its mean is almost zero. However,
the advantage that this method has over some of the less
computationally demanding discrete gust methods is that
the airworthiness requirements concerned are more
uniformly defined
5
.
4.6.1.2 Statistical Discrete Gust Method
The Statistical Discrete Gust Method (SDG) has been
introduced as a method that employs a better description
of atmospheric turbulence than the Power Spectral
Density method for extreme gusts on linear aircraft
10,11
.
This description is based on families of discrete 1-cosine
ramp gusts. The present implementation of the SDG
methodology is based on a similar implementation
9
. It
should be noted that the method was developed as an
attempt to bridge the gap between continuous turbulence
and discrete gusts methodologies and is being
continuously refined, most recently with the use of
wavelets. The SDG calculates design loads.
Figure A4.1 shows a single discrete gust, as used by the
SDG method. Initially, its velocity increases in a 1-cosine
fashion until, at a distance H, it levels out to the value U
which is given by
3 / 1
0
H U U
if H is less than L, the length-scale of turbulence, and
3 / 1
0
L U U
if L H . The value of U
0
is decided by the
equivalence of the design value of
g
in the continuous
turbulence PSD analysis to the SDG analysis as
11
4 . 10
0
g
U

where
g
is obtained from the airworthiness
requirements
5
.
For extreme turbulence the scaling of equation the gust
velocity equation changes to
6 / 1
0
H U U
This is how the SDG methodology bridges the gap
between continuous turbulence and discrete gusts.
Continuous turbulence is assumed to be self-similar,
which is where the 1/3 scaling law comes from. Self-
similarity can be modelled as a stretching transformation.
In the time-domain, if the time axis is stretched by a
certain amount, h, the dependent variable, say y(t), will
be stretched by

h . The similarity parameter can be
chosen such that the function ) (ht y h

is statistically
independent of h. This value for can be obtained by
considering the spectrum, () of the process y(t), when
stretched by h, which in reference is shown
12
to satisfy
( )
( ) ( )

+
h h /
1 2
In the special case where the process y(t) is turbulent, the
Von Karman spectrum applies, i.e.
( )
( ) ( )
6 / 11
2
2
2
33 22
6 / 5
2
2
11
339 . 1 1
339 . 1
3
8
1
2
339 . 1 1
1

,
_

,
_

,
_

,
_

,
_

+

V
L
V
L
L
V
L
L
g
g

Simple algebra shows that the limit of both
11
() and

22
() as tends to infinity (which defines the inertial
subrange where self-similarity applies) is
3 / 5
lim

A
where A is a proportionality constant. Consequently
( ) 3 / 5
3 / 5
1 2

,
_

A
h
A h
For this expression to be satisfied, h must vanish from the
left-hand-side, or
3
5
1 2
Hence for continuous, self-similar turbulence, 1/3.
Discrete gusts are extreme events for which self-
similarity breaks down. They are larger-scale and more
ordered events than the background turbulence within
which they are contained. The similarity parameter for
such events is given by
10
3
3
3
1 D

where D is termed the active volume of turbulence and
has values 3 2 < D . For D=3 the standard self-
similar value, 1/3, is obtained. For a value of D=2.5,
the extreme turbulence similarity parameter is obtained,
1/6. Hence, with a simple change in the scaling law,
the SDG method can be made also applicable to extreme
turbulent events like discrete gusts.
At a particular value for the gust-length, H, the nonlinear
aeroelastic system under consideration will exhibit a
maximum load response. The maximum value of this
maximum response,
1
is an estimate for the design
load, y
d1
. A second estimate is obtained using a pair of
gusts as shown in figure A4.2. Here, there are three
parameters that govern the gust shape, H
1
, H
2
and the
spacing between the two gusts, S. The values of these
parameters are varied until the maximum,
2
, is
obtained. Another two estimates for the design load are
87
calculated using two pairs of gusts and four pairs of
gusts. Finally, four design loads are calculated using
0 4 4 4
0 3 3 3
0 2 2 2
0 1 1 1
U p y
U p y
U p y
U p y
d
d
d
d

