Landing Gear Design Loads
Landing Gear Design Loads
Landing Gear Design Loads
.AGARD
1 ADVISORY GROUP FOR AEROSPACE RESEARCH & DEVELOPMENT
7 RUE ANCELLE 92200 NEUILLY SUR SEINE FRANCE
S I .'
91-08950
The Mission of AGARD
According to its Charcr,the mission of AGARD is to bring together the leading personalities o theNATO ation in le hietld,
of science and technology relating to aerospace for the following purposes:
- Recommending effective ways for the member nations to use their research and deelopment capabilitic, lot ie
common benefit of the NATO community;
- Providing scientific and technical advice and assistance tothe Military Committee in the field of aerospace research and
development (with particular r( qard to its military application):
- ontinuously stimulating advant es in the aerospace sciences relevant to strengthening the common detence posture;
- Improving the co-operation amit g member nations in aerospace research and development;
- Providing assistance to member nations for the purpose of increasing their scientific and technical potential;
- Rendering scientific and technical assistince,as requested. to other NATObodies and to member nations inconnection
with research and development problems in the aerospace field.
The highest authority within AGARD is the National Delegates Board consisting of officially appointed senior representatises
from each member nation. The mission of AGARD is carried out through the Panels which are composed of experts appoiinted
b,the National Delegates, the Consultant and Exchange Programme and the Aerospace Applications Studies Programme. lhe
results of AGARD work are reported to the member nations and the NATO Authorities through the AGARD scries of
publications ofwhich this is one.
i.SBN'
, -835-611-1
Experience has shown that landing gears are often neglected in the early design process of an aircraft because interest is
focu~scd mainly on flight performance aspects. H-owever, the operational utility of an aircraft is strongly influenced by the
landing gear. This importance is not adequately reflected in the available specifications. In addition. operaion froim hiimb-
damaged and repaired runways is not reflected in landing gear design requirements.
AGARD SNIP decided to organize a Specialists* Meeting on landing gear design loads to provide a forum for the exchange of
experiences between the NATO nations with the aim of advancing landing gear design criteria and meihods of landing gear
anialyses. The Meeting was to review existing design practices and specifications. consider the various methods used for load
measurement and data analysis and formulate guidelines for future design procedures.
The Meeting %as well attended in terms oif both number and quality of participants and was fiillowed by a general discussion
T~hesumma]Y if this discussion is gisvn at the end of this book.
Preface
L'cxpericnicc a mnti-6r que lesatterrisseurs sont soiuvent iagliges dlans la phase de iiifinitiiin prcliminaire des asions. parce qut.
Inmtcr~i se pilite surtout stir les aspects de performances de vol. Cependant, l'utilit6 operatiiinnelle d'un asion est fortement
influerice par Ics sp~cificati'ims op' rationnelIles des atrisseur, Cette importance nest pa bien rcfitiie dans lesr Riement,
dispiiniblcs. D~eplus. i ne figiie rin dans les ri~glcments sur les op~rations a partir de pistes cnclommagics par bombes et
reparces.
e Panel Siructures et Maiieriaux de IAGARD a decide de teitir une rfunion dlespecialmses sur les charges de conception des
atterrisseiirs afin diiffrir un forum I'ccha~ige dexpcriences entre pays de l'OTAN dans Ic hut dc faire priigiesser lescriteres ei
ics miethodes de ealcul des atterrisseurs.
Ce Meeting devait passer en revue les nt~thodes dleconc,:ptior. et riglements. examiner Ies differentes prattijues qiiant aux
mesures de charges eii service et formulet des recommandatitins sur Ic, procedures futures de conception.
( e Meeting a regroup6 une assistanyc nombreuse et tie iualite et a 6tc suisi d'une discussion generale. La synthese de cette
discussion extdiinnee ii latfin deccc recucil.
l).Chaumette
Chairman. Sub-Committee tin
Landing Gear Design Loads
16&L-
Structures and Materials Panel
SUB-CONIMITTEE ON
LANDING GEAR DESIGN LOADS
Members
A.Dimnarogonas GR
R.Frevmann it;
EFuente SP
V~iiavotto IF
JiJGlaser CA
G.Gnrininger GE
J.J.Kacprzynski CA
R.F.O'Coninell us
H.Ii.Ottens NL
C,L.Petrin us
A.Salvetti IT
O.Sari IT.
O.Sensburg GE
A.ETovar dc Lemos PO
A.P.Ward UK
PANEL EXECUTIVE
Mr Murray C. McConnell (UK)
. ... .. .. ...
Contents
t-reface/P:eface i
Reference
OPENING ADDRESS
Failure Analysis Case Histories of Canadian Forces Aircraft Landing Gear Compor.entsI
by P.Beaudet and M.Roth
SESSION I - SPECIFICATIONS
Paper 7 withdrawn
Paper 9 withdrawn
Long Time Measurements of Landing Gear Loads on SAAB SF-340 Commuter Aircraft 10
by A.l.Gustavsson, A.F.Blomn and L.Helmersson
V
Reference
Paper 17 withdrawn
Paper 18 withdrawn
Paper 21 withdrawn
Report of Discussions R
by D.Chaumette
Vi
I-I
by
!'jz. P. Beaudet
Directorate Aerospace Support Engineering
National Defence Headquarters
Ottawa, Ontario, Canada, KIA 0K2
and
Dr. M. Roth
Quality Engineering Test Establishment
Department of National Defence
Ottawa, Ontario, Canada, KIA OK2
SUMMARY
Despite the extensive landing gear design analyses and tests carried out
by the d-migners and manufacturers, and the large number of trouble-free landings
accumulated by the users, the Canadian Forces, as well as others, have experienced
a range of problems or failures with landing gear components. Different data
banks were surveyed and over 200 case histories on more than 20 aircraft types
were reviewed in oroer to assess trends in failure mechanisms and their causes.
LIST OF ACRONYMS
Landing gears are designed to provide aircraft support and control when
on the ground (steering and stopping) and to provide a rethod of absorbing the
loads and zti._ associated with l 7i taxiing. Since they are not *;d
in fligh., landing gear components have traditionally been designed to meet design
limit loads using high strength-to-weight materials in order to carry out their
functions with minimum weight and use of space. From a design viewpoint, landing
gear service life is assumed to be principally limited by fatigue considerations,
the time-dependent degradation of the materials being addressed only by means of
prevention.
Despite the extensive landing gear design analyses and tests carried out
by the designers and manufacturers, and the large number of trouble-free landings
accumulated by the users, the Canadian Forces (CF), as well as others, have
experienced a range of problems or failures with landing gear systems. This paper
provides an overview of the structural component failures as experienced by the
CF within the last two decades on more than 20 different aircraft. Table 1
describes the fleet and provides information on the number of landings, which
range from 200 to 1,200 pnr year per aircraft; this information is not recorded
for helicopters.
strength aluminium alloys and steels have most commonly been selectc,J, with
magnesium and tiLtuiU alloys used for special applications only. The
requirements for high strength led to the widespread use of aluminium alloys from
the 7xxx series heat trsated to the peak strength T6 temper for structural
components and 2014-T6 fo- wheels. A variety of high strfngth low alloy steels
which could easily be heat treated to strength levels in -v'.s of 141r Mid
(200 koi) and up to 1930 MPa (280 ksi), were selected. AISI 4340 hs lbeen
commonly specified and 300M selected mo.-e recently.
Only lImiLed use of tnis system couio oe ma.3 h-u? it nrovided general
information without indication of cause or mechanism :f failures, except that
corrosion and cracking were used as generic terms to describe a failure.
0% P.R%11,0
il I
(",
i
i.,
.
mu
Fi.m . n
ait ste i r ar
IS RI IFR
IRAIER
iiis data iank was more precise than AMMIS because the descrip!:ion c" ne
ii[ iceS permitted identification of their causes in a number of zases, :Ait t-'r(
,rc no indications as to the failure mechanisms.
4,,
3 6 1417,
Ni E (>LANIIINI,'
OF11>
TRAINER PATROL
FAJILW
CTO £4;O0(;| ;
2:-
60
The SCC failures cczred in parts made from a material now well known for
its severe susceptibility to SCC. In this particular case, contributing factors
were exposure of the end grains in the hole and possibly a galvanic couple between
the alumininforging and the beryllium copper bushings in the presence of an
elecdrolyte fored by the grease contaminated with water. New parts were
manufactured using alumtnhum alloy 7049 and heat treated to the T73 temper showing
improved SCC resistance. The bushings are now cadmium plated on their outer
surfaces and a multipurpose grease is used.
3.3.w eoai landing gear axle of htransport airbraft
During a training flight landing, one of the axles of t e main landing
gear of a light transport aircraft fractured in a complex helical manner around
the lock pin hole at the bottom centre of the axle (Figs 7a and b). The 0.755 m
long axle was made from AISI H1 tool steel tube and heat treated to an ultimate
tensile strength (TS) in the 1650-1790 MPa (240-2b ki) range except for the
eds which were in the 1240-1380 MPa (180-200 ksi) range. The outside surface
was chromium plated to a thickness of 0.05 to 0.075 ma (0.002 to 0.003 in) and
the lock pin hole flash chromiu plated to a thickness of 0.0075 to 0.0125 mm
(0.0003 to 0.0005 in). The inside surface was coated with a corrosion preventive
compound to MIL-C-16173 Crade 1 specification.
Fig. 6. a: Fail2d shock strut assembly: (A) indicates the eye-bolt with beryllium
copper bushing and (B) the eye-bolt region on the strut housing.
b: The dotted line indicates the extent of the stress corrosion crack
at the origin of the failure.
Detailed examination of the broken axle revealed that the final fracture
process had originated at a very small fatigue crack, 0.5 mm (0.020 in) long by
0.1 mm (0.004 in) deep spanning about half the width of a small and relatively
shallow corrosion pit (Figs 7c and d). Other pits were present around the hole
and they likely resulted from the ingress and accumulation of moisture at the axle
locating hole and pin joint.
Fracture mechanics modelling showed that, because of the high operational
str2sses and poor damage tolerance properties of HIl steel with a mean room
temperature fracture toughness around 37 MPa4T (34 ksi4n) at the part strength
level, very small crack-like discontinuities including corrosion pits, that cannot
be reliably inspected, could cause catastrophic failure of the axle during normal
operation. The criticality of these crack-like discontinuities is further
aggravated by cold temperatures: a 10% reduction in room temperature toughness
is expected at 40"F (5"C) and a 25% reduction at -40°F (-40'C) (5].
Fig. 9. The 75 m (2.95 in) long Fig. 10. Damage to the shock pin
stress corrosion crack in the chromium plating and
lever initiated at a corrosion of the exposed
corrosion pit (arrow) in an steel.
area where the IVD aluminium
coating had disappeared.
3.3.4 Min landing gear truk beam of a large transport aircraft
A main landing gear truck beam from a large transport aircraft fractured
on the ground just after refuelling as a result of circumferential crack located
about 0.14 m (5.5 in) forward of the oleo attachment point (Fig. lla). The part
was made from 4340 steel heat treated to a UTS in the 1800-1930 MPa (260-280 ksi)
range. The part had been treated with a manganese phosphate, primed and painted
on both inner and outer surfaces. No sacrificial coating, such as cadmium, had
been applied to minimize the occurrence of hydrogen embrittlement. A corrosion
preventive compound (MIL-C-16173 Grade 1) was also applied to newly manufactured
parts for additional protection.
Visual and microscopic examination of the fracture surface revealed that
final failure had emanated from a small, nearly semicircular shaped area (12 mm
long by 8 mm deep) (0.47 by 0.31 in) with intergranular features, most likely a
manifestation of SCC (Figs llb and c). The transition from intergranular to
overload was abrupt. The stress corrosion crack had initiated at a corrosion pit
which, at the time of failure, was quite broad but shallow (10 mm by 6 mm by
0.3 mm deep) (0.39 by 0.24 by 0.01 in) (Fig. llc). Extensive corrosion in the
form of shallow pits was observed in the forward portion of the beam in the
vicinity of the fracture and particularly in an area at the inside top of the beam
(Fig. llb).
The fracture at the top of the cylinder bore could visually be determined
to have originated at a narrow fatigue band (2.5 mm (0.10 in) maximum depth by
70" arc) which had initiated at the inside radius at the top of the bore and was
centered on the forward side. The fatigue zone consisted of a 2 mm (0.08 in) deep
fatigue band followed by a narrow region of rapid crack extension and a very
narrow fatigue band (Fig. 12b). The cylinder also contained a longitudinal crack
through its wall 0.65 m (25.6 in) long from the bottom. That crack had propagated
from a 35 mm (1.38 in) long elliptically shaped stress corrosion crack (Fig. 12c)
which, along with some smaller ones, had initiated at the interface between the
7079 alloy and a flame sprayed aluminium-silicon coating which exhibited poor
adhesion in some areas. Thq stress corrosion crack was about 40" away from the
forging flash line.
551
Fig. 13. a: The 0.34 m (13.4 in) long crack emanated from a fatigue crack in the
trunnion radius.
b: 32 mm (1.26 in) long and 2 mm (0.08 in) deep critical fatigue crack.
3.3.7 Wheels
Over the years, a large number of wheels on all aircraft types have been
found cracked aend a few have failed catastrophically (Fig. 14a ). The cracks have
been discovered as a result of deflated tires and by NDT inspection. These
failures occurred after a few years in service for some and much longer for
others. Some of these wheels were made from magnesium alloys; all the others
were forged from 2014 aluminium alloy, heat treated to the T6 temper, sulphuric
acid anodized, and painted.
Many of these failures could be traced to corrosion pitting on the weather
side of the wheels. Some of the pitting was promoted by fretting from parts such
as heat shields which damaged the protective coatings. These pits, sometimes as
small a 0.2 mm (0.008 in), were followed by intergranular cracking and then by
fatigue (Fig. 14b to d). A number of other fatigue failures initiated on the
inside, mainly in the tire bead seat area, again at pits although corrosion was
less severe. In a few cases, manufacturing defects such as forging laps in the
aluminium alloy wheels and porosity in the cast magnesium alloy wheels were at
the origin of fatigue cracking.
1-12
Fig. 14. a: Catastrophic wheel failure from a 0.14 m (5.5 in) long fatigue crack
which had initiated on the weather side.
b: The arrows point to corrosion pits at the origin of a fatigue crack
(x5).
c: Pitting followed by intergranular cracking, the dotted line marking
its extent, and then fatigue (x40).
d. Transition from intergranular to fatigue cracking (x650).
Often wheels are lifed at the design stage in terms of roll distances.
In practice, at least within the CF, wheels are not considered lifed, but rather
are kept in service as long as possible and inspected periodically. In such a
situation, proper maintenance to assure the integrity of the protective coatings,
to remove corrosion and to reprotect the part becomes very important.
ab
a. overload
b. fatigue
c. pitting
d. stress corrosion cracking (including hydrogen embrittlement)
e. wear (including fretting)
f. false call, either an indication of a defect when in reality there was
none, or an indication of a defect that could be tolerated and was
usually inherent in the manufacturing process such as porosity in a
casting, or flash line in a forging.
The following categories were used to classify the failure causes:
AWL-.-
1-14
About 10% of all components investigated were in the false call category
mainly because of forging laps and porosity in castings. Those remaining were
equally split between those which were found damaged or cracked and those which
failed catastrophically. The distribution of these components by material types
showed that slightly more than 50% were made of aluminium alloys, mainly and 1
7xxx series alloys, but also 2024, 356.0 and 295.0. About 40% were made of
steels, 4340 and 300 being the most common, the other high strength steel (HSS)
being 4130, 8630, 8740, 9620, H11. The remaining were magnesium alloys (Fig. 16).
OTHERHSSft
MAGNSFU STEELS
-'qWNNNNNW 1914340
300M
SSTELS
52% sr
ALUMINIUM ALLOYS
OVERLOAD
CORROSION MECHANISMS
PITTINGG
COMPONENTS 38% OrHERMODES
WEAR.OTHER OjOA
FALSE MODES VERO'ATIADI
CALL COROIO
OVERLOAD FLECL
D PrrI[NG
FATIGUE PTI
SC CORROSION
74%__ PITTING.
OTHERMODES
Fig. 18. Failure mechanism distribution for wheels and other structural
components.
The distribution of failure mechanisms for steels, aluminium and
magnesium alloys in general and for each statistically significant alloy type
(Figs 19 to 21) indicated that:
a. Fatigue was the most important failure mechanism for all steels.
However, for 300M, SCC and corrosion, often initiated oy wear ramage
to coatings, were particularly significant
b. Corrosion was the most important failure mechanism for all aluminium
alloys, pitting being relatively important for 2014 and SCC being the
main failure mechanism for 7079-T6. Fatigue was also an important
failure mechanism, especially in 7075
c. Corrosion and fatigue were equally important failure mechanisms in the
magnesium alloys investigated; the large number of false calls resulted
from porosity in cast wheels
OVERALL 21%
O AISI 4340
-OVERLOAD
SFATIGUE
PrnNO
SCC
PINO.OTI IIERMECHANISMS
9WEAR.OTHER MECIIAMtMS
FALSECALL4
Fig. 19. Failure mechanism distribution for steel components by alloy type.
1-16
OVERALL
2%3%
(%
AA 2014
31%
AA 7075
FArAGUEm
413
efING
PrMlNC-OTHERMECHANSMS
=WEAR.OflIER NECHANISMS 4 OOVERLOAD
FALSE
CALL.
AA 7079
S16%
-3r -G<I'E
OVERLOAD
FATIGUE
Prr11NG.OTHER
MECHANISMS
The distribution of the causes of the failures indicated that they were
attributed to manufacturing defects or problems in 30% of the cases (Fig. 22).
Other causes such as design deficiencies, material selection and maintenance
practices were also important for both wheels and the other structural components.
Abnormal landings and improper directives for maintenance practices contributed
only a small percentage.
I I-
OTHER STRUCTURAL
3% 3COMPONENTS 3%
2%
3%
6%
-DESIGN
MATEIIALSELECTION
IM MANUFACURING
IIFIELD
MAINTENANCE
MAINTENANCEDItRECnIVF
=ABNORMALLANDING
3% 2%
OVERALL
3%
3% ~AISI 4340
4%
- MA11NIALSIcURn
C3MAN1JFACS1IUNC
eTeFID MAINTENANCE
M=ANTANE DmRCnvEs
ABNORMAL LANDING
OTHER HSS
Fig. 23. Failure cause distribution for steel components by alloy type.
OVERALL
2% A5%
tAA
7075
-MATERIAL.
SF61J.CflON
MM AW~FACTURING
00
tZJ
HI 1 MAIKTENANCE
MAUIVTNAN(CI
AS1W5MANN
DIRECTIVFS W
AA 7079
Fig. 24. Failure cause distribution for aluminium components by alloy type.
LANDING
II
WAM7LATIRI
.AT R.I.
SElE.TION
PITrING+OTHER MECHANISMS
li OVERLOAD
% -n FATIGLrE
DESIGN Prrn.NG MATERIAL SELECTION
SCC
C7I'n IOTIOIER ECHANISMS
CM WEAR.OTHERMECHANISMS
4% FALSECA.L
Developing a realistic load spectrum is not an easy task. One must not
only take into account the loads associated with landing as well as those
associated with ground manoeuvers (steering, braking and taxiing) as they
contribute significantly to fatigue damage (6], but also the stresses induced by
manufacturing processes and assembly. Nevertheless, designers should be striving
to achieve this goal [7]. In order to standardize the testing for comparison
purposes, it might be appropriate to define realistic fatigue load spectra as they
already exist for some aircraft structures (e.g. FALSTAFF) and some engine
components (e.g. TURBISTAN). Further, in order to satisfy airworthiness
authorities who have different interpretations of similar requirements, it would
appear appropriate to establish E single, universally applicable set of fatigue
design requirements.
Although fatigue failures have been attributed to design deficiencies in
nearly 60% of the cases, corrosion as it accounted for nearly all of the remaining
design related problems should not be neglected. The causes were many: no means
of drainage, inadequate sealing from the environment, design allowing water
entrapment, poor selection of protective coating, a design preventing access for
corrosion inspection, the use of dissimilar metals in contact and assemblies
promoting damage to the coatings. Although preventive measures can be taken for
each of these examples, there is also a need to better understand how materials
degrade with time in the chemical anc physical environment in which the landing
gear will be operating. The physical environment refers to the type of
maintenance and possible types of damage (e.g. scratches, plating deterioration)
that can be incurred during specific maintenance accivities. As mentioned
previously, actual design addresses fatigue through durability testing and
corrosion through means of prevention but it would be wiser to consider the
synergism of fatigue with the envizonment as many failures were caused by fatigue
but promoted by corrosion pitting. The ENSTAFF spectrum, which is the FALSTAFF
spectrum taking into account environmental effects for airframe, is, for example,
a positive step in this direction. A similar approach could be followed for
landing gears.
. m-
i dmS "- 5i, -
l-"_Uft
. 0
...J "idam Pim .N
.Mbi U 6." DIMbUy Um '*
-0 emelDW.b-ThW** .klp dat PW "I--"
-dft V--
*PInw Isis ~ WON 6
.-mRlm iys iide'sWa
MA P1- 4
W-
W~~ -,
N* - ft.w.7 Im Pi5.is h
is * 0.
*. I~y.d.. . 2-7h P
C- -.
"
d-W1
WW .. -AM -A i- ~l
-MAP
-Pnisdhhltyha*c
bur"
f.t."p *Dip d.,.hp.Wt
mhs
As mentioned earlier, the requirements for minimum weight and use of space
often led to trade-offs in material properties, high strength being selected over
fracture toughnesa and corrosion resistance. For example, the H-11 steel selected
for a main landing gear component of a CF transport aircraft is no longer used
in the CF following a catastrophic failure because of its poor fracture toughness.
Magnesium alloys have also been avoided by most manufacturers because of their
inherent corrosion problems. Another good example, probably the most
controversial, was the use of the 7079-T6 aluminium alloy in landing gears. This
alloy was developed in the mid-50s and although its production ceased in the early
70s because of its poor resistance to SCC, it was, unfortunately, still used in
the manufacture of landing gear components (struts, cylinders) for another decade.
Major costs were incurred because of the numerous in-service SCC problems
experienced and the need to replace landing gear components made of this alloy.
Even with alloys developed and selected more recently, such as 300N, SCC is still
occurring.