with p
1
=1.0, p
2
=0.81, p
3
=0.57 and p
4
=0.40. For highly
damped systems the first two design values are more
important, for slightly damped ones the last two design
values predominate.
For linear systems, estimating the maximum response
due to SDG gusts is simple since superposition can be
employed. For nonlinear systems this estimation can only
be performed by means of an optimization scheme,
especially for the longer gust-shapes. The optimization
scheme chosen for this study was Simulated Annealing
13
.
4.6.1.3 Stochastic Simulation
The Stochastic Simulation method (SS) models
continuous turbulence as a white noise input with a Von
Karman spectrum, in the same way as the PEC method.
Hence, the SSB is stochastic and can calculate design
loads, correlated loads and worst-case gusts, given a
target value for the design load. The procedure is as
follows
14
:
1. A Gaussian white noise signal with unity variance is
generated and fed through a gust pre-filter, such as
( )

,
_

+
,
_

+
,
_

,
_

+
,
_

V
Ls
0898 . 0 1
V
Ls
823 . 0 1
V
Ls
083 . 2 1
s
V
L
1298 . 0 1 s
V
L
618 . 2 1
V
L
s G
g
The output of the filter is a time history of
continuous turbulence data. The object is to identify
segments of this time history that lead up to peak
loads.
2. A number of long time-domain simulations are
performed
3. The load time histories obtained from the
simulations are analysed. Instances in time are
isolated where the load exhibits a peak near a
prescribed value or within a specified range. Then
standard durations of time data leading up to the
peak values are extracted, lined up in time and
averaged. The result is 'averaged-extracted' time-
histories of the excitation waveform (input to the
gust filter), gust profile (section of turbulence data)
and load. These have been shown to be directly
equivalent to results obtained by the MFB
methods
14
, if the value of the turbulence intensity
g
is selected appropriately.
To ensure that there is an adequate number of extracted
samples so that the final waveforms are as smooth as
possible, very long simulations are required (1000
seconds has been suggested
14
). Long simulation times
also ensure that the white noise input has a variance very
close to unity and a mean very close to zero. Finally, the
extraction and averaging process must take place
separately for positive and negative peak load values.
The stochastic simulation method, as outlined here
cannot be used on its own since it requires a target load to
be specified, around which it will search for peaks in the
load response. This target load value can be supplied by
another method. The authors of ref. 14 used the MFB
multi-dimensional search procedure to obtain the target
design load value and picked peaks in the SSB load
output within % 8 t of that value. Of course, the object
of their work was to show that the MFB results are
equivalent to stochastic results. In a straightforward
design loads calculation it would be extremely wasteful
to use two of the most computationally expensive
methods to produce the same results twice.
However, it is suggested here that the SSB method can be
used to supplement results obtained by the Probability of
Exceedence Criteria method. As mentioned earlier, the
PEC method will only produce values for the design and
correlated loads. It will not calculate time-variations of
the loads or the gust velocity. The SSB, on the other hand
can produce design and correlated load responses and
critical gust waveforms. Hence, the PEC method can be
used to yield a target value for the design load to be
subsequently used with the SSB method.
4.6.2 Deterministic Methods
Figure A4.3 and table A4.1 summarize the Deterministic
procedures. An input signal to the "first aircraft system",
H
1
, is generated by feeding a pulse through a (von
Karman) gust filter G, with ,G(jf),=[
n
ww
(f)]
2
. The power
spectrum of the input to the first system will thus have
the shape of the von Karman spectrum. The pulse
strength k is variable in the MFB method, and constant in
the IDPSD (k=U

) and SG (k=U

T, where T = length of
gust input) methods. It should be noted, that the gust
filter in the MFB method is only an approximation of the
von Karman spectrum, and in the version used in this
report it is the Hoblit approximation .
The first aircraft system, H
1
, represents the non-linear
aircraft equations of motion in MFB and SG. In IDPSD,
H
1
is a linearized version of the non-linear aircraft, by
replacing the non-linearity by a linear element with an
"equivalent gain", K
eq
. K
eq
is a multiplication factor to the
original gain in the feedback loop, with 0K
eq
1
For nonlinear systems, the three Deterministic methods
apply different procedures:
- MFB varies the strength k of the input pulse to the
first gust filter.
- IDPSD varies the value of the equivalent gain that
represents the nonlinearity in the first system.
- SG varies the phase relation of the gust filter, which
is limited to only four different phase relations.
88
4.6.2.1 Matched Filter Based 1-Dimensional
search
Matched Filter Theory (MFT) was originally developed
as a tool used in radar technology
15
. The main objective
of the method is the design of a filter such that its
response to a known input signal is maximum at a
specific time, which makes it suitable for application to
gust response problems. The method can only be applied
to linear systems because it makes use of the principle of
superposition, which does not apply to nonlinear systems.
However, by applying a search procedure, it can be
adapted to provide results for nonlinear aircraft. The
method is deterministic.
The technique is quite simple and consists of the
following steps
15,16
:
1. A unit impulse of a certain strength K
g
is applied to
the system.
2. The unit impulse passes through a pre-filter
describing gust turbulence (usually the Von Karman
Gust pre-filter).
3. The pre-filtered input is fed into the aircraft model
and the response of the various loads is obtained
(e.g. wing root bending and torsional moments).
4. The response of the load whose design value is to be
estimated is isolated, reversed in time, normalized
by its own energy and multiplied by U