The need for materials with higher fracture toughness is now being
addressed by more stringent requirements, but the concerns about corrosion and
1-22
Much effort is now being devoted in this area and the CF is presently
sponsoring a research and development project to model corrosion processes.
Hopefully, this program will permit better rationalization of the inspection
process. The inspection cycle of many CF aircraft landing gears is presently
based on the number of landing cycles. Although this may be appropriate only when
the fatigue mechanism is considered, it is unfortunately irrelevant in preventing
corrosion damage, since it is a time-dependent process. The use of corrosion
probe monitoring may also be helpful in rationalizing the inspection cycles.
(10) Nagr, A., "Fatigue Crack Growth in Aircraft Main Landing Gear Wheels",
Fracture Mechanics: Nineteenth Symposium, ASTM STP 969, T.A. Cruse Ed.,
American Society for Testing and Materials, Philadelphia, 1988,
pp. 868-882.
(11) Forth, K.D., "Gearing Up for The Future", Aviation Equipment Maintenance
Journal, January 1990, pp. 58-61.
(12) Lifka, B.W., and Sprawls, D.O., "Stress Corrosion Testing of 7079-T6
Aluminium Alloy in Various Environments", Stress Corrosion Testing,
ASTM STP 425, American Society for Testing and Materials, 1967, p. 342.
ACINOWLEDGENII T
by
2. To compound the problem, aircraft ground activities have never generated the same excitement or
enjoyed the same interest as flight operations. Some even look upon landing gear as a necessary evil, at
beat. Consequently, the development of ground loads analysis techniques and requirements, coupled with
operational considerations and design trade-offs is not as advanced as other aspects of aircraft design.
The result of this reduced level of interest need not be catastrophic structural failure. Normal ground
operations rarely produce structural failures which kill people. A likely result of this reduced level of
interest is an excessively conservative and over built design. While anyone can recognize structural
failure, few can recognize the over designed, over weight structure. Therefore, most designers are
reluctant to cease using historical factors, even when they are no longer appropriate.
3. This does not mean that the area of ground loads has been completely ignored. What this does mean is
that due to an uneven amount of interest in different areas, some conditions are easier to tailor than
others. In order to see the application of AFGS-87221A, let's consider a current USAF fighter that was
designed before the specification tailoring concept. We will compare its actual requirements against
AFGS-87221A and the current U.S. Navy document MIL-A-8863B(AS). No, not all of the requirements, just a
selected few that pertain to landing gear: taxi, turns, pivot, landing and towing.
4. Let's first consider the actual taxi requirments for our sample aircraft, as shown below:
5.1.1.a Taxiing - The aircraft is in a three point attitude for three point and
unaymetrical braking and in a two point attitude for two point and reverse braking.
The aircraft is in equilibrium with balancing gear and intertial loads. The coefficient
of friction versus aircraft gross weight is shown on Figure 5.4.1.a-1.
For operation from a paved surface, the aircraft shall be capable of withstanding loads
from both a continuous runway profile and discrete 1.5 inch step and (1-cosine) bump and
4
depression inputs. Figure 5, .1.a-2 defines the paved runway step and bump and dip inputs.
5. We can immediately see that this references a coefficient of friction function and figure that defines
a surface roughness requirement. Without reproducing the braking coefficient of friction figure, it
varies from 0.7 to 0.8 as a function of aircraft weight. For the taxi way roughness the figure is
reproduced here.
2-2
FIGURE 5.4
. 1.a-2
BUP HEIGHT OR DIP DEPRESSION
500 7
400
300 - /
200
IOU
- 80
Z 60
-J 50
> 40
30
20
10
0 5 10 15 20 25
You can see that for a bump wave length between one landing gear separation to three times the landing
gear spread, the bump height goes from 1i inches to 3 inches. At twice the landing gear spread the bump
height is only 2 inches. This does not suggest an aircraft intended for other than prepared bases.
5. Below is the U.S. Navy requirement from MIL-A-8863B(AS), which would have been imposed on the aircraft
if it had been built for the Navy.
3.12.4 Taxiing. Applicable to all types of airplanes. The airplane shall be in the
three-point attitude. The drag loads and side loads at each gear shall be zero. The
sum of the vertical loads acting at the ground shall be equal to twice the weight of
the airplane. Separately for the design of the nose gear, and its support structure
only, the sum of the vertical loads, acting at the ground, shall be equal to three
times the weight of the airplane.
6. Next, let's consider the taxi requirement as it is in AFGS-87221A. This is deceptively simple in
appearance. While it no longer includes the braking criteria, it actually demands the engineer must have
considerable amounts of information on various aspects of aircraft taxi.
3.4.2.1 Taxi
REQUIREMENT GUIDANCE
Define the taxi requirements in terms of general parameters 3.2 and any attainable
combinations thereof. Taxiing loads shall be based on operational requirements such
as taxiway, runway, and tire conditions. Taxi loads shall be established at appropriate
speeds in accordance with 3.2.7. For example, low speed taxi on taxiways and ramps of
paved and semiprepared airfields at speeds up to the taxi limit speed, V and high speed
taxi on runways of paved and semiprepared airfields ar speeds up to the lift-off limit
speed, VL. The appropriate effects of weight, cg position, mass distribution, and
landing gear characteristics shall be included. RTD-TDR-63-4139 Vol I and ASD-TDR-62-555
Vol I provide criteria and analysis techniques for establishing alighting gear dynamic
load. Power spectral density levels for paved, semiprepared, an" unprepared airficlds
are presented in Figures 6 and 8. Discrete bumps and dips for slow and high speed taxi
are presented in Figures 4 and 5.
Reading the Requirement Guidance you will realize that not only must the designer be cognizant of the
proposed aircraft, but also the proposed operational bases and their taxiways and runways. This
brings us back to the question of well defined operational requirements. Obviously, with a two inch
bump for a wave length of twice the gear spacing, our example aircraft was never intended for semi-
prepared field operations.
2-3
7. We shall now address turning. Below, is shown the actual tuning requirement as it appears in the
system specification for our sample aircraft.
6.5.2 Turning - The air vehicle in the three point attitude shall execute steady turns
by the following means:
c. Differential braking.
The ratio of side load to vertical load shall be limited to 0.5 on any wheel and the sum
of the side loads shall equal 0.5W at maximum design weight and 0.3W at maximum limit
weight except that this value need not exceed a value which would result in overturning.
For braked conditions, the drag load and side load on the braked wheels shall be such
that the vector sum of the drag load and side load will not exceed 0.8 of the vertical
load at maximum design weight and 0.65 of the vertical load at maximum limit weight with
a linear variation between these weights. Requirement "b" above shall result in a side
load factor of at least 0.3W at maximum design weight and 0.18W at 68,000 pounds and remain
at 0.18W tc maximum limit weight with a drag load on the braked wheels of at least 0.5 of
the vertical load at maximum design weight and 0.4W at 68,000 pounds and above.
3.12.2 Turning. Applicable to all types of airplanes. The airplane shall be in the
static three-point attitude. The sum of the vertical ground loads on the landing gear
shall be equal to the weight of the airplane. The drag loads shall be zero. The side
loads on each landing gear shall act in the ground plane and in combination with the
landing gear vertical loads, such that the total re.ultant load passes through the
airplane CG. The ratio of the side load to the vertical lcad shall be the same at
each landing gear. The sum of the side loads shall be 0.5 times the weight of the
airplane, except that this sum need not exceed a value which would result in overturning.
3.12.6 Steering. Applicable to all types of airplanes. The airplane shall be in the
statuc three-point attitude with the noise gear swiveled in all possible positions. A
torque equal to the maximum available steering torque shall be applied to the nose gear.
What can we say about either or both of these? Obviously the actual requirement is noticeably more
cczplicated than in MIL-A-8863B. But, does that tell us anything? Nt, neigher specification is a
performance requirement! The coefficients of friction and the load factors tend to suggest they are the
results of an analysis which cnverted performance to specific criteria. However, there is no ways to
understand why the actual specification is more complex than MIL-A-8863B. What did this complexity buy
us?
3.4.2.2 Turns
REQUIREMENT
RATIONALE
The purpose of this requirement is to provide structural requirements for unbraked steady
turns.
REQUIREMENT GUIDANCE
Define the turn requirements in terms of general parameters of 3.2 and any attainable
combinations thereof. Turning design loads shall be based on operational requirements
such as taxiway, runway, and tire conditions. Turning requirements shall be established
at appropriate speeds of 3.2.7. For example, turns on ramps at speeds of up to the taxi
limit speed, V, on paved and semiprepared surfaces. Turns on taxiways at speeds up to
th taxi A1±t speue., vT on paved and semiprepared surfaces. Runway turn-offs at speeds
up to the taxi limit speed, V , on paved and semiprepared surfaces. The effects of weight,
cg position, mass distribution, and landing gear characteristics shall be accounted for.
REQUIREMENT
LESSONS LEARNED
A technique for establishing lateral load factors during ground turning is presented in
ASD-TR-79-5037.
Not only must the aircraft be well defined, but so must the taxiways and runways. In reality, it is
probably more important to define the aircraft speeds besides the taxiways and runways, since in a first
order snalysis the mass of the aircraft is immaterial. Again, the implementation of AFGS-87221A requires
more effirt and insight than using either the U.S. Navy specification or the actual aircraft turning
requirment.
2-4
9. Pivoting is in many ways just a very special type of turn. The actual specification requirement for
our example aircraft is given below alcng with the plot which defines the tire coefficient of friction.
5.4.1.d Pivoting - The aircraft is pivoted about one wheel. The brakes are locked on
the gear about which the aircraft is rotating. The tire coefficient of friction is defined
by Figure 5.4.1.a-1.
FIGURE 5.4.l.a-I
FATIGUE BRAKING COEFFICIENT OF FRICTION
= .8
.6
20 40 60 80
The requirement which MIL-A-8863B would have imposed on the development of this aircraft is given below.
3.12.3 Pivoting. Applicable to all types of airplanes except SKI airplanes. With
brakes locked on the landing gear unit about which the airplane is rotating, the airplane
shall pivot about one wheel, or in the base of multiple wheels, about the centroid of
contact area of all wheels in the gear unit. The vertical load factor at the CG shall
be 1.0 and the tire coefficient of friction shall be 0.8.
Obviously there is a great similarify between these two requirements. However, the Air Force allowed a
linear reduction in the coefficient of friction beginning at 53,000 pounds and becoming 0.7 at 68.000
pounds. This means that this aircraft, designed to Air Force requirements, has slightly less pivoting
capability at higher weights than it would have, had it been designed to MIL-A-8863B.
10. Nov, let's see how AFGS-87221A addresses pivoting. Looking at the requirements as presented, there is
a noticeable difference. We can see that the pivot points must be defined along with thrust levels. Both
the sample Air Force requirement and the current U.S. Navy specification assume the point of pivot is known
or obvious. Also, AFGS-87221A requests thrust levels when neither of the other two do. However, both the
other two specifications give a tire coefficient of friction when the approach of AFGS-87221 does not. The
AFGS-g7221A simply presents a different approach, but this new approach does need power and thrust levels
established by a rationale process.
3.4.2.3 Pivots.
RATIONALE (3.4.2.3)
REQUIREMENT
The purpose of this requirement is to establish maximum torsional load on the main
landing gear.
REQUIREMENT GUIDANCE
If the requirement is not applicable for pivots, insert N/A in all blanks including
those blanks of the subparagraphs. If applicable, insert APP in the blank and define the
extent of applicability. For each applicable subparagraph define the pivoting requirements
in terms of the general parameters of 3.2 and any appropriate combination thereof. For
example, the pivot points are about one main landing gear wheel with brakes locked, or in
the case of multiple wheel gear units, about the centroid of contact area of all wheels
in the gear unit. The power and thrust levels shall be based on a rational analysis to
determine power required to perform the maneuver. The coefficient of friction between
the tires and ground shall be 0.8 and the vertical load factor at the c.g. shall be 1.0.
2-5
Some aircraft configurations, such as a very large transport, preclude true pivot turns,
in which cases a minimum radius turn should be defined in 3.4.2.2 instead of pivoting.
11. Landing as a loading condition requires extensive definition. Additionally the U.S. Navy has a
requirement to be able ro land on aircraft carriers. This adds at least an order of magnitude to the
complexity of their specifications and makes any real attempt to compare requirements difficult.
Therefore, let's just consider landing sink rates in order to get some feeling of the difference. The
figure below is the actual landing sink rate requirement for our sample aircraft.
6
z
30 40 50 ,) 80
Below 35,000 pounds the aircraft can land with a vertical sink rate of 10 feet per second. About 35,000
pounds the sink rate is gradually reduced in order to account for the energy increase due to the added
weight of the aircraft. It is important to remember that this aircraft's maximum sink rate is 10 feet per
second since the same aircraft when designed by the U.b. Navy would have a sink rate of over 20 feet per
second.
12. The requirement from AFGS-87221A, as shown below, is intended to be tailored to any operational
requirement. It can be made to cover all operational landing conditions, while accepting that not all
aircraft need have all landing capabilities.
3.4.2.6 Landings
d. Arrestment ( ).
e. Declerating devices ( ).
REQUIREMENT RATIONALE
The purpose of this requirement is to establish structural requireents for landing opetations
on specified surfaces.
Following the conceptof a rationally tailored system specific structured specification, the Air Force Guide
Specification approach allows the selection of appropriate ondltions. However, what are appropriate
conditions? Sink rates in excess of twenty feet per second might be appropriate for carrier landings or
air assault situations, but what is appropriate for a fighter using a main operating base? I don't know.
Not until the intended user and the design engineers review the proposed operational requirements and all
the trade-offs and interactive design impacts can the totality of the landing requirement be established.
Not only must numerous aircraft parameters be defined, but so must the runway be defined in great detail.
13. The new AFGS-87221A recommends that towing conditions be tailored as it does all other structural
loadings. However, it does suggest as a safe selection the very requirements that have been used for
years and are still current in MIL-A-8863B(AS). These are the very towing requirements that our sample
aircraft used. I don't know how old these towing requirements are, but they have been found in a 1940
2-6
document. Much the same can be said for jacking and hoisting. These requirements are historical, and
their basis or supporting justification has been lost over time. It is especially true in the area of
towing besides Jacking and hoisting that we lack research to assist in developing the well tailored,
system specific, structural specification.
14. With the advent of AFGS-87221A there is a commitment to tailoring the actual system specific
structural specification. However, those of us who use this document must increase our depth of actlvity
in several areas. First, we must work very closely with the final user of our product. Second, those of
us who use AFGS-87221A must realize that specification development is going to be more complicated than
it has been in the past. Due to conflicting operational requirements, it may be necessary to perform
trade studies which appear to be very clone to preliminary design concept studies, in order to convert
performance requirements. The comfort and supposed safety of the old, rigid specifications is over.
Besides, the old approach may have been safe with respect to contracting laws, but they often fail the
laws of physics. Therefore, our old criteria was not as uniformly conservative as we thought.
ETUDE COMPARATIVE DES NORMES F'RANCATSES AIR 2004 ET
AMERICAINES MIL-SPEC RELATIVES AUX CHARGES SUR LES
ATTERRISSEURS
par
J.M. DAUPRANT
DGA/DCA/Service Technique des Programmes Aeronautiques
4 avenue de la Porte d'Issy
75015 PARTS
FRANCE,
RESUMEZ
L',tude comparative des normes fran(;aises (AIR 2004 D / AIR 2004 E) et america'ner (MTL-A-
8862A / MIL-A-8863A) dont fait l'objet cette presentation montre, a parts ir av:vr.
embarqu& (SUPER ETENDARD) et d'un avion terrestre (MIRAGE 2000), !'influence de cosn -rec:
et les repercussions liles a l'application de celles-ci sur les cas de charges, socr
efforts au sol et sur le dimensionnement des trains d'atterrissaqo.
INTRODUCTION
L'6volution des normes militaires relatives aux charges sur arrerrisseurs ost cn~c
depuis on demi-siecle.
C'eat dana cette optique que le STPA (Service Technique des Programmes Aeronautiqu-os,
organisme de ia DGA (D616&gation Gtn~ale pou~r 1'Armnement) au seir, do M.ini.,-r- 1, a1
D~fense franCais, a command6 A la soci~t6 MESS IEP-RUGATTI1, avec le concours -le la soces,
DASSAULT-AVIATION, one e tude comparative des normes francaises et americaines recerdes.
L'6tude demand~e a'inscrit dana le cadre d'une r~Ilexion globale en vue d'une posrikl
revision de la norme actuelle AIR 2004 E.
Cette norme n~cessite en effet des modifications p~or prennire en compte I(5.
caract&istiquea des avions de combat dita de nouvelle gen~ration (en particulier a caise
des commandes de vols 6lectriques).
Lea normes retenoes poor cette estude sont lea normes francaises AIR 2004 D, AIR 2004 F et
lea normes am~ricaines MIL-A-8862A, MIL-A-8863A.
beux aviona - on avson manin: le SUPER ETENDARD et on avion terrestre: le MIRAGE 2000 -
ont s ervi A 6valuer I ,influence des diff,&rentes normes sur lea cas de charges, st lea
efforts ao sol et sot le dimenslonnement des atterrisseurs.
1 PRESENTATrION DES NORKdES
Fiqnre 1
L'61aboration de la norme AIR 2004 E est doe A la conjonction ide plusieurs facteurs
-les mutations technologiques et le dfiveloppement amorcef d~s le debut des ann'.es 1960
des moyens de calcul et de mod~iisation tels que les 416ments finis, ainsi que le
d~veloppement des moyens de simulation et d'essai,
lexperivanca acculse bar .0 0 >er:&5 :,ri e153 .0 :~tils r~ aou ra er s de5
a-ens Ct de s ptr o-.-ss Cr
eni sI o- -CIv-''
OSenle 1q9remen ts r isen des5 prcqrammet n"at i-'rsaux a't T nt Mlra- x , e 1.-v
t i j e cncpela' :on
A Ca n s'-e ''"
n-'5
v'a In V!Iem 1"vei at un e i or ae Sdelt' at 1
-ef I-a''I T,, a eY a"e
5 tint-
MIS-A-A (0-.-
paA u I ctecr-rl.Ia pat ins ller!n-vts ciisrA 5,,
i -deareilasa e I atsslm
EFn/s
en I - ala-e 96-
3,5
'j2FSF DE.
-!F. iXREM,. . c cM l~
d1 m irc
b'ie'renteridul, le z .pproch-mnrt ens me narmes nlens pan tc5tides I:Iftne e 93st en'
sibsist~om't ma igr6 scoit.-One il lust ration est deonre ear In sal-n-au pree'- en F;
r C asi Sttrmes AIR 2X4 E "t MIL-A-8862A.
Sneis .- tnon
adep m'aters _____ _____
Linag Ac IA n 1,31 A
Figuz-a 3
-- '5. c:-r'o ne_-'ec',-aine k.n -ac '! :';- !a <reAIR 27 4 K p- late - I!r
33:
'"1±'-~
t'r''td acMemie3vc'e e des ado a<-na 03.,v<
cot-v l'm-mb- c 123 P-'.'eM '.t ie P'- ''-,' c- a a sclon5'asle dicmaine 'in's
tc-553C5 a:
- -ix, :as-es
naaatic -orlale jdatenr:35 aa",
- can do delcdI,)JaqC A
riets~tcsndu cctro til rnormale (can dn- rotat ion as,
dnc'luq -m eoeft iccler~lt* 1,7 donnans Ia "harq.- vet ica le 6 appl iques en s-atiqn-
sr naque atsterr iSn-iur ( 1,7',s) n-st A revolt A Ia haussle).,
pr ise en cempr e eVerstule. dos 4volat ions au si~ (notammens silr P stes sommairement
reparn-sl
-prise n-v -mprfe 1il 'c Asdr avions matins parndli tqlt'm955 sat ional adapt#.
3-4
L'e -lution des reglements aux U.S.A. est importante tant dans le nombre des normes
redigees que dans les changements apportes.
La force )EIL-A-8862 A relative aux avions terrestres succede par exempie en 1971 A la
norme NI A-8862 ASG; elle se distingue de celle-ci par l'expression d'une valeur de
vitesse de chute extr~me et une approche plus complete des cas de virage et de roulement au
301.
Toutefois, c'ost la norme NIL-A-8863 ASG datant de 1960 et relative aux avions embarqu6s
qui illustre avec plus de norrer6 l'importaflce des changocents enregistres.
C'est avec olle en effet quest introduit le concept de combinaison cultivariable. GrAce A
ce concept, les can de calcul d l'attorrissage ne sont plus definis de rsaniere figee rnais
soot selectionn~s sot Ia base de distributions statistiques de diff~rents parametres
d'atterrissage.
Ainsi !a not,,- 3 T-A-8863 ASG pr~sente les conditions initiales d'impact sous !a forme
d'une cocbinaison de la vitesse de chute, do !'angle de roulis et de la vitesse
d'engagement satisfaisant l'6quation dun ellipsoide
2 2
a(Vv - VVm) + b(TETA -22± C(VE - VEm ) = 1
To- be=
beW Mefl~I be P919eW6 ". Pee..6 e 9~
d- V4.6 3mft 1.22 2l.*1e 1. .t. 11.2040 t W l.A.Ift
V"e en Ca CS6L3. OM. ian vrO U noM.
TA VM-b I.nv 1 LO4y I lY LU
0. LG__AA A.M. Ai b0S _____
am W, 4.0 S.0 I.0 6.4 9.0
~,
we of 0 .1 01
A..7. 2.
_ _ _ __ _ _ _ _ 1 _#
Le 3. -16 .
Figuro 4
11 COMPA"AIBON
Les normes AIR 2004 D, AIR 2004 E, MIL-A-8862A et MIL-A-8863A pr6se'-t6es succinctement dans
le paragraphe pr6c6dent sont appliqu6es au SUPER ETENDARD et au MI?2'GE 2000 afin dl~valuer
leur influence sur les cas de charges, les efforts resultants et sur le dimensionnement et
la masse des atterrisseurs.
L'accent est mis plus sptcialement dans cette s tude sur le MTRAGE 2000 et sur le concept
multivariable developpe dans la norme MIL-A-t,863A.