, the design
gust velocity (which is determined by airworthiness
requirements
5
).
5. The resulting signal is the input that maximizes the
response of the chosen load for this particular
impulse strength, K
g.
It is then fed back into the
system (first the Gust pre-filter, then the aircraft
model) in order to obtain the response of the load
whose design value is to be estimated and also the
responses of the other loads (which are termed the
correlated loads).
6. The procedure is repeated from step 1 with a
different K
g.
The characterization of the method as one-dimensional
refers to the variation of K
g
. The end result is a graph of
peak load versus initial impulse strength. The maximum
of this function is the design load and the gust input that
causes it is termed the Matched Excitation Waveform. It
must be mentioned at this point that the method does not
guarantee that the maximum load for a nonlinear aircraft
will be obtained. As was found in refs. 7 and 17, the
variation of peak load with initial impulse strength for
some types of nonlinearities (e.g. freeplay and bilinear
stiffness) does not display a global maximum (instead it
slowly asymptotes to a certain value).
4.6.2.2 Deterministic Spectral Procedure
This method was first proposed by Jones
18
. In its most
general form it is based on the assumption that there
exists a single deterministic input function that causes a
maximum response in an aircraft load. It states that a
design load on an aircraft can be obtained by evaluating
the load response to a family of deterministic gust inputs
with a prescribed constraint. In practice, this implies a
search for the worst case gust, subject to the constraint
that the energy of the gusts investigated is constant. The
method is deterministic. The procedure consists of the
following steps:
1. A model input shape in the time-domain is
generated.
2. The input shape is parameterized to produce a set of
describing coefficients
3. The coefficients are used to generate the input
waveform
4. The energy of the input is constrained by dividing
the signal by its rms value
5. The constrained waveform is fed into a turbulence
pre-filter and next through the nonlinear aircraft
system
6. The aircraft load response is assessed. If it has not
been maximized the coefficients that generate the
input are changed and the process is repeated from
step 3.
This iterative procedure requires a constrained
optimization scheme, to ensure that the maximum load
has been obtained, and a model input shape. The
optimization scheme proposed originally
18
was simulated
annealing. Another approach
16
is to convert the
constrained optimization problem to an unconstrained
one by means of the Kreisselmeier-Steinhauser function.
As for the generation of the initial input shape, two
approaches have been proposed. In ref. 19 a white noise
gust model is used. The problem with this approach is
that it is more difficult to parameterize a random signal
than a deterministic one. Alternatively
16
, the MFB 1-
dimensional search results are proposed as the input to
the DSP loop, which results in what is called the MFB
multi-dimensional search procedure.
The parameterization process is probably the most crucial
aspect of the DSP method. Input waveforms have to be
described by a minimum number of coefficients to
minimize computational cost but this description has to
be as accurate as possible. Again, two popular procedures
can be found in the literature. The first
19
is to fit the
waveform by a number of half-sinusoid (or cosinusoid)
functions. The other approach is to fit the waveform
using a set of Chebyshev polynomials
16
. In the same
reference, a Fourier series approach was considered but it
was found to be much more computationally expensive.
The most common implementation of the DSP method is
the Multi-Dimensional Matched Filter Based method
which is described next.
4.6.2.3 Multi-Dimensional Matched Filter Based
Method
The Multi-Dimensional Matched Filter Based (MFB
Multi-D) method
16,20
for gust load prediction for
nonlinear aircraft is a practical application of the
Deterministic Spectral Procedure. It was designed to
provide a more computationally efficient alternative to
the Stochastic Simulation Based approach. Reference 16
shows how the method provides almost identical results
to those obtained by use of the SSB but with less
computational effort. The method is deterministic.
89
The MFB Multi-D approach revolves around the fact that
the usual design envelope analysis can be reformulated as
an exactly equivalent time-domain worst-case analysis. In
other words, the search for a worst-case gust load in the
presence of a turbulence field of prescribed intensity is
equivalent to the search for a design load
19
. Hence, the
simplest possible procedure for determining the worst-
case load is to simulate very long patches of turbulence
and to look within the load response of the aeroelastic
system in question for the design load. This is the
stochastic simulation approach that requires significant
amounts of computation.
The worst-case load problem can be simplified by noting
that the significant part of a long turbulent signal that
causes the maximum load is short and can be
approximated as a discrete gust. Hence the MFB Multi-D
method searches for the single discrete worst-case gust
waveform thus avoiding the need for long simulation
times.
The implementation of the method is as follows, also
depicted graphically in figure A4.4:
1. An initial guess for the worst-case gust waveform
(or matched excitation waveform) is obtained by use
of the 1-dimensional MFB procedure.
2. The initial guess is parameterized. In the present
application the parameterization scheme used is
Chebyshev Polynomials.
3. The values of the various parameters are changed
and the resulting waveform is fed into the
aeroelastic system (including a turbulence pre-filter
as described earlier).
4. The resulting maximum load is compared to the
previous value for the worst-case gust load and is
accepted or rejected according to some optimization
procedure. The optimization procedure used for the
present application is Simulated Annealing. The
procedure is repeated, i.e. the parameters are
changed again resulting in a new gust waveform
which is then used as an input to the system, until
the worst-case gust load is obtained.
4.6.2.4 Indirect Deterministic Power Spectral
Density Method
The Indirect Deterministic Power Spectral Density
method (IDPSD)
20,21
, is derived from the Design
Envelope Analysis
5
of the continuous Power Spectral
Density method. For linear aircraft it yields design loads
equal to those obtained by the PSD method but using a
deterministic input, in a similar way to the linear MFT
method. For nonlinear systems it can be extrapolated to a
1-dimensional search procedure, equivalent to the MFB
1-D search but involving a linearized representation of
the system. The method is deterministic.
The IDPSD procedure is very similar to the MFB 1-D
method with two main differences. Firstly, the IDPSD
method uses a different gust filter and, secondly, the
initial excitation is applied to a linearised version of the
system whose output is then reversed, normalized and fed
into the nonlinear system. Hence, the MFB 1-D method
consists of a filtered impulse of variable strength fed into
the nonlinear system, the resulting gust waveform being
fed into the same system. In the IDPSD method, an initial
input of constant strength is fed into a linearised system,
called the first system, whose nonlinear element has been
replaced by a variable gain. The resulting waveform
forms the input to the nonlinear system, called the second
system. The search procedure consists of varying the
linear gain until the response of the second system is
maximized.
The input to the first system is given by ) (t V U

,
where

U is the design gust velocity and V(t) is the

Fourier Transform of the two-sided Von Karman
Spectrum, ( )
ww
, given by
( )
2
2
2
339 . 1 1
339 . 1
3
8
1

,
_

,
_

,
_

+

V
L
V
L
V
L
ww

where is the radial frequency, L is the turbulence

length-scale and V is the aircraft velocity
22
. This input
can be alternatively defined as the Auto-Correlation
function pertaining to ( )
ww
, i.e.

d t R t V
t j
ww
e ) (
2
1
) ( ) (
22

The Von Karman Spectrum can be expressed in a more
practical form as the Auto-Correlation function of the
filtered MFB impulse,
2
) (
) ( ) (
) (

g
g g
u
t u u
t V
+

where u
g
is the MFB filtered impulse gust velocity, the
overbars denote averaging and is an integration
variable. The solid line is the Fourier Transform result
and differs from the Auto-Correlation result (dotted line)
in that it takes negative values away from the peak. As a
consequence the Auto-Correlation result was preferred
for the present work.
The IDPSD Method procedure is as follows:
1. ) (t V U

is formed, say using equation (6).

2. The input is fed into the linearized aircraft model
with linear gain K and the response of the various
loads is obtained (e.g. wing root bending and
torsional moments).
3. The response of the load whose design value is to be
calculated is isolated, convoluted by V(t),
normalized by its own energy and multiplied by

U , the design gust velocity.

4. The resulting signal is the input that maximizes the
response of the chosen load for this particular
linearised gain, K. The signal is then fed into the
nonlinear system in order to obtain the response of
the load whose design value is to be calculated and
also the responses of the correlated loads.
5. The procedure is repeated from step 2 with a
different K.
Reference 21 suggests that the values of the linearized
gain should be between 0 and 1.
90
Table A4.1 Elements of Deterministic Methods
4
Element Matched filter
(Scott e.a.)
IDPSD
(Noback)
Spectral Gust
(Brink-Spalink e.a.)
Impulse
Strength k
k variable k = U

k = U

*T
Gust
Prefilter G(jf)
|G(jf)|
n
(f)
One set (f)
|G(jf)| =
n
(f)
One set (f)=0
For all f
|G(jf)| =
n
(f)
four sets (f)
Aircraft
System H
1
(y)
(Nonlinear)
set of equations
for output y
Linearized
Equations;
Variable
"equivalent gain"
Nonlinear
set of equations
for output y
Calculation
y-norm:
1
]
1

1
]
1

) ( ) ( =

) ( =
*
-
2 / 1
2
-
2 / 1
norm
df jf
s
jf s
dt t
s
y
-----------------------------------------------------
For linear system:
y y
U
k
k df G G
H H k
y
des norm
y
*
1 1
+
-
2
2 / 1
norm
= = if