La norme MIL-A-8863A examine la resistance d'un avion embarqu6 pour les types
d'atterrissages ci-apr~s:
Elle permet de balayer un grand nombre de cas et d'isoler ceux susceptibles d~ctre A
Ilorigine de probl mes structuraux.
L'atterrissage "Pos6-d~collA et arrdt dans les brins sur porte-avions" conduit pour les cas
les plus s6vdres A une valeur de vitesse din chute maximale de 7,64 in/s dans des
configurations d'atterrissage cabr6 (centrage arri~lre) , d'atterrissage trois po-nts
(centrage avant et angle de roulis 6gal 1 0" et 2*) et pour on des cas multivariables.
En revanche, les cas de charges en ce qui concerne les atterrissages de type II et III sont
momns s~v~res puisque les vitesses de chute maximales obtenues sont &gales respectivement A
5,50 rn/s et 3,05 m/s.
L'analyse des cas de pr6sentation fait donc ressortir une diff~rence importante (30%) ent-
la valeur de vitesse de chute la plus 6lev~e obtenue avec la norme MIL-A-8863A (7,64 m/s)
et la valeor retenue dans les Clauses Techniques pour le dimensionnement des atterrisseurs
du Super-Etendard qui est de 5,50 m/s 1.
Il est probable qu'un tel 6cart se traduise pa des efforts importants au niveao do train
principal pour Ilatterrissage cabr6 (centrage arri6re) et des efforts importants au niveso
du train auxiliaire pour Ilatterrissage trois points (centrage avant).
Bien quoascon caicul dlefforts au sol suivant la norme 141L-A-8863A n'ait en fait 6t6 men6
avec ces cas de charges, il est raisonnable din penser que l'application de cette norme
rernette en cause le dimensionnement des trains d'atterrissage du SUPER ETENOARO.
11-2 AVION_2XA3TRX
11-2.1 ~~d..cbr~
La norme MIL-A-8862A impose l'6tude des atterrissages pour une vitesse de chute linrite din
3,05 rn/s.
La norme MIL-A-8863A prend 6galernent en corspte Ia r~sistance d'un avion de type avion
tertestre. Les atterrissages d~finis dans la norrne sont les atterrissages arrondis sor
pistes pr~par6es et atterrissages simul~s par essais en laboratoire.
Une analyse des cas de presentation, analogue A ceile faite pour le SUPER ETENDARDO est
rdalis,6e pour le MIRAGE 2000 et donne les rksultats suivants.
1 cette velour et issue dun compromis entre Ia norms anglaise Av.P.970 (Design Atquirements for Aircraft
for the ROYALAIR FORCE and ROYALNAvY( at is norm AIR 2004 D.
Los cas diewsionnants sont le3 suivants :appontage trois points, pentance 6quilibrant leg deux-tiers du
goid. pour lo train da-tterrissaqe auiilre I appontage sut eas tterrisseurS princisux queue bage St
queue haute pour I. train d'atterriaaage principal)
3-6
a) Train principal
Les tableaux at las graphiques pr~sent~s en Fig.5., Fiq.6e/6b at Fig.7 ralatifs au train
principal du MIRAGE 2000 em~nent divarses ramarquas.
La norma MIL-A--8863A donna un effort en cours d'enfoncement at des efforts de misa en
rotation at de ratour 6lastique comparables A ceux obtanus avac la norma AIR 2004 0. La
diff~ranca la plus significative concarna lleffort au sol maximal (9%), l'effort lat~ral
dans la cas de rip6 int~rieur (21%) at Ileffort vertical dens le cas do ripe ext~rieur
(11%)
La norma MIL-A-8862A est globalemant l6g~rement plus s~v~ra qua las normes AIR 2004 D at
MIL-A-8863A, cas de rip6 mis A part.
La norma AIR 2004 E est la plus s~v~re des normas. Dans le cas du ratour 6lastique at de la
mise en rotation, las efforts au sol d~passant de 29 % las efforts correspondents, celcul~s
A partir de la norma AIR 2004 0; dens Is cas de ripe ext~rieur, l'6cert est consid~rable
puisqu'il est do 68%.
Figure 5
3-7
~0 0
00
S C -,1
ww
-00
m
C4 0
C,!
0 0 0
41
i-,oo
010 N1
3-8
b) Train auxiliaire
Les calculs ont trait dans cette partie au train auxiliaire du MIRAGE 2000 avec application
des normes AIR 2004D et MIL-A-8863A. Les r~sultats sont int~ressants.
Les cas de rip6 sont tr~s s~v~res; uls Se distinguent par des efforts trois fois plus
importants que ceux d~finis A partir de la norroc AIR 2004 D.
F---D -E--
I UIAIRE
Figure 8
i ORMED APPLIQUES
2~04 D1 ML-A-.83
Ix -28.3 -3681
HIDE EN ROTATION yl 0 0
Fz 35,4 4 03L
306 3
EFFORTS 230
RETOUR ELASTIQUE 1Fa! 0 1 0
AUJSOL IFzo 37,2 47.0
en kN ________________________
IFL 0
intarieur -13. 3 4-7 I 4
52 L..........
RIPS 10.2 1 385> L.......IL...
Iext~rleur F 5 3 1a4
__,_ I F 10!2 385 ON
Figure 9a
3-9
ATTERRISSEUR
AUXILIAIRE DU
MIRAGE 2000 .B j
Int6rieur
,y RIPE
ry
r RETOURELASTIOUE
F.
ryMISE EN ROTATION
~ ~ EFFORTAUl SOL.(kN)
-0 -20 0 20 40 a0
Figure 9b
11-2.3 firensionnement
Le dimensionnement de llatterrisseur principal du MIRAGE 2000 a 6t6 r~alis6 via A vis des
cha rges obtenues par application des normes AIR 2004 0 et MIL-A-8862A. Ce dimensionnenent
na pas tenu colopte du cas "hors norme" de pointe d'effort au d~collage.
MASSE
DE L'ATTZRRISEUR on kg 9 0.258 90,939
0.80%
Figure 10
3-10
Le bilan de masse global qui met en 6vidence une augmentation de poids de l'atterrisseur de
0,8% en "d~faveur" de Ia norme MIL-A-8862A (i.e. cette norme conduit A un train plus lourd)
est compl~t6 par une analyse pour cheque 616ment constituent le train d'atterrissege.
Celui-ci r~v~le des diff4§rences sensibles en fonction de la pertie consid~r~e (Fig.11).
ATrERRISSEUR BILANGLOBAL
PRINCIPAL DU MIRAGE
2000 0 ,U1 kg
+ 0.s%
APPLICATION DE LA
NORME UIL-A-8962A
VERIN CONThEFICHE
CAISSON+ AXES
D'ARTICULATION
COMPAS+ AXE
AMOR'ISSEUR
TIGE COULISSANTE
FUISEE
Figure 11
Compte-tenu de ce bilan de masse, et dans la mesure oioles efforts d~termin~s A partir des
normes MIL-A-8863A et AIR 2004 D sont proches, il est raisonnable de penser que le
dimensionnement actuel du train d'atterrissage principal du MIRAGE 2000 ne devreit pas
subir de modifications majeures par application de la norme MIL-A-8863A, et ceci m~ine en
cas de variation de conditionnement nominal de l'atterrisseur (pression d'air et volume
d'huile de l'amortisseur, pression des pneumatiques) qulimpose Isanorme em~ricaine.
En revanche, il eat clair, a la vue des r~sultats pr~sent~s au paregraphe 11-2.2, que le
dImensionnement actuel du train deatterrissage auxilieire du MIRA'CE 2000 serait remis en
cause par Papplication de la norme MIL-A-8863A.
3-I
CONCLUSION
L'6tude comparative command~e par le Service Te. inique des Programmes A~ronautiques, dont
rendait compte cette presentation, permet de .nettre en evidence un certain nombre de
points important s.
Le second point int~ressant porte Sur tes norrsas MIL-A-8862A at AIR 2004 0 et concerne plus
part iculi~rement l'influenca de ces normes Sur le bilan de masse total de J'atterrisseur
principal do MIRAGE 2000.
L'6cart de poids antre l'atterrisseur principal du MIRAGE 2000 dimensionn6 d'apr~s la norme
AIR 2004 0 at lS meme atterrisseur dimensionn6 d'apr~s la norme MIL-A-8862A est faible. Las
r~percussions sent doec relativement peu importantes du point de vue du bilan de masse
total,
L'intrdt qua rev~t le concept multivariabla contenu dans la norme MIL-A-8863A at conserv6
dans la norma MIL-A-8863B doit dtre soulign&. 11 constitue le troisi( me point clef de
1' Atuda.
Ca concept permet one description plus souple des cas d'atterrissage puisque las conditions
initiales dimpacts ne sent plus figoes minis sent donn~es snus ha forina d'une combinaison
multivariabla de plosieurs param~tres. En fait, il offra surtout l'avantage de d~finir las
cas d'atterrissagas de fagon plus rationnalla at plus proche des situations r~ellas.
ANNEXE
4-J
SUMMARY
During the different phases of development from feasibility studies to the final
design, the landing gear designer applies increasingly refined methods of analysis to
derive the loads to which the system is designed. Future design procedures should
reflect such a staged approach leading to designs which are fully optimised with the
aid of rational methods of analysis to meet the complete range of aircraft operating
conditions.
Finally areas of work are identified which need addressing further, in order that
a staged approach can be adopted completely in the future military landing gear design
procedures.
1. INTRODUCTION
A staged approach for landing gear design, which runs in parallel and in phase
with that used for the aircraft is required in order to achieve an optimum design for
the complete system. In the Feasibility Stage, the aircraft design criteria must be
formulated, the basic design ata for the aning gear determined, and possible
configurations of the landing gear studied. At this stage a rapid method of deriving
initial design loads on an empirical and an arbitrary basis is required. The second
stage, or Initial Design, requires firmed up data and should utilise a 'rational' form
of analysis to examine in detail the design load conditions which are critical for the
landing gear and aircraft attachment structure. Rational analysis enables actual
optimisation of the shock absorber characteristics to be made at this time. In the
third stage, or Final Design phase, the structural design and shock absorber
characteristics are fulTy optimised using a refined rational analysis taking account of
the full range of design, fatigue and extreme conditions for the landing gear as well
as the realistic operating requirements of the aircraft.
2. DESIGN PROCESS
A staged design process for the landing gear in parallel and in phase with that
used 'or the aircraft is required to achieve an optimum total system design. The
design process breaks down into three discrete stages, Feasibility Studies, Initial
Design and Final Design.
In the first stage of the design process the following areas of activity must be
addressed:-
compilation of Design Criteria consistent with the operational requirements of the
Aircraft (short field, repaired runway etc.)
4-2
* identification of devices which may influence landing gear design loads (arrestor
system, take-off assistance etc.)
* Update of design data for the aircraft (mass, centre of gravity, aerodynamics
etc.) and for the landing gear (mass, stiffness, wheel tyre and brake data et,.).
Use of dynamic analysis to derive design loads by rational rather than arbitrary
methods especially for conditions leading to critical design loads for landing
gear or aircraft structure.
iOptimisation
of landing gear shock absorber characteristics and structural
stiffness to produce a balanced set of design loadings for the gear and aircraft
attachment structure including both limit and ultimate conditions.
During this phase the envelope of design conditions is arrived at using the
'firmed up' information for aircraft and landing gear configuration. A 'rational'
analysis method is used rather than an arbitrary approach particularly for those
conditions which lead to critical design cases for the landing gear or aircraft
structure.
The last phase of the design process includes the following activities:-
- for landing, permissible combinations of aircraft weight, sink rate roll, yaw,
etc.
Finalisation of fatigue operating conditions (aircraft weight, sink rate and pitch
rate distribution, etc.).
F
4-3
In this final phasP of design, in addition to ensuring that the design is fully
optimised to best meet the requirements, the aircraft operational envelope should be
fully explored theoretically prior to subsequent practical demonstration by rig, ground
and flight testing as hardware becomes available.
To il.ustrate the extent to which improved 'rational' methods for the calculation
of landing gear design loads may be applied and to make comparisons with the
'arbitrary' cases of current requ-ements, a number of examples are considered below
which relate to typical fighter type aircraft and reflect experience gathered from a
number of recent projects. The examples cover the range of arhitrary, design condltlons
such as landing, taxiing, braking, etc. and where possible direct comparisons are made
between the arbitrary and rational approaches.
This can be done by modelling the complete aircraft as a six degree of freedom
system, with fully representative aerodynamics and setting up the appropriate touchdown
-cnditon as a trimmed or Lalanced condition with lift equal to weight, zero pitch
-t t
t. "he compl"' e aircraft model can then be used to generate in a 'rational'
manner the conditions correspo-Ji ng to specific 'arbitrary' cases. The landing gear
system must also be modelled with proper representation of the shock absorber spring,
damping and friction components, the flexible structure of the gear and attachments,
the wheel and tyre assembly dynamics and the wheel spin-up process under the influence
of a slip dependent tyre/ground friction coefficient which gives rise to drag and
spring-back forces usually critical for the design of the complete system.
A number of features emerge from using the rational approach for the landing
tcuchdown which should be taken into account in reviewing the validity of current
arbitrary design conditions.
i) Today's fighLor aircraft touchdown at relatively high nose-up angles and the
'three-point' attitude of the arbitrary cases is an abnormal condition.
1i) Time varying lift forces doe to the sudden decrease in the angle of incidence
caused by rapid sink rate reduction, to aircraft pitch rotation and to lift
dumping devices during the impact phase can significantly increase the energy to
be absorbed by the landing gear compared with the arbitrary consideration of lift
equal to aircraft weight during the whole landing impact phase. (See Figure 1)
iii) Spin-up and spring-back forces should be derived usitc a representative mooel of
the flexible gear and support structure together with a szip dependent tyre to
ground friction relationship with a maximum value of 0.8. (See vigures 2 and 3).
iv) Tyre side forces induced by side slip due to aircraft yaw, roll or lateral
velocity or by deflection of the landing gear can lead to design load combinations
more severe than those derived from the arbitrary design conditions. These are
often more severe on the second gear to touchdown in an asymmetric landing. (See
Figures 4, 5 and 6).
This is the condition applying at nosegear impact with the ground. As already
indicated the normal condition for most modern fighter aircraft is for the main gears
to touchdown with the aircraft in a nose-up attitude, this is followed by a period
during which the aircraft pitches under the influence of aerodynamic and ground forces
until the nosegear impacts the ground. This period of pitching onto the nosegear can
be long compared with that of maingear energy absorption and the influences of the
changing pitch attitude and aircraft lift are therefore correspondingly greater on the
nosegear energy absorption requirements. Additionally during this interim period
between maingear and nosegear impact landing procedures such as pilot selected or
4-4
automatic de-rotation, drag chute deployment or reverse thrust selectors can play an
even mor, influential role. All these events must be properly modelled to arrive at
the initial conditions and time varying forces and moments required to rationally
predict nosegear design loads.
From the use of a rational approach to the analysis of the nosegear impact the
following points emerge which are considered relevant to a review of the arbjtrar1
defined requirements.
ii) The lift at nose landing gear impact is probably near to zero through a
combination of the instantaneous aircraft attitude and the fully deployed lift
dumping devices.
iii) Spin-up and spring-back forces should be derived using a representative model of
the landing gear and support structure together with a slip dependent tyre to
ground friction relationship with a maximum value of 0.8.
iv) For nosegear design asymmetric landing conditions are generally of much less
significance than the effect of de-rotation procedures referred to above.
v) Optimisation of the shock absorber characteristics should be made using both the
most critical of the rationally derived de-rotation conditions and the typical
fatigue conditions and should aim to produce a balanced set of design cases,
principally for vertical and drag loads in typical, design and extreme landings.
This is the condition covering the decelerating transition from landing touchdown
and de-rotation to steady taxiing and is likely to affect nosegear loads particularly.
Any retardation devices such as drag chutes, thrust reversers or mainwheel brakes need
to be included in a rational analysis of this phase with particular attention paid to
their time dependence, both their sequence and rate of application. (See Figure 8).
The following are likely to be of significant influence to the landing gear loads.
This condition represents the accelerating take-off run and should include the
sequence of events from brake release to take-off rotation.
The phase of the take-off run most likely to benefit from use of a rational
analysis approach seems to be at the point of aircraft rotation, where aerodynamic
fo-ces can lead to significant incremental maingear loads being developed. These could
b~come critical for landing gear design if taken in combination with runway obstacle
traversing considered in section 3.6 below.
The various existing landing gear requirements include design cases for a range of
ground manoeuvring and handling conditions which evolve from a vast amount of
experience and have been shown to be realistic by application on many different types
of aircratt. Of the many conditions covered two are worthy of further consideration in
the context of a rational method of analysis.
Firstly, the condition of sudden braking which needs not necessarily occur in the
landing roll out phase (3.3 above) but could be associated with very slow taxi speeds.
Adverse phasing of the aircraft pitch frequency and the time to develop peak brake
torque can lead to transient loads on the nosegear in excess of those derived from the
4
4-5
Secondly, for the turning condition arbitrary requirements are based on a latera.
acceleration level of 0.5 g which can lead - landing gear desng cormitors
particularly for multi-wheel configuration maingears on large airc aft. it is
suggested that use of an ai.-!'_ p-p~rly accounts for the taxiing nanoeu;vye
and the rational distribution of tyre forces on multi-wheeled gears consistent wto
the associated roll and yaw, could lead to a reduction in severity of design 1bade for
this case.
Single and multiple repairs with realistic variations of repair height, repair
spacing and aircraft speed. (See Figure 91.
Super-position of single and adversely spaced repairs with steady state ocdbtivue
such as braking or reverse thrust.
Super-position of single and adversely spaced tepaits with landing and de-rotatio:
phases using typical values of significant parameters such as sink rate and pltc.
rate. (See Figure 10).
Super-position of single and multiple repairs with the extreme cases from the
various phases in 3.1 to 3.5 above.
ii) Superimposing repair capability on top of every existing design conoition will
severely penalise the landing gear design.
iii) The most damaging design conditions result from the super-position of repair
requirements and landing, de-rotation, or take-off rotation phases which produce
particularly severe increments in vertical and drag (spin-up and spring-backt load
combinattons.
The preceding section 3. shows that existing methods of rational analysis cal, be
applied to the various design conditions of the arbitrary method of deriving landing
gear design loads to illustrate the differences between the two procedures. A fuither
extension to the use of the rational approach is to examine the basis of the typical,
design and extreme conditions used to determine fatigue, limit and ultimate loads
respectively.
lateral drtft case derived from the maximum vertical load. Together wits other
non-landing conditions covering braking and turning these loads determine the landing
gear basic strength level.
In a rational analysis values of parameters other than those specified for the
design conditions can be examined to determine their influence on gear loads and maybe
for their restrictive influence on aircraft performance, where the rationally derived
loads exceed gear strength capability. Such use of rational analysis has identified at
least two areas whcie further examination is suggested.
First is the case for inclusion of asymmetric landing conditions in current design
requirements. It has been found that introducing typical values of roll and yaw into a
rational analysis of the landing condition, which already includes effects such as lift
reduction due to aircraft rotation, etc., can lead to sideloads in excess of the
arbitrary design levels. Further the reduction in sink rate required to compensate for
these asymmetries can be quite restrictive on the aircraft's operating envelope (sink
rate reduction 3.7 m/s to 3.2 m/s).
Increasing the roll and yaw values to correspond to maximum crosswind landing
levels can lead to even greater restrictions on the operating envelope (typically
3.2 m/s to 2.6 m/s), a level which ma- be incompatible with aircraft short field
approach procedures.
Secondly, it can be shown that the loads resulting from rational analysis wi'h
parameter values appropriate to extreme (or ultimate) conditions are generally below
the level of the 1.5 ultimate factor specified in some requirements. This suggests
that ultimate strength factors could be reduced where it can be shown that predicted
loads for extreme conditions are loss than 1.5 times limit loads.
Fhe benefits of a design approach involving reduced utsmae attongth factors and
the consequent increased risk of structural failure hould be assessed against the
alternative of designing to ensure continued functioning of ti-e sheok absoitbe at
extreme conditiorns as in some helicopter requirements.
. C IONS
lNCUS
fhe
MeI e-l for a rapid method of derivsng initial design locs on an bIt ary'
cas' s wt I cont inue to exist ani conssitlcat ion should t heref" - be gIvn to the
%- ilt y af 'he currently d'ftIt- landig gear I sgn cases nt aome ieqI'ti enr'
ht'2l ,mpar sons made In soct Ion 3 It ighl ight some differences in I andInr iat
lesign I ads derived ft r, t he ' rrb i f rary' met hod of en IsIng reqa teeen, ld
'rational' analysis method takinq account of all the oerant parameter .ta r tho
variou design condit tans.
i The vtlue of t 'rati ntl' tithd of al; yIs I vryt clv--i hotih eItrtmitI
landitng gear -isign I -,,c and aircraft rpet rins envelopes, and con dirat I i
should te given to mu tog this a more rerial e.41) Iement for landing g-,t dena.
DU L'ci 0
Dl 0
0l 0 u
CCz L
1 U 4
00
z z
L Z0
wz 0 Q0
0
> cc
LLo 4 > 0
U- -
W6 o
21-9
0
T-
w O 0 0 f0
0<L
->z -J
F
4-8
250-
MAINGEAR
FORCES o ,PNU
AT AXLE F-N
(kN)/ /'
0 -
-5-OUTBOARD/ j
150
SPRING BACK
0 003 0.30 009 0.12 0.15 0.18 021 024 0.27 0.30
TIME (SEC)
1. SPIN-UP 84 86
2. OUTBOARD SIDE FORCE 66 55
3. INBOARD SIDEFORCE 88 67
4. SPRING-BACK 135 135
5. VERTICAL 220 220
FIGURE 3
4-9
RATIONAL ANALYSIS
ASYMMETRIC LANDING TIME HISTORY
200-
150- STARBOARD
MAI NGEAR
VERTICAL
LOADS (kN)
50
RATIONAL ANALYSIS
ASYMMETRIC LANDING TIME HISTORY
200-
100-
PORT F
MAI NGEAR/ \/x / - -
LOADS (MN
RATIONAL ANALYSIS
ASYMMETRIC LANDING MAINGEAR LOAD SIGNATURE
YAXS=AS20
06,w 00o
OUTB./INB.
z w-
* z~fl -
02 ox0
o d
z LL 3: w
WEL - -- 4-
-- _-j-_- z-__
00
0 a
10 0 -
< <<
cc0
I-III
o / zX<
0 C
4-12
MAINGEAR
LOAD
DISTANCE
FIGURE 9
NOSEGEAR MAINGEAR
REPAIR REPAIR
IMPACT IMPACT
MAINGEAR
VERTICAL NOSEGEAR
LOADSN,
TIME
FIUR 10
5-1
SUMMARY
Under Project 'Nave Bounce' (HE), the USAF successfully determined the level of
surface roughness that could be tolerated by most aircraft in the inventory. The
runway roughness capability of each aircraft was determined by developing a
sophisticated computer model of each aircraft. In most cases these computer models
were validated with aircraft tests on rapidly repaired runways. This multimillion
dollar effort has resulted in the definition of surface roughness criteria (repair
criteria) for each aircraft. The knowledge gained as a result of all of the testing
and computer modeling has led to a much more thorough understanding of the complex
interaction between the flexible structurelanding gear, and rough pavement.
The first part of this paper discusses the development, laboratory qualification
testing, and taxi testing of an improved F-15 rough field landing gear which provides
a significant improvement in rough field and sink rate performance over the existing
F-15 landing gear. This landing gear design utilizes passive, internal strut
modifications to achieve this performance without any effect on reliability and
maintainability. The second part of the paper discusses the advantages of an
automated Personal Computer (PC) based process for selecting the minimum operating
strip (MOS) and for determining the minimum level of runway repair required. As
backup to this automated approach, a novel technique for quantifying the ability of a
given aircraft to traverse rough surfaces is also discussed. This method assigns a
'Vulnerability Index' (VI) to each aircraft. The VI is a reflection of the
aircraft's abIlitv +t 'bzorb he energy that is transmitted from the pavement to the
struts and the airframe. This method will give the base commander a tool for making
good "intuitive' decisions 4n the event that the automated process cannot be used.
It can also be used as a validation technique for the automated method.
Landing gear designs of the past 40 years have stressed efficiency in weight and
volume, and performance with respect to landing impact loads. While landing impact
performance has been adequately addressed in landing gear designs to date, load
alleviation following landing impact and during taxi and takeoff has received much
less emphasis. Some production landing gear designs have been developed which
provide a rough pavement capability, but these typically resulted in an adverse
effect on weight and volume. Several rough pavement landing gear designs have been
laboratory tested, but most of these either provided minimal improvement or were not
easily incorporated without affecting weight, volume or reliability/maintainability.
Enhancements in the area of computer modeling, particularly the change in
overall aircraft response to rough pavement due to changes in landing gear
characteristics, have resulted in a significant improvement in the ability to balance
a landing gear design to provide both excellent landing impact and rough pavement
characteristics. In addition, an improvement in laboratory testing capability
through the utilization of a computer driven hydraulic shaker has greatly improved
both the speed and accuracy of computer model validation. Exact rough runway profile
shapes at various simulated forward speeds can be input directly to the landing gear
in a drop tower and the resulting landing gear response measured. The combination of
enhanced computer modeling techniques and laboratory testing capability coupled with
extensive landing gear design experience has resulted in the development, fabrication
and very successful laboratory and flight testing of an improved high sink rate rough
field landing gear system for the F-l5 aircraft.
Development of this particular landing gear design was initiated by the
Cleveland Pneumatic Company (CPC) in response to Air Force inte 'est and efforts in
this area. CPC's approach was to utilize passive design concepts in order to
minimize any adverse complexity, and resulting effect on reilability and
maintainability. They utilized their experience in oleo relief valves and multiple
stage air curves along with an extensive landing gear dynamic response model to
arrive at designs for both the maen and nose landing gears with essentially the same
predicted reliability as the existing F-15 landing gears. All modifications to the
landing gears were internal and consisted of replacement of several internal
components without requiring any modifications to either the outer cylinder or
piston. In the 1984/85 time period, the improved main and nose designs were
subjected to a series of simulated rough runway tests at the Air Force's Landing Gear
Development Facility (LGDF) which verified the enhanced performance offered by the
new designs.
The STOL and Maneuver Technology Demonstration Program (STOL/MTD) contract was
awarded to McDonnell Douglas in October 1984 with the F-15 aircraft selected as the
5-2
demonstrator vehicle.
In addition to demonstration of 2-dimensional
thrust
vectoring/reversing nozzles, integrated flight/propulsion control, and advanced pilot
vehicle interface, the demonstrator was to have a landing gear capable of both rough
field (multiple 4.5 inch bumps at 60 foot minimum spacing as well as a 5 inch
cosine shaped dip of 50 foot length) and ircreased sink rate (12 ft/sec) operation.
The CPC landing gear design was chosen for this program in March 1986, having already
demonstrated the bump capability and requiring only minor modifications to achieve
the increased sink rate performance.
The nose landing gear was successfully qualification tested at the LGDF in
September 1986. The test weights were based upon a STOL/MTD aircraft takeoff weight
of 50,000 pounds and a landing weight of 35,000 pounds. While the takeoff weight was
less than that of an F-15 C/D (68,000 pounds), the landing weight was the same as an
F-15 C/D. Drop tests were cinducted utilizing both wheel spin-up and simulated wing
lift. Even at a 14.1 ft/sei sink rate, the loads were well below limit values as can
be seen in Figure 1. The energy absorbed at this sink rate was twice that to which
the landing gear was designed as the energy absorbed is proportional to the sink rate
velocity squared.
Rough runway laboratory testing was conducted in a drop tower with a hydraulic
shaker providing a simulation of ground profile traversal at simulated forward speeds
of 30, 50, 70, 90, 110, 120, and 146 knots. Ground profiles that were simulated
included a 5 inch 1-cosine
^
shaped dip of 50 foot length, multiple 3 and 4.5 inch
bumps with an 80 fo t w--clng, and a single 7 inch bump. These profiles are shown in
Figure 2. As can be s-n in 'I re 3, the peak gear loads for all of the rough
runway tests were significantly below the design limit load for the nose landing
gear. Although testing of the nose landing gear to the limit load would have
provided an assessment of the maximum allowable bump height, the fact that this
particular landing gear was to be utilized on the STOL/MTD flight test aircraft
precluded such tests. A final test for the nose landing gear was a sudden free
extension test which verified the improved rebound damping. Post test inspection
revealed that no damage or deformation of either external or internal components
resulted from the qualification testing.
The main landing gear was then successfully qualification tested in the
January/February 1987 time period. The nine inch stroke of the main landing gear
constrained the improvement in landing impact sink rate as compared to the sixteen
inch stroke of the nose landing gear, but a 12.5 ft/sec vertical sink rate was
obtained in testing nonetheless. Both the nose and main landing gears were designed
for a 10 ft/sec vertical sink rate. As in the nose landing gear drop tests, wheel
spin-up and simulated wing lift were utilized to enhance the simulation of landing
impact. For the tail down attitude drop tests of the main landing gear, a 12.5
degree wedge shaped platform was placed under the drop tower. The 12.5 ft/sec sink
rate tail down drop test data is shown in Figure 4.
Rough runway testing of the main landing gear was conducted in the same manner
as for the nose gear with the same bump and dip profiles being utilized. As can be
seen in Figure 5, the main landing gear performed very well when subjected to the
simulated rough runway profiles. In the case of the single 7 inch bump, predicted
loads close to the limit precluded testing of the 120 and 146 knot cases, especially
considering that this bump height was in excess of program requirements. A key
aspect of the performance of the rough field design that was verified in these tests
was the fact that the largest improvement over the production F-15 C/D landing gears
occurred during the highest load test cases. This was expected as the designs were
optimized to *chop off' the load peaks that occur during repair traversal. The main
landing gear tests also verified the improved rebound damping characteristics
designed to minimize aircraft bounce at the higher landing sink rates.
Actual taxi testing of the F-15 STOL/MTD aircraft was conducted in the summer of
1989 at the Air Force Flight Test Center at Edwards AFB. For these initial rough
runway taxi tests, two 4.5 inch bumps with an 80 foot spacing were installed on the
runway using a combination of AM-2 aluminum matting and plywood tr achieve an
approximation of the shape of the 4.5 inch repair as tested in the laboratory. In
order to determine the prcdicted loads for the taxi tests, a computer model of the
landing gear dynamic response characteristics was validated with the laboratory test
data and then incorporated into an F-1 aircraft response computer model. The
aircraft response model had been previously validated with taxi data from the Have
Bounce Program. This computer model was then utilized to generate all of the
predicted loads for the upcoming taxi tests to ensure that no limits would be
exceeded.
Aircraft gross weights of 40,000 and 50,OOO pounds were tested over the two 4.5
inch bumps. An upper limit of 90 knots resulted from the location of the repairs on
the runway and the need for overrun area in case of an abort. The results for the
40,000 pound gross weight testing are shown in Figure 6. As can be seen from this
data, the loads were quite low relative to the landing gear limits and the resulting
improvement was minimal. For the 50,000 pc nd testing, the improvement relative to
the F-15 C/D aircraft with the current landing gear was significant, especially for
5-3
the speeds in which the current landing gears generated loads near their limits.
This data can be seen in Figure 7.
Fu.-'e testing that will be conducted over the 4.5 inch repairs will include the
effects of hard braking, landing gear performance during engine thrust reversing, and
two point attitude testing. Although the F-15 STOL/MTD aircraft will be limited to
50,000 pounds, the performance of the improved landing gears at the 68,000 pound F-15
C/D maximum gross weight has been verified during the earlier laboratory testing.
Significant improvements relative to the current landing gears resulted from the
higher loads that were generated by the high gross weight.
In summary, a significant improvement in rough runway performance has been
demonstrated by the CPC rough runway landing gear designs. This capability was
achieved with a minimal weight increase (less than 30 pounds for the shipset) and no
projected effect on reliability and maintainability. These designs can be applied to
other landing gears with the amount of improvement dependent on the landing gear and
aircraft configuration and available strut stroke.
The primary purpose for developing the computer models under the HB program was
to establish rapid runway repair Surface Roughness Criteria (SRC) for each aircraft.
The itent i a to dealiver t he SRC to the Operational Commaands thrugh technics,
manuals and in the form of 'overlay" charts depicting the allowable bump heights and
apacings versus distance down the runway. This is done for each aircraft at a cross
spectrum of gross weights, density ratios, headwind components. etc. Although these
SRC are accurate, they are also complex and become conservative when multiple
aircraft types operate on t,,e same MU6. Repair crews must be trained to use the SRC
and MOS procedures. Another drawback of this approach is that the addition of future
aircraft or even derivatives of the existing aircraft to the fleet requires Tech
Order revision and the addition of numerous overlays and perhaps more training. It
is not an ideal system.
Another method could be to automate the process. Figure 8 depicts an approach
that would solve many of the deficiencies of the current technique. The approach is
to modify the mainframe computer nrograms developed under the HB program to run on-
site using readily available PCs. The primary benefits of an on-site computing
capability are as follows:
MORE ACCURATE& Using a menu driven on-site computer program to simulate the
actual aircraft and conditions will be more accurate. Important aircraft parameters
such as gross weight and center of gravity and ambient conditions such as density
ratio and headwind component can completely alter the repair criteria. This approach
also reduces the conservatism built into the present 'static' system that is induced
when multiple aircraft are superimposed on the same MOS.
- RETALIATE FASTER If actual aircraft and ambient conditions are used in the
computer simulations, quicker repair times will result because less conservative
repairs will be identified.
- FLEXIBILITY An on-site simulation capability will provide the flexibility
needed to allow gross weight and other aircraft changes. Using the current method,
this would have to be approximated through interpolation.
- PROVIDES OPTIONS: One of the highest payoffs of this method is that it
provides the base commander with a tool for answering important *what if* questions
and getting accurate answers in a short response time. For example, if the runway
repair crews could not maintain the quality of runway repair specified, the commander
may want to reduce the aircraft gross weight to a point where a safe takeoff can be
made. Assessment of the effect of relocation of the runway threshold to permit
successful takeoffs is another option. The ability to quickly assess options is a
powerful benefit of this capability.
- UPDATE; As new aircraft or derivatives of existing aircraft enter the
inventory, updating the SRC will be a simple matter of mailing a floppy disk
containing the new computer model to each main operating base.
- MORE EFFICIENT On-site, menu driven computer programs designed to select the
MOS and specify the required level of repair will be much easier to use than the
current technique. Consequently much less training will be required and since the
process is simpler, there is less chance for error in a wartime situation. Also, the
computerized technique would do away with the requirement for tech orders, overlay
charts, and subsequent revisions.
* MONITOR REPAIRS: Repeated aircraft operations will result in degradation of
the repaired runway surface. Measured runway profile data can be fed into the PC
based computer programs to determine if rework is required and to what extent.
The following are 'rule of thumb' guidelines for repairing and operating on an
MOS based on the lessons Iearned in the Have Bounce project. These guidelines are
intended to be used in the event that the PC based system is not available or as a
validation check to the PC based system results.
S= V x 2.5
Where S is the spacing in feet from the trailing edge of the previous bump and V is
the aircraft velocity in feet per second. Aircraft speed can be obtained from the
flight manual and should include density ratio effect and headwind component.
* Any bump closer than the calculated S above will not exceed 1.5 inches itrom
the undamaged grade.
Use 'soft field' takeoff and landing procedures. The purpose is to unload the
nose landing gear as much as possible and unload the main landing gear with wing lift
as soon as possible.
• Minimizing aircraft weight will result in maximum runway roughness capability.
CALCULATION OF VI
CONCLUS IONS
In summary, both the technology of landing gear strut design and the
understanding and analysis of aircraft response to runway roughness have evolved
dramaticilly during the past decade. The benefits which would be providea by
incorporating this knowledge far outweigh the small costs involved. Increased
aircraft capability coupled with accurate and timely assessment of aircraft/repaired
runway compatibility would significantly enhance post-attack sortie generation. It
is recommended that the knowledge gained be incorporated into design handbooks,
military specifications, and operational procedures.
/
15 -
10- /
5/
0 2 4 6 8 10 12
STRUT STROKE (IN)
iIlIP
78I' - - -1t5 IN
0 F- oI IN
___-___--____________-
STRUT1W LTAKP f I.II
15IN 75,FT -iJ 3ZIN
5 ff -p - 1 - N
0J0 10I IP
1 O1 - I2 I
20 -
10
40-
20- *
-20-
60 -- "- _ - -_
50-
40--
30"
20
10
0
0 20 40 60 80 100 120 140 160
FORWARD SPEED (KNOTS)
60-
NLG LIMIT
20
0
0 10 20 30 40 50 60 70 80 90
SPEED (KNOTS)
MLG LIMIT
80 -
60
40
NLG LIMIT
20 ; . ,
0 10 2C 30 40 50 60 70 80 90
SPEED (KNOTS)
DAMAGED RUNWAY
PROFIECATAON SIMULATION
CRITERIA
OUTPUT
" G0 I NO GO
" BEST LOCATION FOR THE MOS
" QUALITY OF REPAIR REQUIRED
AIRCRAFT RESPONSE
LOAD "I/
LOAD
I IME
FIGURE 9. H
TYPICAL TIME HISTORY PLOT GENERATED By "TAXI"
OB
5-I1)
FEET
V500'
VELOCITY (KNOTS)
FIGURE 10. PLOT OF "VELOCITY SWEEP" SHOWING TYPICAL VULNERABILITY INDEX
by
D.C.Thorby
J.Johnson
A.B.K.Auld
B.T.Nevman
.J. Brooker
Dynamics Group
Aerodynamics Department
British Aerospace (Military Aircraft) Ltd.
Richmond Road
Kingston-upon-Thames
Surrey KT2 50S
United Kingdom
SUINAT
The paper describes the special landing gear requirements of the Harrier family of aircraft, and is
based on modelling and testing experience over a considerable period of time. Only topics peculiar to the
VSTOL (Vertical and Short Take Off and Landing) aspects of this aircraft are addressed.
Of the four possible modes of take-off, ramp-assisted (the "ski-jump*) presents unique landing gear
problem.. This is described, covering the design of ramp profiles and the procedures used to establish
service operating limits for the landing gear, including the effects of ship motion.
The particular problems associated with vertical landing are next discussed. This mode of landing can
produce landing gear side loads potentially much higher than are normally possible in a conventional
landing with forvard speed. Clearance procedures using a multivariate approach are described.
The Barrier has also been cleared for operation on unprepared rough fields. The Monte Carlo method,
applied to the results of numerical modelling using computer-generated surfaces, is described.
Other topics discussed are: runway directional stability, and the load and directional stability
implications of converting to radial tyres.
1. n eTraolcYI(M
The original Harrier concept has been described in many publications. Several marks have been
produced, but as far as the landing gear is concerned there are essentially two versions: the Harrier I
produced by British Aerospace, and the Barrier II produced jointly by BAe and the McDonnell-Douglas
Corporation. Dowty Rotol Ltd. are the main contractors for the landing gear. The gear units for the two
aircraft (except for the outriggers) are geometrically similar, but the Harrier II units are stronger with
stiffer strut spring curves designed for a sink speed of 15 ft/s (4.6 m/s) at higher weight rather than the
12 ft/s (3.7 m/s) used for Harrier I. Figure I shows the very similar landing gear layouts for the two
aircraft, and figure 2 sketches the Harrier II units.
A21 Barriers have four possible modes of take-off: vertical, short, conventional and ramp-assisted,
and three landing modes: vertical, slow and conventional. In addition, some versions of both Barrier I and
Harrier II have been cleared for operation from unprepared fields. This normally means grass fields, but
interlocking metal strips over grass are also used. The vertical landing requirement particularly has
dictated the use of an unusual landing gear arrangement, essentially a bicyclv with outriggers for roll
stability. The nose and main gears carry roughly equal static loads, each about 45% of the weight, while
the outriggers together take the remaining 10%. The unique rotating "vectoring" nozzies of the Rolls-Royce
Pegasus engine mean that the total engine jet force ay be used for forward thrust, jet lift or to give a
combination of these, as in "partially jet-borne flight".
Many aspects of Harrier operations involving the landing gear are perfectly conventional, and are not
discussed in this paper. Operations standing out as unusual are firstly, ramp-assisted take-off, or the
"ski-jump". This produces a normal acceleration in the order of 3g, virtually all reacted by the landing
gear. Secondly, vertical landing gives a landing gear side loading mechanism quite different from
conventional landing. Rough field operations are not unique to Barriers, but the methods used to clear the
GB Kk 5 version of the Harrier II may be of interest. Finally some remarks on the often neglected area of
directional stability on the runway, and the related question of changing to radial-ply tyres, are
included.
6-2
HARRIERI
HARRIER11
2. RAP-ASSISTM TAM-OFF
This is now the conventionul mode of take-off from the ships of several navies. The corresponding
landing on to the ship is always vertical. For those unfamiliar with the "ski-jump", Appendix 1 gives a
brief explanation of this interesting topic together with an outline of the method used to establish the
optimum exit angle for a particular application. Ramp-assisted take-off may be from land, as demonstrated
at the Farnborough and Le Bourget salons some years ago, but it is in naval applications where the
technique shows dramatic advantages.
Devising the best shape for the ramp is an interesting problem. If the landing gear were rigid and the
speed constant, a circular arc would clearly be the correct shape, since the aim is to impart the maximum
possible vertical momentum to the aircraft, while at the same time minimising the applied load. Figure 3
shows what happens in practice if a circular arc profile is used. Because the centrifugal load is applied
suddenly, an oscillation is set up. This has two adverse effects: it increases the peak gear loads,
requiring a reduction in endspeed to avoid full strut closure, and it could make the launch pitch rate
somewhat variable.
Yi
IA
ACTUAL
LA
dt' -
STATICLOAD
DUB TOVSIDZTAT IS
An analogous problem was known to the early railway builders, who found that a straight length of rail
followed by a curve of constant radius caused trains to oscillate in roll. Their solution was to insert a
"cubic transition", so that curvature and hence load increased more gradually. The same cure works with the
ramp, as illustrated in figure 4.
Y tMIC
[2 FLTT CRcLA' c
d.',
STATICLOAD-. i A
Later, an alternative way of preventing the heave and pitch oscillations, avoiding the additional ramp
length caused by the transition curve, vas found. This is described in the patent application (reference
1), and figure 5 gives a brief explanation of the aethod. It consists, essentially, of adding a correction
profile equal to the closure history of the landing gear strut under constant load to the circular arc, or
Whatever profile is required *statically". This has been made to Work vith the Harrier, in spite of there
being two landing gears vith differing characteristics on the same track.
CIRCULARA
ARC
yx
x rt
CONPRBSSION HISTORY
OP
STRUTAND TYREVITO
CONSTANT
APPLIEDLOAD
t (. xlv)
The process of optimizing a profile for a given application is in practice iterative, using a
mathematical model fully simulating the launch process. The criterion currently used is that no strut or
tyre should fully close during the launch. Load might be thought to be a more appropriate criterion but,
because strut closure velocities and hence damping forces are very lov on the ramp, load and closure are
virtually interchangeable and related by the strut spring curve. Full strut closure Would imply large
indeterminate loads, and this is currently avoided.
6-5
Since a typical ramp reaching 12* may be traversed in much less than a second, the launch pitch rates
could reach very high nose-up values. Fortunately this does not actually occur with the Harrier, as shown
in figure 6. The nose-down moment producd during the short time that only the main gear and outriggers are
on the ramp provides an almost perfect correction leaving a desirable, slightly nose-up, pitch rate at
launch. This permits the optimum angle of attack for the seml-ballistic, partially jet-borne, phase to be
obtained quickly, and earlier than from a flat short take-off (STO). Thus an important secondary advantage
of ramp take-off is that it allows aircraft rotation earlier than would otherwise be possible.
NOSEWHEEL
LEAVESRAMP
- 20
KALIN
WHEEL
" 10 VHSRAMP
10
5 5
0,I
7.2 7.3 7.4 7.5 7.6 7.7 7.8
TIME(SECONDS)
The service user must be provided with safe limits foc ramp take-off. These involve performance
limits, which are beyond the scope of this paper, and landing gear limits. In practice the latter are
produced by computer simulation, validated by ship trials. As far as the landing gear is concerned it is
normally only necessary to measure strut closures in these trials.