A
= . . =

1
]
1

"Critical
gust
profile" w(t)
For linear systems
same profile for
matched filter and IDPSD
Aircraft
System H
2
(y)
Nonlinear set
of equations
Y
des
Variable k
y
des
= [y
t
]
max
Variable gain
of H
1
(y)
y
des
= [y
t
]
max
SG stops
here:
Four values
for y
norm
,
T
y
y
(max)
=
norm
des
91
4.7 Appendix A4.2 Description of Aircraft
Models
Three symmetrical aircraft models have been considered
in this research. The first one is a simple model of a
large transport aircraft with two degrees of freedom, pitch
and plunge, and a load alleviation system that feeds back
the centre of gravity acceleration to aileron deflection.
The model is shown in figure A4.5. The functions C(s)
and D(s) are the transformed Wagner - and Kssner
functions representing unsteady aerodynamic loads.
Output y in the figure is the centre of gravity
acceleration, and output z is the centre of gravity
acceleration caused by aileron action only. This model is
called the Noback-model in this report.
The second model represents an aircraft with "Fokker-
100-like" characteristics. This model has the two rigid
degrees of freedom pitch and plunge, and ten symmetric
flexible degrees of freedom. This flexibility is
represented by the first ten natural modes of the aircraft
structure. Aerodynamic forces are calculated with strip
theory, and unsteady aerodynamics is accounted for by
Wagner - and Kssner functions. The wing has 27 strips
and the tail 13; the fuselage is considered as one lifting
surface. The Wagner - and Kssner functions are
calculated at 3 locations on the wing and at 1 location on
the horizontal tail.
The gust penetration effect and the time delay of the
downwash angle at the tail with respect to the wing are
included. Taking these two effects into account, makes it
necessary to apply time delays to the gust input, and to
the state variables (because the angle of incidence at the
reference point on the wing is a function of all states)
respectively. Especially the latter considerably increases
the total number of system states.
A Load Alleviation System is implemented in the model
that feeds back the load factor to a (symmetrical) aileron
deflection. Figure A4.6 shows the aircraft system with
the feedback loop to the aileron input. The configuration
of the Fokker 100 model used in this report is:
m
a/c
= 40,000 kg I
y
= 1.782 10
6
kgm
2
V = 220 m/s, altitude = 7000 m
centre of gravity location at 25 % mean-
aerodynamic-chord.
The third model has been distributed at the Gust
Specialists Meeting of March 1995. It represents an A310
aircraft, containing plunge, pitch, and 3 symmetric
flexible degrees of freedom. Unsteady response is
assumed instantaneous, and gust penetration is not
represented. The aircraft with control system is depicted
in figure A4.7. The centre of gravity acceleration is fed
back to both the ailerons and the spoilers through a
feedback gain of 30 degrees per g load factor. Ailerons
and spoilers have the same authority: deflections between
0 and 10 degrees. This means that the nonlinearity in this
control system is "non-symmetric"; the control surfaces
can only deflect upward. The load quantity outputs of this
system are the increments of:
- Engine lateral acceleration [g].
- Wing bending moment [lb.ft].
- Wing torque [lb.ft].
- Load factor [g].
92
Figure A4.1 Single SDG Gust
Figure A4.2 Pair of Statistical Discrete Gusts
93
Figure A4.3 Process for Deterministic Methods
94
Figure A4.5 Noback Aircraft Model
Figure A1.4: Graphical description of MFB Multi-D procedure
95
a/ c r esponse
l oads r esponses
0
t r i m
u [ 3 ]
sel ect _dn
[ t , w]
i n p u t
0
el ev at or
x ' = Ax +Bu
y = Cx +Du
St at e- Space
Mu x
Mu x
1
Ta. s +1
Ai l er on
-K
- 20 deg/ g - 10<y <10
Figure A4.6 Fokker-100 Model
u[ 4]
sel ect _dn
1/ . 3048
m2f t
pi / 180
deg2r ad
st at espace1
To Wor kspace1
out p
To Workspace
nfi l t(s)
dfi l t(s)
TFF gust
f i l t er x' = Ax+Bu
y = Cx+Du
St at e- Space
1
taus.s+1
Spoi l er
Sat ur at i on
Mux
Mux1
Mux
Mux
mai ngai n
Gai n
[ T, i npt ]
From
Workspace
1
t aua. s+1
Ai l er on
Figure A4.7 A310 Model
96
5 A More Global Approach
5.1 Why a more global approach
It comes from the necessity to get rid of insufficiencies of
classical load regulations, the main lines of these
regulations being:
Limit loads are defined as "maximum loads"
expected in service.
Regulations prescribe the set of loading conditions
(ex.: manoeuvres), or directly the computation
procedure (gust, ground loads), to be considered for
finding these "maximum loads".
Ultimate loads result from multiplication of limit
loads by a prescribe safety factor.
The sources of difficulties are principally:
The chronic lack of exhaustively of regulation
loading conditions set up from flight experience of
past programme.
Already with conventionally controlled aircraft
manufacturers had to add "company" design load
cases, for instance to cover countered maneuvers
where the pilot, remaining inside limit values of
"official" load factors and control surface deflections,
could easily make severe structural loading.
Matters worsen when new technologies come, which
has been met, in particular with:
the design of fly by wire combat aircraft and
the associated concept of care free piloting,
where "maximum loads" can be reached every
day as result of extremely complex and various
dynamic maneuvers, far from regulation
maneuvers.
the design of re-entry vehicles with their "hot
structures", where limit conditions result from
combinations of mechanical, thermal loads, and
aging conditions, closely depending on
structural design.
The need to clarify the meaning of the word
"maximum loads" ; its have been often restricted to
loading conditions corresponding to maximum values
of "general load" components, notion becoming
insufficient when "long beam theory" is not relevant
(e.g. delta wings), where local structural failure
modes are not only led by "general loads".
Still more severe difficulties occur when thermal
loads, or any physical or chemical environmental
conditions, or aging and fatigue effects, must be
considered in addition to mechanical loads.
The safety factors philosophy
first it is a need to clarify the present safety
factor rules when other physical effects
(thermal, environmental, aging/fatigue, ) are
added to mechanical loads, where several
components of safety factor must appear,
corresponding to each physical effects.
more fundamentally we have to open the debate
of safety factor evolutions with innovation,
with the progress both of design solutions and
of analysis process, knowing that we are to day
unable to quantify, inside the present global
safety factor, separated contributions of loads,
manufacturing, strength, , or of any other
uncertain elements.
Faced with these questions since the mid 70ies with
MIRAGE 2000 programme and after with RAFALE,
DASSAULT AVIATION have developed and
experienced the "more global approach", already
presented to AGARD SMP in 1984 and 1996 (ref. 1 and
2) and reminded hereafter to be proposed now to the
RTO community.
To note that this approach, including extensions to
thermal loads, have been carried by ESA and CNES for
design loads of HERMES space shuttle .
5.2 Limit Loads
5.2.1 Basic principles of the "more global
approach" for limit loads
They are:
To keep (even to reinforce) the limit load definition
of classical regulations:
Limit loads are the maximum loads expected in
service .
To consider that it is not necessary to prescribe any
particular set of loading conditions within
regulations.
"Maximum loads" must come from scenario analyses
of missions/flight conditions/ environments, suited to
the designed product.
In practice, this don't prevent aircraft designer from
building a set of
"reference design load cases",
under his responsibility and to demonstrate that these
"reference design loads" envelop the maximum loads
expected in service.
To clearly define the meaning of the sentence :
"Maximum loads expected in service" ,
and to propose a practical process for their
determination (see hereafter).
5.2.2 "Maximum loads" through "Load Severity
Indicators"
The notion of "maximum Loads" has a meaning only
through the effects of loads induced on the structure:
A load case is referred to as a maximum load case as
soon as it produces the maximum value of at least 1
failure mode strength criterion.
97
Which need in theory :
To identify of all structure failure modes liable to
occur under mechanical loading (local stress - or
strain - induced ruptures, local or general buckling,
non-allowable overall deflections, ), and more
generally under all other physical effects (thermal,
aging, ).
To allocate to each one of these failure modes of a
scalar strength criterion calculable in function of
the loading conditions and of the structure design.
When necessary the strength criteria may take into
account thermomechanical and aging effects.
To sweep all "expected" loading conditions (see
6.2.3) calculating each of these strength criteria.
To reduce the effort of monitoring thousands of local
strength criteria, we have introduced the notion of :
"Load Severity Indicators".
Which are few tens to few hundreds of scalar indicators
standing in monotonic relation to a structure area strength
criteria, whatever the loading.
As "load severity indicators" are generally chosen:
components of stress or strain in pilot points,
internal reactions (e.g. : loads on the wing or control
surface attachment bearings),
classical "general loads" components (shear force,
bending moment ) on particular sections.
*
Computation management will be simplified if the
severity indicators remain linear functions of the loads ;
they can then be calculated at low cost in function of
flight parameters, starting from a matrix of "load
severity indicator operators" giving the relation with
flight mechanics state vector, this table being built prior
to maneuver computations.
The strain gauge distribution of flight test aircraft will
attempt to reflect the choice of load severity indicators,
thereby providing for calibration and validation of the
operators and thus, of the whole load computation
process.
Once "load Severity Indicator operators" are
built/calibrated/validated, the computer cost of maximum
load case selection comes cheap, corresponding to linear
combinations of "load severity indicator operators",
downstream sweeping of:
flight mechanics simulations, (numerical simulations
/ real time flight simulator),
environmental aircraft responses (gust, turbulence,
),
ground load conditions,
etc ,