6-6
The rolling notion of the ship has potential aircraft stability implications, although no problems
have been found in practice. It is the pitching motion of the ship vhich is more important since it may
have a favourable or unfavourable effect on both performance and landing gear operating limits, depending
upon the point in the pitch cycle at vhich launch takes place. Controlling the moment of launch to improve
the statistics is theoretically possible, but is inherently more difficult than vith a catapult launch due
to the longer time interval (typically ten seconds) betveen the start of launch, on the flat deck behind
the ramp, and free flight. An allovance based on ship pitch amplitude, assuming random timing, is currently
made. The loading effect on the landing gear due to ship motion Is appreciable and is found to be caused by
Coriolis force (a function of ship pitch velocity and aircraft speed) as much as by the direct vertical
acceleration. Figure 7 illustrates this and shovs the predicted MS total normal acceleration per EMS
degree of pitch for a cruiser-sized ship. Modelling of ship notion takes advantage of the large frequency
ratio between the ship pitching motion and the rigid modes of the aircraft on its landing gear, and the
short time taken to traverse the actual ramp. During this time the total aircraft normal acceleration can
be regarded as constant, at values typically betveen 0.5 and 1.5 g, i.e. the effect of the motion can be as
much as 0.5 g either way in addition to the usual gravitational 1 g. The simulations are carried out at
fixed values of "gravity" (e.g. 1.0, 1.1, 1.2, etc.). Each of these can be interpreted as a range of ship
pit(l and aircraft speed combinations, giving considerable economy of computation.
Landing gear operating limits are kept entirely separate from aircraft perfore_ _.nsiderations.
9.e8, 9v
0.30
(g/050,
0. 28
0.26
0.24
0.22
0.20
60 70 0 90 I00 1to 120
AIt~RAFSPRED(10JITS)
While these can be important in a catapult launch they are even more so in a ramp launch vhere the
landing gear is certain to be highly compressed for the majority of launches. Although the usual hydraulic
pulsing problem associated vith sudden release had to be considered in the main gear, it vas the sudden
release of bending mmnt from the nose gear that caused difficulty in early development. In its fully
compressed state the nose gear has about 400 m of trail. Since the vheel load is very high at the end of
the ramp, it can be seen that a bending oscillation is inevitable vhen this is suddenly released (figure
8).
6-7
500
The cuze for this was extremely simple, consisting of a run-down 4ection designed to unload the nose gear
In one natural period of the oscillation, or about 0.044 s, (figure 9).
TANGN - _ 421-
2000.-~
Figure 10 shovs that the oversving is zero if this is achieved, ad can be kept small for the practical
range of transition times. This time varies, of course, with exit speed. All Harrier ramps must be fitted
with the run-down section, which in practice may be straight, about 2 m long, and tilted down about 125 m
(about 3.5*) from the ramp exit angle.
e S 5 2 2S
T
TIM - tn
TRANSITION
NATUSALPUIISO
•TIM
FIGUR 10. REDUCTION IN OVURSVINGAGAINSTTINE FOR LOADREDUCTION
3. VERTICAL LANDIN
Of the three landing modes, only vertical landing will be discussed here. The sultivariate approach to
limit and fatigue loads (based on NIL-A-8863A), described below, was also used for the slow and
conventional landing modes.
Whether a tyre is rotating or not at touch-down makes a fundamental difference to the way side load is
developed, and side loads are potentially much higher in a vertical landing than in a conventional landing.
In the latter, if the aircraft is yawed relative to the runway, each tyre has a slip angle and this gives
rise to a side force. The rotating tyre however acts as an efficient damper and the side force developed
rarely exceeds a small fraction of the vertical load. This is reflected in current qpecifications where the
design side force is typically between 25% and 40Z of the maximum vertical load on the tyre. No such
mechanism operates in a nominally-vertical landing, particularly in a case where there is lateral velocity
but absolutely zero forward velocity. When there is no laterally-acting shock absorber the lateral energy
must be absorbed by the elasticity of the tyres and possibly that of the landing gear in lateral bendi g.
These have an effective "stroke" of only 50mm or so and are practically undamped. Very large side forces
are thus possible, but an ultimate limit is set by the coefficient of friction between the tyre -ad the
runway. Under dry conditions this can be as high as 1.0, so the side load is potentially as largo as the
maximum vertical load on the tyre.
Ii~
....
....
~~m ~~~.. ..........
........ .... . . . .
6-9
The landing gear side forces determined from a mathematical model are only as accurate as the basic
data used. The empirical relationships developed by Smiley and Borne In reference 2 still appear to be the
beat basis for modelling tyre side forces when specific measurements are not available, and for
extrapolating measurements to conditions not tested. Figure 11 illustrates the empirical model applied to a
rotating tyre of a particular size. "Cornering power" is the slope of the side force versus slip angle
curve at low slip angles where it is nearly linear. Steady state forces are given by the expressions
presented in reference 2, and it is stated that the side force builds up exponentially with distance, with
a *time constant" in the order of one footprint length. In the case of conventional landing, this is rapid
enough to regard as instantaneous, but can be taken into account in the case of nomin :lly-vertical landings
with some forward (or rearward) velocity.
400
300
200
1100
II I I
0 2000 4000 6000 a000 10000
RADIALLOAD
(LB)
9000
RADIAL
LOAD
(IS)
66000
p4000
200
o to 20 30 40 50
SLIP AGLE (050)
Reference 2 appears to predict steady-state side forces of cross-ply (bias) tyres due to yawed roiling
with remarkable accuracy. Figure 12, for example, shows the "cornering power" of cross-ply tyres measured
by two different suppliers; the curves fall each side of the Sailey and Borne prediction for this size.
Radial tyres have, in our experience, somewhat higher cornering power as shown in figure 13 which is based
on data from the same two suppliers. It is worth noting that a third supplier claims that side-force
characteristics can be 'tailored" to requirements, and that the cornering power of a radial tyre can
actually be lover than that of the correspo-ding cross-ply tyre.
6-10)
600
400
____
RADIAL,
LOAD(LB)
FIGU
12. COUIIM POWI OF
TWO HhERS
OF CROSSPI.TTRE CofPARD
WIT NACA
4110 PREDICTION
60 0T
M
SUPPLIER'A'
500
200C
100 _ _ _ _ _
RADIALLOAD0
(LB)
FIGM
3 . 1UM&
P(N OF TOD HMUZ
Of RADIAL TT ll (offAREDWMT NCA
4110 MtHRD 143
4
Reference 2 also predicts lateral static (non-rotating) tyre stiffness and, based on one supplier's
data, the predictions are in the order of half the measured values for cross-ply tyres as shown in the
following table. The table also shows the static lateral stiffness of the same supplier's radial tyres
which, as would be expected, are less stiff than the corresponding cross-ply tyres.
RADIAL LOAD
30 kM 75 kN
These figures suggest that some caution is necessary in using reference 2 to predict static lateral
stiffness, and measurements are required in critical applications.
NIL-A-8863A, which is the specification applicable for ground loads on the AV-8B version of the
Harrier II, requires a multivariate analysis for landing limit loads. Table I of the Specification lists
the variates to be investigated, their mean and standard deviation values etc., and the range to be covered
both individually and in combination.
Individually, each variate need only be considered within the range giving 0.001 probability of
exceedance each way, or about 3.1 standard deviations each way for an unskeved normal distribution. For
combinations of variates, the joint exceedance probability PT need only be considered up to the value given
by one variate at 0.001 probability with all the others at mean value, i.e. 0.5 probability of exceedance.
Thus:
n-
0.5 1
PT - P0
where P0 is the individual minimum probability of each variate, and n is the number of Vd~iates.
Figure 14 shows these limits for two variates V and V with the other variates at fixed values. C is
a constant, determined by the fixed values chosen for the other variates.
5
P(VB < VE . (VA) VA O C 'V '
A).P(V
k
~t C
AREAT0 BE tNWsrTAsV
r- ......
vBT
s [lf)
o denote, 1,v
i a )
()SINK SPEED
()PITCH ANGLE
(iii) RtILLANGLE
(iv) LATERAL VZLvCITY
All cambinadions ot these variates can be cuvered by a series of load contour plots like iigures 15 and 16.
Here sint speed und pitch angle are temporarily fixed, and load contours *'tc.are plotted on a graph of
roll angle gainst lateral velocity.
12 A. u 12
2 7
FIGURE 15. MAIN GF.IR VERTICAL AD AMt) CLOWuRE UWTOUtR lI'Y '1PKED OF 12 YT/SKC ANT)PITCH ANGLER
OF 8*
6-13
-12 8 4 0 4 0 2
ROLL ANGLE (0501
GEAR
MAIsorN TO STARMR
sIP LOAD
V"\\
I g,:
-8
12 -
'ROLLAMGL9 (DIG) R 2
VIGURE 16. MAIN GER SIDE LOADWIT SDK SPMW OF 12 FT/SEC AND PITCH ANGILEOF He
Figuie 1? shows roll and lateral Velocity combinations producing limit load boundaries for the various gear
onits at fixed sink speed and pttch angle, in relation to the probability boundary required in this case.
For fou.rvaiateS the value of PIas defined above is 0.000125.
MentJ'vn should be made of a series of landing impact parameters measured on the AV-8A version of the
Harrier I by MADC. The se were invaluable in defining the multivariate paraneers for the later AV 88. All
relevant impart quantities were extracted from 70 ma film and expressed in statiltlcal terms: mean,
-randarJldeviat ion, skew and hurtosi s were lised for statistically significant numbers of landings. A
similar exercise vas oar ied out by the Royal Navy and the RAE. but using a smaller film size. Data of this
kind f eg ianding load piedictioata sound statistical basis mod should be gutheced whenever possible.
Fatiguieloads, applied by the supplier to the landing geaz and by the airframe mantfacturer to the
aic!raue ah-:tments.can be generated in a similar way, Io order i limit !he loadings to a reasonable
number , rS. probability density curve of each variate must he 'lumped" as shown in figure l*8, in this raxe
o five di.%cret e valu.es.T~he overall "a te rno of cases to 't- ron to produce the fatigue loading schedule,
for a V..fft , Ia landing mole. can then he built out from all combinations of such discrete values.
6-14
SSIDE WAD
SIDEAI WGDNOABRA
2 LOAD VERICAL SD OD
4 '. SIIDELLADD
12 Nr VER0TICAL12
ROLL ANGLE(DEG)
46.61
23.61 13.61
3. 11
ROLLANGLE
Ship no1i0n can change the statistical parameters for vertical landing. In order to arrive at a
reasonable rationale, it is necessary to consider how a vertical landing is perforsed with a Harrier. This
1s ac complished by first hovering over the deck station upon vhich it is desired to land. The vertical
position is fixed in space, rather than relative to the ship. A conistantrate of descent Is then set op by
reducing the throttle setting slightly, then returning it to the hover position. Given this procedure, it
is seen that the statistical varlition of impact parameters should be a combination of two uncorrelated
ditr ibutions. one doe to the aircraft (similar to operation from land) the other due to the ship, which
can be defined from ship data, Independently of the aircraft.
6-15
Some versions of Harrier I and Harrier II have been cleared for operation from unprepared fields.
This generally means grass fields with or vithout an interlocking metal strip covering. The ground
roughness is quantified by a method believed to be unique to the Royal Air Force and known as the "Bump
Spectrum". Production of the bump spectrum for a particular rough surface is illustrated in figure 19, and
figure 20 shovs a typical result produced by computer although the process is simple enough to implement by
hand. The method works very veil in practice and the bump spectrum plots can be related to the more
fundamental Power Spectrum method.
La
POINTS (L 5,hl) ,
(L 2 .h 2 ) 9TC AR PLOTTD ON A ViMP SPMCAM DIAGRAN
35
30
25
20 A A
0 to 20 30 40 50 60 70 so 90 100 110
UNDUIATION
LEW (M)
4.2 TM PILOSOPHY
Rough ground clearance of a particular aircraft presents a statistical difficulty even vhen, as in the
case of Harrier GR Nk 5, fully instrumnted trials vere carried out. During such trials it is practical and
reasonable to use five or maybe six different operating sites. Each strip, b.,vever,can only be expected to
contain a few significant bumps and a clearance based this would always be open to the objection that
anoter strip meeting the bump spectrum requirements could fully close the landing gear. In order to
overcome such objections it would be necessary operate from a very large numter of strips. With a well-
validated model this should be possible by computer simulation. Pursuing this line of thought, the
following rationale was developed:
1) The trials were only used to validate the model. The surfaces used were:
(a) A memotk rummy to check aerodynamic loads in ground effect by using the landing gear load
lnstrumenta ion to measure the lift and pitching moment directly at various speeds.
(b) An accurately known diacrate bm, similar to a standard runway repair bump.
(c) Grass and metal-covrmi grams surfaces with increasing roughness as masureJ by the bump spectrum
and progressively increasing aircraft weight.
6-16
2) Having validated the model, long lengths of computer-generated "rough ground" were produced based on
Gauzsian white noise. It was found that a single integration with respect to distance produced surfaces
indistinguishable from samples of real ground, as judged by the following criteria:
(a) Similar Power Spectra. This follows from the way the surfaces were generated: single integration of
white noise.
(c) Using the model it was also demonstrated that the ratio of RES load to RNS closure for each strut
was similar for simulated and real surfaces.
3) In principle a large number of take-offs and landings should then be simulated. It was, however,
demonstrated that a series of constant-speed runs could be substituted, provided the model was adjusted
to include the vertical forces and pitching moments due to thrust and, during landing, braking. Thus
acceleration or deceleration has no effect, in itself, on response, but the associated steady forces
must be included in the model. The use of constant speed runs had the advantage of making the time
histories of loads and strut closures etc., stationary and therefore more amenable to statistical
processing. It should be noted that the model was not linearised in any way.
4.3 RESULTS
Some adjustments to the model were required to achieve good correlation with tests. These were:
(a) Friction forces, which had been based on static friction measurements had to be considerably
reduced. This is reasonable, since friction tends to reduce with increasing rubbing velocity.
(b) In common with other workers it was found that a single gas spring curve would not give satisfactory
results over the range of closure velocities experienced on rough ground, and it was necessary to
represent the thermodynamic behaviour of the gas in the struts.
With these adjustments there was generally very good agreement between model predictions and test results,
as shown in figures 21A and 21B. In these plots the solid line is the computed response using the measured
surface profile in all cases, and the broken line is the measured aircraft response. The computer model was
provided with measured aircraft velocity and engine fan speed time histories as part of its input data.
9.0
0 ~6 7 a 9 i 1 1
TINS (SE)
3.0 .. - -
0 1 2 3 4 5 6 7 a 9 10 11 12
TIKE (SEC)
Having accepted that computer-generated surface profiles could be used and that accelerating and
decelerating ground runs could be replaced by a series of constant speed runs, the clearance procedure was
reduced to0one of time series analysis. The simulated ground surface (figure 22 for example) when
"traver sed" by the model at 50 knots produced the main gear closure time history shown in figure 23.
0.46~
M 0.02
.0
0.0 0.1 n.2 0.3 0.4 0.5 0.0 01 0.8 0.9 1.0 1.1 1. L. 1.4 1.5 1.0 1.? 1.0
m 12.0
10.0
0 4 6 12 16 20 24 2B 32 36 40 44 48 52 56 601 6'. 68 17
TIRE (SEC)
These time histories could be processed in any desired way, the most obvious being to determine amplitude
or peak probability distributions. Both were tried and, although peak distributions are theoretically
des irable, amplitude distributions worked better in practice. Figure 24 is a plot of occurrences in narrow
amplitude bands for main strut closure, the most critical quantity in the case of the Harrier at the
critical speed. Figure 25 is a cumulative amplitude plot of the same quantity on a grid such thata
Gaussian distribution of amplitudes would produce a straight line. It will he seee that the main gear
closure plot is not straight, reflecting the oonlinearity of the spring curves since the "ground" surface
input was Gaussian.
4000_________ ____
3500 _____
3000 _
1500 _
2000
500 __ __
"aImc'.os..m (IN)
10.0 z z.
o.o-.... .-. I
9.0
IN A PROBABILITY
FlrGU 25. VIL GU CIA5URE CMATrV A tIE DISTRBUTION VITO grAPOLATION TO CLOSUE LDIIT
Various methods were tried to relate the probability plots to servic- conditions. All gave similar
results; the most satisfactory appeared to be the following:
(a) It was first decided that an acceptable probability of full strut closure would be once per 1000
operations, equivalent to 4000 second. in the critical speed range.
(b) A fictitious Gaussian distribution is fitted to the upper tail of the cumulative closure amplitude
plot. This is represented by the straight line extrapolation in figure 25. Only the upper ,xtremity
of the distribution is of importance, of course.
(c) Since the main gear response is essentially narrov-band, with a mean frequency of 1.8 Hz, we may say
that the extrapolated Gaussian amplitude distribution corresponds to a Rayleigh peak distribution,
and that this may reach full closure once in 7200 peaks (i.e. 1.8x4000). This sets a limit to the
standard deviation of strut closure that can be permitted at any particular aircraft weight.
Typically, 4.8 standard deviations must be allowed between mean and full closure.
5. RIMATDIUhCTMLt&L STAJILITY
The unusual landing gear arrangement of the Harrier has made us conscious of the importance of the
directional stability of aircraft on runways, a phase of operation often overlooked, since the usual
tricycle arrangement tends to be inherently stable. It is well known, however, that the aircraft of fifty
years ago having castering tail wheels, were directionally unstable at low speeds.
Directional stability on the runway can be assessed by relatively simple modelling, but simulator
work, with a pilot in the loop, seems necessary for good results. Reference 2 appears to predict tyre
cornering power (the dominant effect) with reasonable accuracy, as discussed in section 3.1.
An interesting example of a runway stability problem, from the Harrier (Kestrel) prototype days may be
mentioned. Following a very light landing, it was possible for the aircraft to roll over slightly,
producing the main gear loading sitMdtion shown (exaggerated) In figure 26.
6-20
'"'
FIGURE 26. AIN GEARIN ROLLEDATTITUDE FOLLOING A LIGHT LANDING
This resulted not only in the loss of (stabilising) side force from one tyre, but also the cornering power
of the remaining tyre was reduced by overloading. As shown in figure 27, instead of two tyres operating at
point A on the cornering power curve, there was only one tyre at point B. The problem was solved by
introducing a main gear modification permitting the strut to compress with almost no load until both
outriggers had made contact with the ground, thus avoiding the rolled attitude.
iA
VZMCAL WAD
All marks of Harrier are in the process of converting t- radial ply tyres. These tyres have seveial
advantages over cross-ply or bias tyres, as demonstrated in the automotive field. Vhen converting an
existing aircraft to radials, load and directional stability implications aeed to be considered.
Instances of increased landing gear loads due to changing to radial tyres can be predicted
theoretically. The most obvious is in a conventional or slow landing with appreciable yaw at touchdown
where increased *cornering power" (i.e. side force per unit slip angle) can be expected to tncreas wheel
side forces. Also wheel side loads In low-speed turns can be expected to be redistributed. Conversely, the
reduced lateral static stiffness of radial tyres would be expected to reduce side loads in vertical
landings with lateral velocity, possibly throwing more load on to other wheels, when these are not tted
with radials.
Although the mass of a radial tyre tends to be lover than the equivalent cross-ply tyre, the moment
of inertia is g.nerally higher, with implications for spin-up drag.
6-21
The increased cornering power of radial tyres can theoretically have an effect on runway directional
stability, particularly on an aircraft such as the Harrier where not only is there an unusual distribution
of static load between nose and main gears (close to 50/50), but also tyre forces generally are large in
relation to aerodynamic forces. As with road vehicles radial tyres at the rear are stabilising and at the
front destabilizing (assuming an increase in cornering power). At the very least aircraft converted to
radial tyres should be assessed by an experienced test pilot. It is also necessary to establish clear rules
about the mixing of tyre types.
7. COICLDIC IRNARKS
Ramp-assisted take-off with the Barrier created a new loading action for the landing gear. The
resulting loads are readily predictable and no difficulties have been encountered using gear units not
originally designed for the purpose.
Vertical landing differs fundamentally from conventional landing in that landing gear side loads,
produced by a different mechanism, are potentially as large as the corresponding vertical loads.
The Monte Carlo approach to rough ground clearance on Harrier II resulted in some economies, since it
permittel the number of test flights to be reduced without loss of confidence. A similar approach could
probably be used for the more common requirement of clearance for operation on repaired runways which might
be regarded as random surfaces not necessarily having Gaussian amplitude propertiea.
Runway directional stability calculations, at their simplest, are relatively easy to carry out and it
is surprising that they do not receive as much attention as flight cases. They are indispensable in the
case of unconventional landing gear layouts and can be useful also in predicting the effect of burst tyres,
steering system failures, etc.
The effects of fitting radial tyres to established aircraft types must be considered seriously, taking
into account possible load increases and changes in runway directional stability.
8. gp~
9. ACK009LUMINff S
The authors are grateful to British Aerospace for permission to publish this paper and for the support
of the Procurement Executive, UK Ministry of Defence. Any views expressed are those of the authors and not
necessarily th.se of British Aerospace.
6-22
APPIDIX 1
First consider the take-off of a conventional aircraft (figure Al). The runway is used to accelerate
the aircraft to flying speed on the ground. Aircraft rttation then produces enough wing lift for unstick.
D T T
1,v UNSTI
Figure A2 shows a Harrier "short take-off". This aircraft does not rotate in the usual sense; instead
the engine nozzles are partially rotated so that the combined upward components of engine thrust and wing
lift produces unstick, even though the wing lift is much less than the weight.
T-
LL
4 - 550-
Introduce a ramp, as sl..,,, in figure A3, with the end 150m from the start of the ground run. First
consider the hypothetical case of no forward acceleration or lift increase after leaving the ramp. In this
case, the aircraft would describe a ballistic, parabolic, trajectory and re-land approximately 100m from
the start point. This in itself is of little use but note that a "runway" 1000m long has been created.
- ~ - FARAROIA
1000.-
Actually, as shown ;n figure A4, the aircraft does accelerate considerably during the ballistic phas,'
and flies out of the parabola because of the build up of i..nglift coabined with the vertical component of
engine thrust. It is using an effective "runway" much longer than th,Ireal runway and the ramp-assisted
take-off run can be less than one-quarter of that required for a flat Slu, exolaining the expression "the
runway in thp sky". It is thus feasible to launch heavy aircraft from very small ships.