marking as limit load case conditions where maximal of
"load severity indicators" are reached,
and/or :
checking that these maximal remain under the level of
"reference design loads" chosen a priori.
5.2.3 "Maximum Loads Expected in Service"
That means that we have to sweep all possible scenario,
during an aircraft life, of missions / maneuvers /
environments /, computing previous Load Severity
Indicators, and selecting, as design load cases, loading
conditions where load severity indicators are maximal.
When relevant, it can correspond to probabilistic
analyses in the spirit of Continuous Turbulence
regulations( e.g. FAR 25, appendix G)
to determine from mission analysis limit value of
"load severity indicators", corresponding to 1
average exceeding per aircraft life .
to ensure that the limit load set (or the "reference
design load" set of the manufacturer) envelop these
limit values.
5.2.4 Application to design of "fly by wire"
aircraft
It have been detailed in reference 2, the principle is to
integrate the designs of structure and of Flight Control
System via the following iterative process :
Start from a first set of "reference design loads"
from aircraft manufacturer experience
reflecting flight quality requirements
Design of airframe
supported by F.E./Aeroelasticity analyses /
optimizations
delivering "load severity indicators" operators
and their associated limit values
Design of F.C.S.
to maintain "load severity indicator" responses
below their limit values for all possible scenario
of missions / maneuvers / environments,
or
to define new limit load cases ( airframe
design iteration).
5.3 Ultimate load definition and Safety Factors for
multiphysical effects
When limit loads contain only "mechanical effects" the
definition could remain "as is" :
98
Ultimate loads result of multiplication of limit loads
by a prescribed safety factor.
When others physical effects (thermal, aging, ) occur
in limit conditions, specific safety factors must be
applied successively and separately on each of these
effects (the others remaining at their limit values) ; for
instance:
on heat fluxes or on parts of heat fluxes or on
resulting temperature fields.
on life duration for fatigue/aging loads.
for each kind of other physical/chemical
environmental conditions.
The nature and the levels of these specific safety factor
must be adapted for each type of vehicle liable to meet
these special physical effects, levels could result from
probabilistic considerations ( see 6.4.2 ) .
Another requirement for these multiphysical effect safety
factors is to keep possible a verification test in the
ultimate conditions; it leads to avoid safety factors on
"calculation beings" physically inseparable by test
conditions as with the present thermal stress safety factor
of AIR2004-E and other regulations.
5.4 Safety factors evolution with innovations
5.4.1 The particular case of fly by wire aircraft
Knowing that the flight control system, with a "care free
piloting functions, can protect against limit load
overshoots, a debate may arise as to the pertinence of a
change to the safety factor (currently 1.5) ; such
discussions come up against great difficulties :
The current safety factor covers aspects other than the
occurrence of load conditions that are severer than
the limits ; they involve, amongst others :
potential flaws in the load computation models (force
fields applied to the airframe) in function of loading
conditions (flight mechanics state vector ).
every unknown differences between the airframes in
service and the one that was qualified (non-detected
manufacturing or material defects, various non-
detected corrosion-, fatigue- or impact-induced
damage types, etc).
For all of these factors, there are non sufficiently
conclusive probability models available that give the
load or structure strength overshoot statistical
distributions ; we do not know how to quantify these
factors separately within the global safety factor.
The global safety factor of 1.5 can be justified
quantitatively only by the acquirements of
experience, based on observation over half a century
of a globally satisfactory structural strength of
aircraft in service ; but this safety factor cannot be
decorrelated from the rest of the environment of the
used construction techniques, analysis methods and
verification process. Any partial change that occurred
in the technical environment requires a demonstration
to establish that there is no regression in Safety
(cf. qualification rules for composite materials),
although this would not mean that any likely gain in
one point can be exchanged against a reduction of the
margin in another point.
A further element for debate bears on the advantages
that might be drawn from a potential safety factor
reduction:
For new projects, the potential gain in terms of
structure mass is likely to be slim, the safety factor-
to-mass exchange ratio will remain far below
proportionality (fatigue sizing of metallic parts,
design to technological minimal for large areas, areas
with design-sizing aeroelasticity constraints, ).
The discussion is somewhat more open, for existing
and proven by flight service airframes, when
considering any specific or circumstance-related
maneuver performance characteristics improvement.
5.4.2 Towards probabilistic approaches
At long range a complete reconstruction of structural
analysis process would be required , to get out of the
above mentioned piling of safety margins, resulting from
ignorance of the part, within present global safety factor,
assigned to any innovation of design solution or of
analysis method .
This long range research could be founded on a full
probabilistic approach, considering all items of airframe
qualification : loads, types of design ,calculation and test
process , manufacturing process, flight service use,
fatigue & corrosion and any other aging effects, control
plan , , and human error possibilities everywhere inside
the process .
It is a subject in itself, which could be proposed to further
RTO discussions .
References :
1. C. PETIAU, M. DE LA VIGNE Analyse
Arolastique et Identification des Charges en Vol
AGARD conference proceedings No 373 -
"operational load data" - Sienne 1984 .
2. C. PETIAU Evolution de la philosophie des
charges de dimensionnement des avions militaires
.AGARD report No 815 - "Loads and requirements
for military aircraft" - Florence 1996