IV
Conventional aircraft have used the ski-jump technique but, in comparison with the Harrier, the
reduction in take-off run is considera.ly less because engine thrust cannot be vectored.
6-23
For a given mark of Harrier, it is possible to produce c:. rts like figures Al, A6 and A7. Figure A5
shows the minimum speed at the end of the ramp to give a safe launch as a function of weight and launch
angle. This chart would be unaffected by the method of launch: for example it would apply equally well to
an aircraft catapulted at that speed and angle. Figure A6 shows the maximum launch speed from the viewpoint
of the landing gear strut closure as a function of weight and ramp radius. Figure A7 shows the achievable
endspeed as a function of weight and deck run.
RaMPA,.LE
6.
10*
12-
16"
LAWCHVEIGUT
RAIU
10011
LANC6VTIC DNT
200m
LAUUM WRIGHTr
Deck run is the most fundamental limitation and it is also clear that exit angle and radius are
related by:
H = R(l - Cosa)
for any given ramp height or length, the other main limitations (see Figure AS).
aB
FIGUREAR. RAMPGEOMETY
16
14
10
AVAILABLEDEK RUNl
It is easy to show that an optimum ramp angle exists, defined as that which will permit the maximum
aircraft weight to be launched, and is the angle where all three limitations listed below coincide:
The optimum angle may be plotted, as a function of available deck run and permissible ramp height as shown
in figure A9.
Having determined an approximate optimum angle, simulation methods are used to refine the estimate,
taking into account further variables such as wind over deck, ship motion, etc. Although useful as an
approximatiot., having simple geometry, purely circular arc ramps are not recommended for the reasons given
in 2.1.
CONSIDERATIONS ON OPTIMALITY OF
LANDING GEAR ARRANGEMENT AND DESIGN
A.J.Krauss
Germany
1. INTRODUCTION
much effort has heen spent and is still being spent on development and improvement of
optimization procedures and computer codes.
The formal task of optimization is to quantify that set of design variables which both
satisfies a set of constraints and yields the absolute maximum or minimum of an objective
function. Cne of the intrinsic problems of optimization is that the objective must be
expressed as a numerical quantity, which often leads to conversion problems (in which way
can one convert a quality into a nonnegative real number?). An other even more serious
problem is that formal optic''7ation requires that the dependence of the objective on the
design variables must be analytically defined. In relation to these problems it appears of
secondary importance that in most cases the optimization process will stop at the local
"optimum" which is closest to the starting design, albeit there might exist better optimum
solutions across the surrounding ridges of the objective function. Before formal
("automatic") optimization methods are called in, the design must therefore be developed to
a starting point in reasonable vicinity to the real optimum.
The author does not intend to weigh the importance of the vatiouis functions of landing
gears. Rather there will be the attempt to proceed from "simple" functions to more complex
functions'while keeping in mind that all these functions cannot truly be separated from each
other nor from the aircraft and its characteristics.
This function should be provided in all phases of taxi-out, take-off, landing, taxi-in,
and ground operations such as reverse braking. Deficiencies w.r.t. protective function
detrimentally affect total weapon system effectiveness and life cycle cost by e.g. increased
air-raft downtime due to more frequent structural damage.
2.1.1 £tCa;g,' Cround Roll
The protective function requires a minimum clearance between the lowest point of the
aircraft (including external stores) and the plane ground beneath an "all flat" landing
gear. The residual ground clearance is recommended to be at least 0.15 m (6", Ref.i) against
fixed and movable parts in their most critical position. Under normal operating conditions,
this clearance is deemed sufficient to cope for elastic defle2tions of landing gear and
aircraft as well as for arrestor cables bouncing up behind nose and main landing gear. It
the "all flat" condition was caused by a failure, then this clearance leaves the operator
with a fair chance to remove external stores from the aircraft before jacking or hoisting.
This ground clearance requirement determines the least possible lengths (nence least weight,
approximately) of the landing gear legs with a given wheel size.
Para 2.1.1 has dealt with protection in a virtually ground-parallel attitude when the
only requirement w.r.t. main landing gear position is that the effective point of ground
contact must not be farther forward than the aircrafts most rear C.G. position with landing
gear extended.
Unless the aircraft is equipped with a tail bumper the extreme landing pitch attitude,
on an "all flat" main landing gear, sets the requirement. Ref.l sees that extree 3ttitude
at 90% of the maximum 'ift coefficient; however, with computerized flight controls and
vectorized thrust this may well be exceeded. There should also be consideration of failure
cases such as flaps-up landings.
Although, on paper, landing pitch attitudes could be limited by limiLed pilots' view
the designer should not hastily soften up landing gear position requirements on this ground.
It might prove relatively easy to improve pilots' ability to compensate poor view (e.a. by
electronic means) while cost for repositioning of the landing gear could be prohibitive.
Thus considerable landing performance potential can be lost due to an early mistake.
Ground clearance aspects could be treated without referring to the aircraft's C.G.
position. However, fulfilment of the following protective function requires consideration of
C.G. position both longitudinally and vertically.
This requirement is not only based on protection against ground contact but also on
aspects of flight safety and flight performance. Flight safety is enhanced if at all
permitted landing touch-down attitudes regardless of runway friction the ground forces on
main landing gear induce a pitch-down moment. Flight performance is enhanced if pilots may
land their aircraft at high pitch attitude and correspondingly low approach speed without
having to expect reversal of landing gear pitching moment.
Fig.3 illustrates that a limitation of landing pitch attitude may lead to considerable
increase in landing speed and consequently loss of A/C landing performance.
This loss could eventually be exp-essed by an ealyLical function and be included in a
formal optimization procedure.
If any possible a tail-bumper should be avoided on a combat aircraft due to various reasons:
With C.G. in its most adverse position (rear and high) relative to the main wheel
ground contact point in static position, the angle K (see insert on Fig. 4) could be
determined to not less than 39 deg by the simple requirement that a braking ground drag
coefficient of p - 0.8 should not completely unload the nose LG.
However, detailed dynamic analysis of reverse braking yields much less restrictive values
for K.
Results for a combat aircraft (Fig.4) show that in this case a tail-bumper can be avoided
if K > 17 deg.
The geometry of this boundary condition is shown on Fig.5. 'he dependence of affordable
reverse speed on the angle K could also be cast into a penalty function and be used in a
formal optimization procedure.
Cautionary note: The angle K is measured in the earth-fixed coordinate system. Static
pitch attitude of the aircraft must be taken into account to arrive at the appropriate angle
in the aircraft-fixed design coordinate system.
Strictly speaking, only special landing gear designs such as a "Jump Strut" may enhance
nose wheel lift-off. However, this function is antitheticel! to para 2.1.4 Reverse Braking
and is therefcce due to be discussed here.
An aircraft can be lifted off the runway as soon as Lift exceeds Weight. For
conventional HIOL aircraft lift is basically a function of airspeed and angle of incidence.
In order to achieve shortest possible take-off ground roll distance maximum allowable
incidence should be attained at the time when minimum lifL-uLZ bpeed is reached. Tu attain
the incidence in general requires rotation of the aircraft from a by and large horizontal
position. However, this rotation can only take place if an.'when the aerodynamic moment
about the pitch axis overcomes the opposing moment produced by vertical and drag ground
forces on the landing gear. Nose wheel lift-off speed mst be less than main wheel lift-off
speed. Since rotation takes finite time, the required difference l'woen no- #he-l lift-off
(NWL-) speed and Main Wheel Lift-off (MWILO)speed becomes larger with larger A/C thrust to
weight ratio.
The following equation yields the nose wheel lift-off speed with two simplifying
assumptions, viz. that the aircraft is rigidly supported on the ground and that lifting
forces and pitching moments are produced aerodynamically.
V -. [... (2.2-1)
p A Cm._. .c/(l,._ + p., I" h G) C,_..
It is apparent that the easiest way to achieve any desired nose wheel lift-off
speed is manipulation of the longitudinal distance from C.G. to main LG ground contact,
L.,,. However, this may lead to operationally unfit WG configurations (e.g. with
respect ot raveLoe braking). To avoid this it is urgently recommended to perform dynamic
simulationb which include ttie flexibility of LG and tyres. This is much closer to
reality than the assumption of rigid support and yields larger affordable L,,, than eq.
2.2-1.
With modern aerodynamically unstable aircraft designs dc./dc is positive for
controls fixed. with regards to VNWLo it would therefore be beneficial to engage
artificial stabilization only after nose-wheel lift-off.
If the design of the aircraft features powered lift and/or powered moment (e.g. by
thrust vectoring) there should be no problem with VWL o , as can be seen from Eq. 7.2-2:
Fig.6 and Fig.7 show that there might be no free choice between a levered and a
telescopic main LG design. Whilst at moderate touchdown attitude all requirements of
para 2.1 could be met by both designs (Fig.6), an increased touchdown attitude enhances
a telescopic design. Telescopic design provides for a static wheel position which is
close to the most forward position limit and thus is least detrimental to V.W.
In adverse conditions (e.g. gusty sidewind, low ground friction coefficient) a good
aircraft response to nose wheel steering input is desirable to keep the aircraft on
track, especially if this track is a narrow Minimum Operating Strip on a repaired
runway.
The performance criterio proposed here is "yaw acceleration per lateral friction
coefficient at nose LG". This ratio can be shown to be
W l. *l.
I, WHEELBASE (2.3-1)
Eq. 2.3-1 can be rearranged by substituting the distribution of static load on nose and
main LG:
W
/- WB . FPROPno, • -FPROP,, ,) (2.3-2)
I,
FOP,o,,. • (I-FPPo,,.)
has a maximum of 0.25 at FPROPn,, - 0.5, i.e. at equal distribution of static load,
which is ceitainly not achievable with conventional LG arrangement. However, within the
usual range of FPROP,,, (0.08 to 0.15) there is an almost linear increrse of the
product with FrROP,,,*, i.e. with increasing I,,/WB.
Braking relieves main 1G vertical load. The magnitude of this relief depends on 1G
geometry and on the braking drag force.
It appears that the degree to which max. available braking drag coefficient can be
converted into A/C deceleration is an appropriate performance criterion:
Brake Efficiency BE -n,1#.,
F, -F,_
hence
BE- (F5 ,W o
Since
braking efficiency becomes a function of LG geometry and max. available braking drag
coefficient:
BE - 1 ..
0 /(WB + " ho) (2.5-3)
BE - l-(l. _i + 1 + M - k. h,).
-h)/WB (2.3-4)
)
aBE/dlm.,, - -l/(WB + P -k, " h_ ,
With about 10% of aircraft weight, nose LG is relatively lightly loacie 'tatpr
ic y.
However, load shift during st-dy braing say pjudu e a nose wheel load which is mo~e
tan twice the static load. If brakes are applied rapidly there is an additional dynamic
load increment which in a first approximation may be equal to or even larger than uie
steady load increment. It can be larger because the progressive characteristic of the
usual airsprings effect nonlinear dynamics.
Fig.8 shows computer simulation results for a rapid (0.2 s Brake rise time) and a
slow (1.0 s) brake application. However, unless automatically controlled, slow brake
application cannot be enforced in an operational environment and can hence not be
assumed in calculating design loads.
Eq. 2.3-5 deserves discussion because it offers 6FI,- " 0 for I.... - 0 or h. - 0.
However, h., - 0 is a physically meaningful solution (i.e. low C.C. position) whilst
Io - 0 indicates that A/C weight is supported by nose DG alone and that, regardless
of 08- k,' the brake drag force on main W. is zero.
f - c..'.. + c *.
f - ClWB/t,, + C2 • W) (2.3-6)
where,
C1 -Co..'l.l~/WB + C..i,.-l,.,/WB
and
c2 (= ] + . WB
For conventional WG
2 arrangements the contribution of landing gears to pitch moment
of inertia (term C2.WB in Eq. 2.3-6) will be in the order of 5% to 10%. Hence under
normal circumstances an increase of wheel base will yield an improvement of pitch
natural frequency. Eq.2.3-6 yields f-0 both at WB-0 and WB-.
J l- /(4.-..C2)0.25
0 5
f - f/f=,. - IW " (4"C2"I ).25/(IY. + C2"WB2) . (2.3-7)
W, damping also is apt to reduce dynamic nose WG load increment due to brake initiation.
There is a great many of possibilities to provide LGdamping, ranging from simple
constant orifice oil flow restriction over metering pin through semiactive to fully
active damping force generation. For sake of simplicity we shall consider constant
orifice damping only.
Brake application disturbq ctatic equiLibrium of the aircraft on its WG and initiates
pitch down motion of the aircraft and diving of the nose DG.Assuming constant ID spring
stiffness and A/C mass properties, kinetic energy of the pitching moment will be roughly
proportional to the square of the nose IG load increment6 due to steady braking, and peak
pitch rate will accordingly be roughly proportional to F,.,k.d from Eq. 2.3-5. Peak
strokino velocity of the nos- LG then is proportional to pitch rate times
rI raulic damping force is dependent on stroking velocity squared. The size of the
damping orifices is in most cases chosen on landing sinkrate and/or on repaired runrway
considerations. Therefore it may be assumed that the hydraulic damping coef" 'ert is
not r. '-
L'" . . .a-. - oje W& da..; .o Lv a&4smv
i ,=rortionalLo pe armsoaing velocity squared.
S-7
force to the
As a convenient performance measure one may use the ratio of peak damping
nose LW load increment from Eq. 2.3-5:
F - PI k/6FBr.k,d ~F.-,k.d -
13
F (2.3-8)
./wB
l+p .,k,.'hl WV'
3. CONCLUSIONS
There are some more landing gear functional performance criteria left to be treated
in the future, especially w.r.t. ground roll stability and control with integrated
consideration of ground forces and aerodynamic forces and moments.
REFERENCES
Angle between Vertical and a Straight Line from Main Wheel Ground
Contact through C.G.
P Coefficient of Friction
P Air Density
WAngular Rate
Abbreviations
CG Center of Gravity
HTOL Horizontal Take-off and Landing
LG Landing Gear
MLG Main Landing Gear
MqLO Main Wheel Lift-off
NW Nose Landing Gear
NWLO Nose Wheel Lift-off
)1-9
1.1y :ot-dted o 1.
Fig. 2 MWG Axle Position w.r.t. Pitching Moment from to bet therear
Landing Impact ifthh
LANDING SPEDL
Fig. 4 Result of Dynamic Analysis of
* 0,8Reverse Braking
hxc.2mn
iy *3m
01-
05 0 2
INITIAL ROLLINGSPEEDBACKWARDS
(rn/sI
static
vert.
0 2
0.2
Oh .o.BRAKE RISE TIME
TIME(s)
and
L. lelnersson
Aircraft Division
SAAB-SCANIA AB
SAhI 9 LINKOPING
Sweden
SI MMAARY
lis paper deals with strain gauge measurerients of forcrs acting on the nose gear anti main gears of the coitute aircraft
SAAB SF 34(0 During initial flight test, forces in 6ie longitudinal, transversal and vertical directions were measured for sarous
manoeusres such as take-off, landing. xiing and towing The investigation revealed high transversal loads at the main gears at
1ouch down File nose gear is most severe, strained when steenng during taxiing run and when the aircraft is towed connected to a
tractor with a tow-bar The resuslt froit such initial nicasuretnents formed the basis for a subsetl1eni investigaiton with tin line data
acquisnitn (l landing gear load, on a cssitntuter aircraft in service at Swedair AB The data:icqi isition system and the data analysis
riethods are described in soite dea I Thedata alstucsiitlii was coitinLUIy ameilutu durnigstearlv six months including various
parameers such as different aicrail .eiglt and static laidling carloads. Results frii these in a leltens are presented as
cuniulative exceedances of ongitudinal. Iranssersal aini sertical loads obtainedfrom the rain flow countanalyses perfored oin
linedurng the nteasurements
1. IA, Mt CT101
IlisEs icalls aniin g gear ltad,h,,,chc rc Irsal,d sarel,' incslit csrascn [I[ sni rt until 11 -' , isa r, si c it c ht
1 1
es.lsdhss a main lailine cear Isis sp ir-rnil l 21 Still 25 ' ar'lIter Its lsdinuis , ar i 'ti
one
csxed
of illtrs: s
ii. iiclus t l '
ilitiCntsr lhilit icih Ii e c hmlottccv d hi'1, c lii l
:, 1ii II"
Is n xlecil airliners, the landing gear weight co stitutes about 2-1 per oe,M the lake oifvislit and thelanhing iears itlc
hasr to trasel tlotal di lances in the same orderof magnitude as buses arid trucks do (hiring a tie title In additico. high nusisisiI
sls t are reaLhedat take offand landing 141 Furthermore, the landing gear environment features tetiperaturc cycling. soisture.
Ie c, cg ltiJild oil ishdslots frotn sand which has to be conridered in tie to ,-ue designbessiles the siiii axial I sails in ig I
three h,adstsntptnents have beer defined for tha nose gear ana the main gear. A vertical torquemay also, be included
When densing the ground load spectrum for theSAAB SF-340 commuter airliner little statistics werc as ilahle in the literature
forits i aegiirs- if aircraft Statistcs had to be taken from larger aircraft. e.g. the Airbus A .(Ii measurements 5.1Orstialler arcraft.
cg AR2 i catgors Ifl('onse-uently, this research project was started which aims at pritucing landing
l. geat aladstatis ts or
tonttmuter airliners
This paper firstly summarizes the flight test meisurements of vatious nianoeuvres, performed in order to obtain time histories
of the loads acting on the nose and main landing gears. Secondly. load statistics fron in service measurements obtained during six
months of actual usage are presented.
,train gauges were applied to the aircraft nose and left main gears in order to measure orctes acting on the gears durng ground
operaion as seen in Figs. 2 and 3,
Ten strain gauges forr.ed, on each gear. four half bridgs and one full bridge he half bridge, were mounted on he wheel
hubs in order to measur .trains due to bending of the hul. caused by vertical and longitudinal forces The bridges iredenoted
7
X ZH and Z in Fig 2 and -orrespond, mainly, to the forces X and Z acting on left and right wheel. rsspectivelv. The fnll brdge. -
was miunted on the shock stnut 1,--nn" a shear bridge which was beltev I to give signals proportional to Y loads
\IH gauges were applied below the shock strut to avoid the influence of the varichle length ot tie strut oi the lorc, caulalted
front the Mteasured strains Ilence. the strain signal would 0- almost proportional to the tore beisve its.C and rond
lhe: st'ai gaege siitald were arnalificd and recorded on a tape recorder using ['(' l tihlqtcl " tidFl caiiic siuih 'vie
inpl, d 9 a freqtincnc ot (4l /
Tlie antn ot this flight test was to recoru tie torces acting on the landingt gear dunng take oft. lairl andl ground otcrtiovi
Ct,,eqintly. a careiul calibratin tad to lie pectionied to evaluate the relatio b,,.tween itearld tiaimi, anti aplited io I ; bip.
t
the loads applied duling ihe calibraiton o1ustti intis sced 11 a wray Zhat norresponds io the 1.,ads applied iii tic test on lie ruita
as closely as posib,
Dunn,_, the calibration. the applied load .,as recorded together with the sintuitaueously oceuritg stratus :i all bridges Thus. in
[he general case, a non-iinear syslem ot equaillons were obtained which relates tine force comnponentsi, Iit' dI,,e sivgntds "lhc
alibration loads were applied step, ise and the obtained non linear calibration curves were assumed to h, stepu ise linear between
each calibraticri point This can be ustified due to the very smooth, almtos '!near. relationship between strains and loads. The y5,-
o1
(it non-littear equation w a, ived for each title-step of recorded strais duritg the flight te,t to aield ie desired force coot-
pontents Ioweven, the cross influences froa tine , rce to another strain gauge bridge were negligible as loug as the loads are
saItinietrically distributed between the wheels of each gear and, hence, the bridge signals were proportional, to respectuve force
Sesen calibration cases were carried out for each gear comprising symnaetrical and unsymmetrical forces in Iith positive and
negatie x and y directions and in the positive z direction. It is to be noted that force components (| atid 1: 1lir each asheel were
cs aluated but in imis paper only tlit resultants F and F, are presented together with F
l'he strain gauge instrumentation for tie initial flight test described in the preceeding section u-as sliglils, changed for the in-
,cr ie cmeasurements. In these latter measurements, the half budges for measuring the vertical and liangitidinal forces on each
separate wheel hub in the flight test resurettients were rearranued into two fill bridges recording !h.: !otal vertimal all!t
iir , respe-stivel,, The gauges apalted tor ilhe transversal load. - _e connected eSactly as for the light test ntaeasUmeIncits
I lie '" gauge on the tose gear wascovered w ih iun ebonite protection rig. This rng was applhed bLa INseit w as t'red that tile
,
nose gear shocik strut would ie fuil cotnpressed. during fir instance ... rd orakiogs. and. hence. delitollsi t,.- strain gaugeg
GencrallI. the strain gauges were carefully protected agaist environmental finluencessuch as temperature and tiositire. tollowinu
:he outilnes in Ref 141.
liae strai gauge signals were attplified -,nd processed on-line w ith three SA IFI" ,MAS-rixes i a cmputter based data acquin-
rion systet) istalled in the aircraft uvion.ic rack 1he :rin ith car acint of six cha nc s consilttes 75 kg and measure ;It nnt
as 250 inti x 15() Tint, which fulifil, the naXntull siZe equiremneits due I lititted inslallattou n Iltilie
The on line analysis scheme is the so-called rain- flow count algorithil 17- I lhe si,.e tf !he Markov-tiatices are 64 x<04 in this
case. Clas nutiber tine is the lower liait and tile uptl litat of the tteasutrintg ratges. The itleasurtr ratge' are displa' ed i
Table I together wbth class-widths lie. accunacy) expressed miten, of forces, for eaci of ihe six channels. Ihcse niteasnn2 raniiCe
and also filter frequencies were established using the flight test measurements as a buss.
When data were retrieved, bride- balances were checked and found stabe The strain gauge bridges were also calibrated
lectrically at these instances, lie Markov-iiatirices stored in the data acquisition stsien hardware (RAMi were transfened i a
'briefcase computer' and stored on microassettes. Lastly, the data were transferred to a VAX comaputer for final analysis and plot
ting.
'he loads during the caltbrion were introduced in a way that was expected to correspond to the loads applied in realits- as
closely as possible. The ,erilcal loads were applied by the aircraft's own weight neasu-ed by a scale pisitined under each landITug
gear. The load was then increased by releasing the jacks carrying the aircraft.
The longitudinal and transversal loads were applied with hydraulic jacks and the landing gear being calibrated posed on a roller
bed. Concerning the nose gear, a special device was niounted to the wheels ct prevent the gear from rolling and turing around its
steering ani during calibration. It should be noted that this load applic,,i., whe-a le load is applied over the wheels, is not repre-
sernaivi, of towing which is performed w,h a tow-bar attached to the ends of tie nose gear wheel hub.
During the calibration, the applied load was recorded together with the simultaneously cicuring strains in all bridges The
linearity and one-to-one relation between loads and strains, obtained during the flight test. was confirmed. Thus. the use of bridge
signals without corrections for cross-relations between bridges is justified as long as the load appncation is symmetrical over the
wheels The calibration curves and the linear relationship used to convert bridge signals to forces are given in Ref. 181.
io-S
Ihese cases were performed as tolows. During towing a tractor connected to the aircraft nose gear wheel hub with an ordinary
to, bar towed and pushed the aircraft over a hangar threshold. The threshold was passed at different angles belween the ;winig
paih and tlic threshold. approxittately 90 ',70' and 50'. respectively. Dung the towing and pushing recordings both the nose
tear .adleft ttain gear passed the threshold. lit this case. the treshold consisted of two parallel 40 mnm high steel bars separated 165
Tll
Pic taxiing case consisted of several events is pivoting. synmminetrical and unsymetrical braking, engine run-up and turns. The
csents took place ott the runv,ay The brakings and the steering manoeuvres were carried out at two different taxi speeds, one re
terrsd toas normal taxi-speed and one lower. During thesteering manoeuvres the aircraft followed a zigzag path
Four take-offs and landings were performed. The aircraft status at touch-down are presented in Table 2. The landings, per-
tornied in relatively high crosswinds, can be classified as rather hard landings which would occur seldom in service (every hundred
to every ten thousand landing according to sink speed statistics 161)
the n service nw-isurements presented in this paper were carried out during the time August X. 196, to January 16. 1987,
,.,hi,h was divided into five periods. These periods are summarized in Table 3. At the end of each period, data were read out from
the shipboard data acquisition system and storel. Also, the number of landings and the number of load sheets collected and pro-
essed are displayed in Table 1. together wth static loads derive(' front the load sheets. The total number of landings were obtained
ton theaircraft's flight log. tn Ref. 8,the we:ght, the number of passengers and the static vertical landing gear loaddistribution
ire given for the five periods, as derived from the load sheets. The mean values for each period of the static vertical loads. ctni
itle ground reaction factors described later On inthis paper
Iroot the load sheets data were used in order to obtain
p'uted
ie data acquisition system was connected to a switch which was intended to be manually turnedoff during tossing This
hean,that all ground loading cases except towing were ieat to be included in the moeasuretmetis, i.e. landing. take- ff. braking
and tistellaneous cases such as ground air-ground transition and others, as listed in Ref. 191.
I-trals*, Table 4 shows which recordings that have been included in this study (marked with X). Some data were lost due to df-
tEretlt reassins Regarding transversal loads, the strain gauges were applied to the shock striti which appeared to be unise Firstl.
!1he urfa c is vets hardanti smooth which cotplicates tile application of the strain gauges: the gauge bonding isvcryuncertain
',esoidl, , the seaing used firstly seemed to dissolve in the hydraulic oil.This problem was overcome later. lirdl\. coticerting (lie
oei ear. the shock strut almost closed durinug the o!'eratiion as expected. Ihiwever. despite thic ebouute protection
ring,thestrain
gauges were destroyed since tie ring was twisted around the shock strut. These three circ attstances caused all transversal loaddata
atthe nos gear tobe discarded and, also, the two first ones of the main gear.
Reenring the atmount ,uf lost vertical loaddata of the main gear, it is explained by trouble with a pin plug Also, contact
n
prohieiii during data extracion causrd the lossof Fi and F during period four.
x y
The results were originally presented in Re' 1101as time-history plots and rain-flow cout data in tabies and graphs. Far from
all that data will be presented here but the interest will be focussed upon some of the most revealing results.
Die severity of nose gear longitudinal loads during towing and pushing has been pointed out earlier 191.The high load inten-
,ities emanate from the connection of the tow-bar to the nose gear. However, the results have to be interpreted carefully since the
load application during calibration was different from that vhen using the tow-bar. F measured during towing was therefore found
to be too high, by a factor of about two, as is further discussed in the subsequent section.
For obvious reasons the nose g-" ,-ill be strained at starting and stopping during towing as seen inFig. 3. Furthermore, when
the main gear passes an obstacle, in this case the hangar threshold, a high peak load on the nose gear occurs as seen at the arrow in
10-4
Fig. 3. From a fatitue Doint -f view load ranges rather than individual load peaks are of relevance. Therefore, data may be pre-
sented as in Fig. 4. This figure displays half cycles of ground reaction factors from rain-flow count analyses of the individual
towing records. The ground reaction factor is defined as the ratio beteen the studied load component and the static vertical load ,4i
that gear, e.g. e x . /F-zstat
Figure 5 shows a record of the main gear loads from the zigzag taxiing case ti normal taxi speed. Ihe varittioni in loads die to
steering is clearly seen. By combining the peak values of each individual steering manoeuvre and evaluating the ratio (F -F )/1-
1
a good correlation was found between F and F for the main gear when separating left and right turns as seen in Tabie 5. e di
ference between rig.'., and left turn results dimurtished when the taxi speed was lowered. Rain-flow count ranges of the lateral loads
at zigzag taxiing are presented in Figs. 6 and 7, together with lateral loads at touch down, for the nose and left main gear. The large
values shown in Fig. 6 are believed to be related to cross-relation of loads acting in different directions, as one of the wheels is
lifted off the ground during manoeuvres at large angles.
Finally, a typical landing record is shown in Fig. 8. The three events, maximum vertical load at landing impact C .1.).spin-up
(S.U.) and spring-back (S.B.) can be seen first for the main gear and, slightly afterwards, for the nose gear. In Table 6. several ratios
introduced in Ref. [91 are presented. It should be noted that the load F does not necessarily occur simultaneously as F but at the
same landing. Also, worth noticing is the difference in the recorded d'vnamic F in this investigation and that of fl[91. In the
xS.B.
latter, the dynamics of almost the entire landing gear is included since the strain gauges are applied high up on the shock strut
whereas the present study only includes the inertia of the wheels: the strain gauges are positioned on the wheel hubs. The ratios for
the nose gear seem to be more unanimous than for the main gear. This may be aconsequence of more uniform conditions when the
nose gear reaches the ground, i.e. both main gears in contact with ground. As mentioned above, the lateral load factors at touch
down are presented in Figs. 6 and 7 asrain-flow count ranges. It is clear that the load factor ranges are smaller at the main gear in
general and that steering load ranges at the nose gear is much more pronounced compared to the landing loads than for the main
gear. On the other hand, one shall bear in mind that the static vertical load at the main gear is about 5 times higher than at the nose
gear. This obviously influences the magnitude of the ground reaction factors for the gears under consideration. The main gear maxi-
mum transversal ground reaction factors at landing was -0.53, 0.35, -0.33 and ).39, respectively.
The results presented in Figs. 3-8 were obtained primarily to gain experience for the planning of the in-service measurements
discussed below.
Although the runway conditions ranged from dry or wet summer condi:;ms to 's-r.:e .cnditions ".vih snow and ice no sig-
nificant differences between the spectra for the different periods could be discerned.
For this reason the mean spectra, after removal of unrealistic cycles due to identified problems with the measurements as
described below are presented herein. The number of cycles are normalized to 91) () flights, which is the aircraft design life, and
the range cycles refer to full cycles. The load levels in the spectra represent the middle of the classes in the rain-flow count registra-
tion. As the used instrumentation did not realistically monitor the towing loads these have been excluded in all the presented results.
Spectra for ground reaction factors obtained by dividing the measured forces with the mean vertical static load during each
measuring period are presented in Figs. 9-13 for nose gear x- and z-directions and main gear x-, y- and z-directions, respectively It
should be noted that the actual vertical static load for an individual flight may differ roughly + 10 1 from the used mean value. Due
to the correlation between high loads and high vertical static loads this will result in the ground reaction factor spectra being conser-
vative at the high load end and unconservative at the low load end (with the total effect on cumulative damage being conservaive i
The remaining effects of the interference from Y-loads in the Z-loads measurements, descrihed belo,. will have an additional
conservative effect on the Z-loads spectra.
Figure 9 shows the x-ground reaction factor spectrum for the nose gear after removal of some recorded unrealistic load cycles.
These unrealistic loads are due to towing and nose gear shimmy. As already noticed in the flight tests summarized above the
equipment used for calibration of X-loads on the nose gear was not representative for actual towing during service. Therefore. it
was decded to manually switch off the data acquisition system during towing. Unfortunately, however, this was forgotten in a tlum-
her of cases. For periods 3 to 5 the relatively few unrealistically high nose gear x-loads due to towing are easily recognized against
the rest of the loads in the spectra (spin-up, spring back and runway roughness) and are excluded front the spectra.
For periods I and 2 the numbers o unrealistically high nose gear x-loads are so high that they cannot be explained as towing
loads only During these periods nose gear shimmy was, however, reported. It is reasonable to assume that the measurements will
give exaggerated X-loads for this condition. analogously to the exaggerated towing loads measured when the nose gear passes a
0 .
threshold at an angle other than 9 * The frequency 15-20 Hz during nose wheel shimmy would also produce the registered num-
ber of cycles.
10-5
As the instrumentation is judged to give unrealistic loads for this condition, and as nose gear shimmy should normally never
occur in-serice, the nose gear X-load data from periods I and 2 are discarded.
Figure 10 shows the Z-ground reaction factor spectrum for the nose gear, also after removal of some unrealistic load cycles.
The presence of towing and nose wheel shimmy during the measurements does not appear to significantly affect the nose gear Z-
loads.
The nose gear Z-loads spectra for periods 2 and 3 look very different from those front the other periods with an unrealistically
high number of cycles (roughly 1000 per flight) between the static ground load and zero. This is caused by a large number of trough
values being registered as zero rather than as some positive value. This moust evidently come from an intermtttent lack of electrcal
contact dunng ground runs and the cycles are thus removed from the spectra. The relatively large nutmber ot cycles to zero for
periods 1. 4 and 5. which are not cycles to the static ground loads are believed to be due to bouncing of the nose wheel due to the
combination of a relatively hard shock absorber and runway roughness. This effect has been reported for other aircraft also, see
Ref. [51.
The Z-load spectrum for the nose gear (and to a lesser extent for the maTn gear) contains some unrealistically high negative
loads. The only negative Z-loids that occur in service are the toassloads when the shock strut reaches its stop In extension.
FAR/JAR 25. '7 'T.ebound landing condition" specities a limit load factor of 20 g acting on the unsprung weights of the landing
gear tor thts case. Assuming that these requirernents represent the maxinum realistic condition, and considering that only
massloads front weights outboard of the strain gages will be measured by the FFA-instruttentation. results in a limit for realistic
negative values in the measurements of -3.6 kN for the nose gear (-11.3 kN for the main gear). Cycles with larger negative values
are thus removed from the spectra.
A discussion of the origin of the unrealistically high measured negative values is needed, however. Unrealistically high nega-
tive loads were present in the spectra from all the periods and the numbers were roughly the same as the number of X-loads front
towing. It would thus be tempting to explain the loads as vertical loads introduced by towing but the time histories from the initial
flight tests for towing and pushing across a hangar threshold gives no support for this.
A factor that may sometimes strongly affect the accuracy of the Z-loads measurements is that the strain gages on the wheel
hubs, calibrated for Z-loads, will measure bending due to Y-loads also. In fact the leverage from the Y-load at the wheel-ground
contact point to the strain gages is roughly 3 times that for the Z-load. As long as the Y-load is equally distributed between the left
and right wheels of the gear this will not affect the measured Z-loads. For cases with an unequal distribution of the loads, such as a
banked drift landing, significant errors will result, however, The difference in loads between the wheels is likely to be greater for
small Z-loads than for large Z-loads and can explain the unrealistic negative loads which were measured.
The surprisingly high positive Z-loads which have sometimes been registered are believed to be mainly due to the nose gear
touchdown after landing. This is supported by the fact that the nose gear shock strut was sometimes almost fully compressed during
the measurements.
Figures 11-13 show the ground reaction factor spectra, for the main gear. in the x-. y- and z-directions, respectively. The only
editing perfortmed on the rteasured data is that cycles with z-loads outside of the litimi or realistic negative saluesl- 11.3 kN), ee
uiscussion for nose gear above, are removed from the spectnrn shown in Fig. 13.
,
Finally, a comparison between the ground reaction factor spectra, normalized to the aircraft design life of 9 (011 flights, for the
nose gear and the main gear is shown in Fig. 14 for X-loads and in Fig. 15 for Z-loads The ground reaction factor in these spectra
are significantly higher for the nose gear which agrees with the results fron other aircraft, see Re, [5 1.
5. A( KNOWLEDGENlENTS
Fit.ancial support from the Swedish Board for Technical Development, the Swedish Civil Aviation Administratitn and SAAB-
SCANIA AB is gratefully acknowledged. The authors are indebted to Bengt Johansson arid Rickard Vinthagen at SAAB-SCANIA
AB. to Karl-Erik Viklund and Flans Karlsson at SWEDAIR AB, and to lo Norrbont. Ivar AhlstrOm and Inez Ingstrom at the Aero-
nautical Research Institute of Sweden. for helpful co-operation during the course of this work.
6. REFERENCES
[I I Gustavsson. AT: "A Literature Survey of Ground Load Statistics for Landing Gear Fatigue Design Purposes". IFTA TN 1984-
9. The Aeronautical Research Institute of Sweden. Stockholm, 1984.
121 McBrearty. J.F.: "A Review of Landing Gear and Ground Loads Problems". North Atlantic Treaty Organization Report 118,
1957
131Campbell, G.S.: *"A Survey of Serious Aircraft Accidents Involving Fatigue Fracture. Vol. 1. Fixed-Wing Aircraft",
Aeronautical Note NAF-AN-7, NRC No 21276, National Research Council Canada, April 1943.
141Ladda, V., and Zaschel, J.M.:' "Long Time Measurements with Strain Gauges to Investigate the Landing Gear Loads on A~rbus
A300". (In German: "Langzeituntersuchungen toit Dehnungsmesstreifen cur Ermittlung der Fahrwerkskrafte am Airbus
A3003", Messtechnische Briefe 15,Heft 3,1979.
11Buxbaum, 0., Ladda, V.. and Zaschel. J.M.: "Operational Loads at Nose- and Main Landing Gear of an Airplane of Type
Air-bus A300 B2 during Normal Service of the Deutsche Lufthansa". (in German: "Betriebslasten am Bug- and Hauptfahrwerk
etnes Flugzeuges von Typ Airbus A3() B2 wahrend des Einsaizes bei der Deutschen Lufthansa '1, Fraunhofer-Institut fur
Ben-iebisfestigkceit (LBF) Darmstadt, Final Report No. 3691. M~tch 1981.
161 Christian, R.D.:"FAR23 Fatigue Substantiation Procedures". GAMA Committee Report SAE 710403, March 197 1.
171deJonge, .1 :~ The Analysis of Load-Ttme Histories by Means of Counting Methtods". NLR MP 8201WU, Report. National
Aerospace Laboratory NLR, the Netherlatnds, August 1982.
181 (ia'tavsson, AlI.: "in-Service Measurements of Saab SF:341 Landing Gcar Loads". FFA 'TN 1987-48 The Aeronautical
Research Institute of Sweden, Stockholm, 1987.
91 Buxhaum, 0.: "Latiding Gear Loads of Civil TIransport Airplanes". Proceedings of' the I1th ICAP-Symnposium held in
Noordwijkerhout, The Netherlands, ICAF Dv' Nn 1216, May 19831.
1101)Gustavsson. Al. and Johansson, B.: "SF-341. Development Ground and Flight Test Report - Study of Landing Gear Loads".
SAAB Report No. 72GTS4322. SAAB SCANIA AB, Linkoping, Sweden, 1985.
Table I Summary of measuring ranges and accuracy Table 2 Landing conditions during
during in-service measurement, flight test measurement.%
Table 3 Summary of the measuremtent periods with corresponding static loading conditions
Table 4 Showing which recordings that were successful (X) and Table 5 Relation (F.z - F z,stat )/Fy of left main gear durng
which ones that failed for different reasons (-) zigzag taxiing
m n
Period F F Fm F F" F" Turning Normal speed Low speed
x z2 Z ditection Mean Standard Mean Standard
1 X X X X deviation deviation
2 X X X
3 X X X X Right -0.86 0.12 -1.06 0.15
4 - - X X Left -1.28 0.13 -0.94 0.13
5 X X X X
No. F F F F F
x,S.U. z,L. I. x,S.U. z,L.. z.L. I
11 0
6.00 extowi ng
5.00 path
400 4 .00 threshold
3.00 K i lthreshold passg~~
5s.00
3.00 50"OS
2.00 ~ , 2.00
1.80 -threshold
1.60 passing goo
/ U
t.0OO0
1.20 ek
1.0
0.80-------------- Z
0.60
0.4,0 0.
Wn 0.00 .. . -
0.25
M 00
OIS 500
0.5 3.00
2-00
-1.00 ' .
0.05
Figure 3 Time-histories of towing over a Figure 5 Main landing gear loads at normal
hangar threshold at 900' angle speed during zigzag taxiing
between towing path and threshold
2.5 kOS 6.00 "101
Zig-zag, "normal" speed 500
2.0 3.00
2.00
1.5 00.00
-i.00 5s
-2.00 - s
1.0 -3.00
Landing
L 3 kN3.00 ool
N0 7T1- 11111
4 2.60
22 20
0.5 2.00
1.40 f J,\
00 1
100 1000 8.640 1., _______________
1 10
0.00 -- l
CUMULATIVE FREQUENCY PER RECORD -0.0
-0.4
Figure 6 Cumulative frequency distribution of range- so
pair ranges of transversal ground reaction I___________________
factors, e ,at nose gear at zigzag taxiing -1AOr-401.,I
e U 2.00
.Landing3
0.5- 1- 0 00
anding 2 03
0.3
0..2
0. 81.......
0.2 ....
-0.8
-1.0
-.2... .
0.2
Range exceedance
.0
4.4 ........ ... ............. ...... ......... ......... ......... ........
3 ..0 . ...... ......... .... ... .... .... ..... ........................ ....
1.0
, 5. c ........ ...........
1 02. .... .. .... -e o k d*T roug h ex c e e d bn c ii
3.0 -
.......
35.... . .. ......... .... .. . . .
1.0 - -. .. ......................................... .
r. r
4 7
100 101 102 10 10 10 106 10 10
Range excesdance
0.2
to6
2 4
10 10! 10 10~ 10 10 o7 108
-0.2 -. ...
-1.40 ......
1.2
0.8 ....... .
0.2 ....................................
0.0
2 4
100 101 1n 103 10 105 106 107 108
Range exceedance
-0.2
-0.4
-0.6 ...........
-0 A ........ ..
-1.0.. ..... ..... ..... ...
Mo. Po 1. 6826
Min Trough .1290
2.0 .
0.8
0.6 . ....
0.4
0.0
10
-0.20-
101 10 1b3 jo ~ 10 1 06 107 1 08
1.8
1.4
1.2 - ..... .
1.0
0.8 -.....
0.6 -. . . ..
0.4 -...
7
100 101 102 103 104 10 106 10 108
Range exceedance
0.8
0.6
0.2
-0.2
10.6 -
14 -
1,2
0.8
0.6 .
04
02
00
100 10, 102 103 10 105 106 107 108
Range exceedonce
Figuare 14 ComzpanA~n between spectra for nose gear and main gear fm 0066X) Il izhti*
'1(1ing eXcludedt X -grund reaaon factor. FFA in- srx -cc nicaxurenuent,
nose geor
5 .0 .....
.........
2.0
13.5
1.0
. . . . . .... eko~Touhecea
500
45
4.0
3.0 -
25
2.0
1.0
3 4
100 101 102 10 10 105 106 107 108
Range exceedonce
FilvreIS (>mmpanismn beweo specir3 fortnow' ge~aranti mam giir for90 M1() fighi,
liing -xclsided. Z. grsmundreationl factor. FlT-Ain- service Iliasirtnent.
11-I
by
SUMMARY
Prior statistics of airworthiness authorities indicate that landing gear (safe life design) often fail during
scheduled aircraft service. Therefore investigations have been carried out during the last two decades in
Germany with the aims to determine the operational loads acting on the landing gear during service and to
define the load cases which have to be taken into consideration for fatigue investigations related to landing
gear an~d airframe.
Statistics about failures on landing gear for civil and military aircraft, relevant load cases as well as
information about essential fatigue requirements are presented thoroughly in the first part. In the second
part some results of different landing gear load measurements ar' compared and discussed. These results
presented in form of cumulative frequency distributions for the load cases taxiing aind landing impact
originate from the following measurements:
In the third part it is reported about the impact of towing and push back operations on the nose landing
gear using conventional and advanced towing methods. In the conclusion the primary results of the landing
gear loads measurements are accentuated, examples for the disposition of landing gear fatigue tests are
considered and essential future actions concerning load monitoring activities including hard- and overweight
landing detection are reviewed.
1. INTRODUCTION
The only components of the aircraft that are categorized as safe-life are parts of the landing gear [ 1 1,
since a completely fail safe designed landing gear structure (no multiple load path design possible) as well as
a damage tolerant designed landing gear structure (sufficient damage tolerance qualities can not be achieved
because the detection of cracks is more complicated than it is for other aircraft components) are inapplicable
up to now. Therefore the landing gear in its entirety is safe-life designed and represents a life limited
structure. In opposite to the damage tolerance design cracks in the landing gear structure are not admiss-
able. During fatigue tests the safe-life requirements under consideration of proper scatter factors (for
landing gear nearly a factor of 5) have to be demonstrated.
Normally the landing gear have to be replaced before the aircraft reaches the Economical Repair Life
(ERLI. Due to runway roughness and maneuvers, (ground operations, take off and landings) repeated loads in
x, y, and z-direction and torsion moments are acting on the landing gear simultaneously.
As a result of large leverarms great bending moments will be induced in the airframe. Additional severe
environmental conditions (skydro, salt, water, large gradients of the temperature etc.) lead to corrosion
problems in some cases.
The landing gear therefore represents a highly loaded structural component. This is the reason why it is
necessary to use ultra high tensile steels for the axles and legs. But this is combined with the disadvantages
that these materials are sensitive concerning fatigue behaviour. Partly this could be avoided by introducing
residual compressive stresses in critical areas 1 2, 3 1.
In ( 4 1 has been summarizt that up to 82 percent of all fatigue failures on landing gear are caused by
taxiing. Even changes in aircraft 'ight within typical service conditions have practically no effect on the
fatigue life of landing gear but an icrease in taxi speed from 25 to 50 km/h on the same runway roughness
leads to a 35 percent increase in fatigue damage to the landing gear
In a contribution about air transport landing gear maintenance from an airline point of view it was noted
that the general design of landing gear usually is very good but the details of the gear create all sort of
problems [ 5 1.
The distances covered by landing gear on taxiways and runways are enormous (assuming that the landing
gear is not replaced before reaching the ERL). Taken into consideration that during 1 flight 8 km will be
covered on ground, a total distance of nearly 400.000 km for an short haul aircraft will be accumulated. A
distance which normally will be reached by commercial vehicles.
As mentioned above the reasons why landing gear could fail have been confirmed by prior statistics. They
come to the conclusion that landing gear often fail during scheduled aircraft service. Therefore during the
last two decades in Germany investigations have been performed to determine the operational loads acting on
the landing gear and to define the load cases which have to be taken into account for fatigue investigati n..
In "his contribution results of operational landing gear load measurements performed on aircraft A300B2, A310,
A320, VFW 614 and F-104G will be presented and discussed.
11-2
1. S
qtatistics of landing gear failures during service
In the MIL Specification a failure statistic from the period 1968 to 1978 has been published. In total 328
structural failure accidents have been analyzed. 65 percent of accidents have occured during ground
operation and 44 percent of them during landing touch down and landing roll. A breakdown of accidents
shows that 50 accidents were attributed to the landing gear structure, see Fig. I. Six of them were attriblted
to fatigue problems, that means 8 percent, see Fig. 2 I 6 1.
MA iERIAL 44 5 11 3 i 3
1 NADEQIArtM IG, I i 0 O I o
PIO 1968 to 1W NOT IN A' HIiANIi %TI1 2 o
0 . i2
1ThCiiNiAL ORDE
SNTCNAOWN 3 I 0 0 0 a
FACUR1
t28 3P2 IOEAKI)Ow OF OF F0ILBE
CIORROSION i . I Z 7 5 0
MAJOR 41% UI'RSTREN(GTH 7 07 0 o 0 0
CRACKEr 6 6 I 3
MR S% EfQL A 0 3 i 0 Z a
O%,LOAD 5 1 2 i
0
LANDING
RiLL I I I i 0 4 3
A similar result shows an investigation of 529 fatigue cracks that have occured during service of military
and civil aircraft in the same period. 5 percent of them were detected on landing gears 1 7 1.
An analysis of serious age related airliner failures in the period of 1977 to 1987 shows that 18 airline
accidents of civil aircraft were caused by landing gear failures possibly as a result of fatigue and corrosion
problems f 8 1.
A statistic of 95 aircraft accidents during scheduled passenger flight in 1988 implies that 6 accidents were
attributed to structural landing gear failures [ 9 ].
Essential guidelines how to fullfil the fatigue requirements have been choosen from the FAR-AC 25.571-1B
for civil transport aircraft.
Fig. 3 shows the definition how to handle safe life structures and how to establish the typical loading
spectra. Fig. 4 gives information how to perform the fatigue evaluation, to establish the scatter factor for
safe life designed structures as well as their replacement times. The specification MIL A-87221 describes in
similar terms the fatigue evaluations 1 6, 10, 11 1.
lISMA ES
THETYWA LOADIG SPMCTRm EDIN
THELOADINGPEC 14 SOULD BE BSD ONAWASJ. diitoull it thni ia nton if Intalats pd"iia a eut dtlgnio
DATA.- DAA
niAiittliAL - oppioso
byn-
fn ndlil I insnm. ,ofa rw 1- t ci, If of
ilmt 1ian ,iii aena
m in tnlo it d it naIhn
opt oi
rtntehd md It n lnm mayn nsna ioiidnioim
THEPIOIPIAL LOAMST4ATSHOULD
BEO E £3 - c RPj 1 M Pp-motninl lm 1y1d tbimtd 1 lilhtitd
-an .y-
I i ond n 020 i
-tdlod dnfoni ii. to nilian*din n p aid imnd*r
OOC LOADS
(IA ,) LAS Wfh3 , T 2i' 1 i529iThvoe
teplan, tto in to tndd if dnlou dab ond~oaimn
E 3WEIAUFT INI TRUST
- REVE AND 0,6 l SElF. RE.-" E"- S 10
P10.3 ESSENTIAL
FATIGUE RIEOIR~MENTS ( 10,111 P10.4 ESSENTIALFATIGUEREQUIRE~iNENTS
110 1
11-3
For fatigue evaluations more than 16 different loading conditions have to be taken into account ( 1, 20, 21,
22, 23 1. Fig. 5 shows the loading conditions. As a result of the landing gear load measurement on A300B2
[ 12 1 the different loading conditions have been confirmed but the push back load case has been observed
during the measurment for the first time.
Fig. 6 and 7 indicate the sources of the data that will be the main part of this contribution. The most
important data concerning statistical reliability is represented by the A300B2 landing gear load measurement,
with the investigation of about 2200 flights as well as the A300, A310 towing and push back measurement. An
important part in this contribution results from the landing investigation of A320. The contribution will be
completed by the VFW 614 landing gear load measurement as well as by results of a fighter aircraft landing
gear load investigation [ 12, 13, 14, 15, 16, 17, 18 1.
FG6 TESTAIRCRAFT
2 ¥IM614 I.ANDNt, GEAR LOAD MtSLREM'{2 I I I I. UQO.A.IQ TOWIN ) rus BAiK 5EASIREMENI' 3I
'FF0.OU*p. NEARLY( OMPARARI.S
T 1-P011O5 SCHEDrIEDSERVI('FOF I H A1) AF
NORMAL.
SERVICE 64 30 01 1
'F F11If': 104 1 TOWAR HIGH .EFTRAITOR
III 10C F 1IFI r MAI11: 62
The sign convention of the loads is given on Fig. 8. In the direction of the arrows the loads are defined
as positv.
Fy7 Fy F
F. F F FF 1 F, A-
2.3.1 General
The landing gear loads have been measured by st- ., gages calibrated and combined to produce pure
vertical, lateral and longitudinal loads using Skopinski's method J 13, 14, 19 ] cn nose and L/H main gear.
A typical example of the measured ",ad time histories during landing and taxiing is shown on Fig. 9. The
maximum values during landing 1.ave been marked. These maximum loads have been analyzed by means of the
extreme value distribution ' 4 1, the loads during taxiing by means of the statistical counting method level
crossings[ 25 1.
MAIN GEAR
_ _ _
_ _ _
__
01F,
F..
F,
2.3.2 Landing
The acceleration of the landing gear wheels to a circumferential speed which corresponds to the aircraft
horizontal landing speed causes a longitudinal drag load, the so-called spin-up load. Caused by the energy
built up in the main gear this leads to a so-called spring back load in form of an attenuated oscillation. In
reason of the elasticity of tl' landing gears side loads occures during landing. Due to the vertical sinking
speed of the A/C at the moment of touch down a landing impact in vertical direction comes off. Distributions
of the sinking speed for different types of A/C are shown on Fig. 10, from which the vertical load can be
derived.
The hatched scatter band represents operational data for transport aircraft 1 26 1 and is in agreement
witl. ( 27 1. The distribution marked by 1 depicts the spectra predicted for A320 as well as DC1O [ 26, 28 1.
Curve 2 results from an A320 landing gear investigation performed during flight tests, it lies in the upper
region of the scatter band but under the predicted one. Curves 3 to 5 show landing sinking speed spectra
for different military aircraft f 6 J.
11-5
0LMS0 S EA 0f5555
12
Oi - -
o- 0
On main landing gear equipped with tandem wheels (bogie gear) torsion moments directed to the vertical
axis of the plane become evident because the spin up loads will never act on the four wheels simultaneously.
The description of the loads acting on main landing gear as mentiond above is valid in principle also for
nose gear. The distinction is that the nose gear loads will be mainly influenced by the counter rotation of the
A/C during landing due to the elevator deflection initiated by the pilot.
In total 2200 landings of A3OB2 and 80 landings of A320 have been analyzed. During 76 landings of A300B2
a second characteristic landing impact in vertical direction was discovered on the main landing gear. On the
nose landing gear additional vertical landing impacts were found during 972 landings. During one landing a
sequence of 10 nose gear impacts have been observed [ 12 ].
The statistical results presented in form of frequency distributions for both the main and nose landing
gear of A300B2 compared with those of A320 (for A320 were available only drag - and vertical loads on main
gear and vertical loads on nose gear) are shown on Fig. 11.
The main gear results indicate that the spin up loads as well as the vertical loads for A320 are appreciably
higher than for A300B2 while the spring back loads are comparable in magnitude. Possible reasons for the
differences could be that the A320 has a "hard landing" characteristic based on the oleo spring of the main
gear combined with a relatively high breakout force and that the results have been taken from flight tests
that means severe service conditions. The vertical nose gear loads of A320 are in accordance with those of
A300B2.
Comparing main- and nose landing gear loads together it is obviously that the spin up loads as well as the
aide loads on nose gear are relative higher. This knowledge is important since, in the past, in the absence of
nose landing gear load measurement data, the main landing gear data has been assumed to be valid also for
the nose landing gear. Another important fact for the nose gear is that 1.5 landing impacts have occured per
landing as mean value.
2.3.3 Taxiing
The frequency distributions determined for the load case taxiing contain loads caused by runway rough-
ness as well as by ground maneuvers (turning, braking). The sample size of the basic data for the analysis
were 300 flights of A30012 and 104 flights of VFW 614. The results of A300B2 have been compared with those
of VFW614. From the VFW 614 measurement were available only drag- and vertical main landing gear loads and
side- and vertical nose landing gear loads, see Fig. 12 [ 12, 16 ].
The comparison indicates that the main gear drag loads of the VFW 614 are much lower than those of
A30012. The reason could be that the braking loads of A30012 are clearly higher during service. It has been
observed that the A300B2 pilots in general prefer to take the first exit after landing if possible. The vertical
loads however are comparable in magnitude and frequency.
The comparison of nose gear results show that the side loads of the VFW 614 are slightly higher while the
vertical loads are lower than those of A300B2. The lower vertical loads of the VFW 614 could be explained by
the higher main gear braking loads of A30082 because braking loads on the main gears create an increase of
the incremental vertical nose gear loads. In addition Fig. 13 shows a separation of the A300B2 taxi spectra for
all different load cases.
11-6
I-0-
0.2
0.2 0.2 -
0 20.
10" 10- 10" 1 10~
-3 10' i'
FREQUENCYPER LANDING
FREQUENCYPER LANDING
7---
o.6
i.
o Ioo , oo t o t o t* , ~ t o t
1. lo
~ ~ ~ ~
taxiing~~~~~ atr ~ ersetgon
wer deie.Tela ecinfcostatmastela&aerltdt
t vetia
0e stti ladn gerladf1, 8
F,, y.
1 . 00
lo
.6W
FIG.14 FREQUENCY DISTRUTnONS OF'THELOAD FACTORS FIGi 5 FREQUENCY DISTRIBUTIONS OF'THE VERTICAL
0ONTHEMAIN LANDING GEARDERIVEDFROM A LOAD FACTORDUJRING TAXIING OF'TRANSPORT
FWQTERAIRCRAFT DURIG UNRACED TAXIN( AIRCRAFT [6,12, 16, 29 1
[17,181
2.3.4 Turning
To establish the turning spectra (pivoting is included) for the main and nose landing gear the measured
side loads have been low pass filtered (Fy main gear = 0,25 Hz, Fy nose gear = 0,35 Hz). The turning spectra
separated for pre flight and post flight turning are shown on Fig. 16 and 17. The number of turnings can be
taken from the spectra with 4 during pre flight and 6 during post flight taxiing, see Fig. 17 [ 12 1.
Fig. 18 gives the turning spectrum derived from a measirement on a fighter aircraft. The load factor n
represents a ground reaction factor that means that the loal is related to the static vertical landing gear
load [ 18 1.
2.3.5 Braking
For establishing the A300132 braking spectra the load time history of the drag loads has been low pass
filtered by 0.25 Hz. That means all load variation over 0.25 Hz were cut so that only the braking loads
remains. The spectra are presented on Fig. 19. The numbers of brakings were determined with 3 pre flight
brakings and 3 post flight brakings [ 12, 301.
- A3OflB2]A 1
* 0.24 -- - 02 .-----
0- 0.2~
O 00
-0.2~~ -- 0.2-
A300 82 A300 B2
0.16 - UN!0.16 . lHTR
0 0
0
0-
LOADOI rACTO 0. .
PRSOOERCNO OR PER
[
O 0
A300B82 A300 B2
"-0.20 0.40
A braking sp ctrum derived from measurements on a fighter aircraft is presented on Fig. 18. It shows that
per flight 10 brakings have been observed. The load factor n. is a ground reaction factor that means the load
is related to the static landing gear load in vertical direction 1 18 1.
Fig. 20 shows a correlation between A300B2 braking loads occured simultaneously on left and right main
landing gear. The relation indicates that exact symmetrical brak'ng occures seldom because brakngs during
taxiing mostly are combined with turnings by supporting the nose gear steering [ 12 1.
Fig. 21 represents the correlation between braking loads on the left main landing gear and the load
increase in vertical direction on the nose landing gear. It implies that the braking loads on the main gear
cause an increase of the vertical incremental load by a factor of 1.8, that means FZNLI = -1.8 F1 112.
( I L ... -
0A
A.300 0 -2 IF0 -a o 0 - 3
A300 B
..
o 3
0
4 o-2 o 8 0080 ... o 0 00 PI 0
X 0
O0, 0 0 0
00,2. 0 00 O Y1'METilC52
.04 A0 0 Bo,AAN005
0
0
0 . z 0.
TO f 0I'' 1 0.0.
0,0 -0 , - 02 -0.3 -T4 0,0 -01 -0.2 - 0.3 - 0.4
Fs. L4H RELATEDTO UMIT LOAD F X LIH MLO RELATEDTO UMIT LOAD
FIG.20 RELATION BETWEENBRAKING LOADS ON THE FIG.21 NPACT OF THEBRAKING LOADS ONTHEMAIN
IJHAND RM MAIN LANDING GEAR [12] LANDING GEARTO THEVERTICAL LOADS ON
THENOSELANDING GEAR 112 ]
2.4 Statistics of ground operations
As shown at points 2.3.4 and 2.3.5 the number of turnings were in total 10 per flight and the number of
brakings 6 per flight. These figures have been determined directly from the turning and braking spectra,
respectively.
In addition the taxi times have been derived for A300B2 and VFW 614 in form of frequency distributions
see Fig. 22, The results are comparable for A3002 and VFW 614. The mean values for pre flight taxiirg are 6.7
min for A300B2 and 8.0 min for VFW 614. The post flight taxiing times are 4.0 min for A30012 and 6.0 min for
VFW 614 [12,161.
Further more some statistics of A300B2 landings have been established. Fig. 23 shows the distribution of
time differences between touch down of the right and left main landing gear. The distribution indicates that
during every fourth landing (25 percent) a two point landing has been observed. The time differences vary
between 0 and 2.5 seconds (L/IH gear first) and 0 and 2.0 seconds (R/H gear first).
Fig. 24 represents a distribution of time differences between touch down of the main landing gear (which
ever comes first) and the nose landing gear. During 70 percent of landings the time differences vary between
6 and 10 sec. The max. time difference detected during the measurement was 20 seconds 12 1.
\\ \ NNC 2
0-,o
MTOT TAXSWG
PB
POSTR.X0TTAOrNO
STRIE
TASIN
i • ST FL*WT. "i
90
I -
1 6-
0,-
2 3 456 8 10 20 3040 60 0(0
TAXI TP IN MI
2
2.0 1A,0 002 A30 2
10 <
0 0t4
-2.0 -1.0 0.0 1.0 2.0 1 1 1 1 1 1 0
ENCES IN SEC
TIME DIFFEr
TIME DIFFERENCESIN SEC
Modern civil airports are equipped with boarding piers.This requires a parking of the aircraft at or near
the pier, so that passengers can leave the aircraft directly to the terminal. After boarding a special maneuver
is necessary, a so-called push back. That means moving the aircraft backwards to a position from which the
aircraft can taxi to the runway under it's own power ( 32, 33, 34 1.
The towing procedure of aircraft from the terminal to the hangar and vice versa normally will be per
med by towbar and tractor. To save fuel, reduce noise and shorten the towing times (compared to towbar
towing) in the last years the so called high speed tractors were designed. The maximum towing speed will be
about 50km/h. It is planed for the future that the tractors will also be used to move the aircraft from the
terminal directly to the runway and back after landing. Examples of the different towing methods are
presented on Fig. 25 32 ]. o
PUSH- BACK OPERATIONS A300
1 06 A310
0 M *S TPUOACT4
-. 0,2 TWCONVENTIONEL
MET- 04
-C .----.-.
1 10 C le I
o(Hoo OIG MEFH;
oDig
TOWIN OPERATIONS
+06 T0 .
ADVANCE d .. meSKE 05OCT54
TOWNG5ETHOD 0
02
~~o~oo~ooL 2
(HiOISpss. owogI -
-0.t
10SJ~C
0 08X Iowa0
The importance of these load cases for the nose landing gear first were presented by the results of the
measurement on Airbus A300B2 [ 12 1. Normally the nose landing gear of the aircraft is unbraked but during
towing and push back - the nose gear is coupled to the tractor - the nose landing gear acts as though it
were braked, caused by braking and acceleration of the tractor driver. This leads to further towing and push
back investigations during normal aircraft service of Deutsche Lufthansa at Frankfurt and Air France at
Paris, CDG.
Fig. 26 show the frequency distributions of the drag loads on the nose landin.g gear using towbar and
high speed tractor. During push back the measured drag loads are comparable in u;Rgnitude and frequency.
The measured negative loads are in proportion to the nose gear spin up loads. During towing the loads are
higher using high speed tractor. The loads are influenced insignificantly by the aircraft mass but severely
by the acceleration and braking characteristics of the tractor.
Recently during a push back maneuver of an ATP aircraft at Bremen airpo t the nose gear and the fuselage
attachment were havy damaged ( 32 ].
1-13
3. C- CLUSION
In this contribution a lot of results taken from long time measurements of different aircraft have been
presented. It is to be hoped that the contribution clarly :ndicates that the landing gear as one of the mast
important components should not be neglected and that more attention should be paid to the landing gear
design including fatigue evaluations. The different loading conditions of the landing gear as well as the real
loads acting on the landing gear during service could be shown.
These facts have been elucidated not only by the failure statistics, but also by accidents and problems
caused by landing gear of modern aircraft in the present. A surprising result was that the magnitudes of
putsh back loads on the nose gear are in proportion to that of the spin up. This is important because the
loadintg condition spin cap is a design case for the nose gear up to now.
To make it easy to use the results for further treatments or transference for new design of landing gear
and airframe structures the operational data have been presented in relation to the limit loads.
Correlations of the landing gear loads (simultaneously occured) have not been presented in this contribu-
tion. In [ 12 ] results of correlations were investigated thorouighly. From that it can only be noted, that a
definite correlation between the loads does not exist. But depending on different frequency bands a better
,orrelation becomes evident.
Ihe comparison of A300B2 results with those of VFW 614 shows in general a good accordance although the
A3OCR2 measurement only represents a scheduled in-service investigation.
The results of A300B2 for the loading condition landing impact show, that the nose gear is relativ more
highly loaded than the main gear.
The push back and towing measurement was the first investigation to establish load spectra based on
actual airline service.
All spectra as presented are transferable to military transport aircraft and the spectra derived from
fighter aircraft to combat aircraft, respectively.
Based on the results of the A300B2 landing gear load measureme . the fatigue test set up has been
revised and completed at that time.
H- a typical fatigue test set up for the main gear of a transport aircraft could be disposed is shown on
Fig. 27. It represents a typical, concerning the frequency simplified (compared to the real loading&) flight by
flight test spectrum for the main landing gear. It is separpted into the take of ind landing spectrum. The
different load cycles are dependent on the loading conditions. Landiri gear actuator loads (simulation of
retraction and extention) also will be applied.
em I i mm
I i
aiiiI
Cm i i, I i im
1- -tI
i °I
05,oiI i i I I I
A~
was;,o~o I A!4
em I i :l
I loin
I 8 M i I
"c..01st years some activities concerning load monitoring of the airframe during aircraft service have
en de, eloped.
A DA-proposal represents the Operational Loads Monitoring System, OLMS for A/C A320 which is under
development. The working principle of OLMS is shown on Fig. 28. It indicates that OLMS only is the monito-
ring device integrated in the wholp Airframe Condition Monitoring Procedure (ACMP). The aims of OLMS are
the determination of load _pectra of all aircraft components and coi relations between loads occured simulta-
n-ously [ 35, 36 1. In addition a erecial function, the event identification during landing and flight will be
performed. That means detection i' hard landings and limit load axeedances.
-- ~ ~ ACE
R
011M,-] . 0P5R-S.
FV
HP
1Al,1 __ AP
H- D- 1 --
It is intended to incorporate the landing gear completely into the monitoring activities in the near future.
These will be solved when in connection with the development of the A320 Weight and Balance-System (WBS)
proper landing gear sensors will be available.
4. REFERENCES
1] J.E. Wignot
Structural Design of Transport Airplanes for Transient Environments
Lockhead-California Company
Burbank, California
ASME-Paper-AMD36, 1979
F2 ] Young, D.W.
Aircraft Iaiding Gear
The Fast, Present and Future
Proc. Inst. Mechanical Engineers
Vol. 200 No. D2, 1986, Pages 75-92
F3 1 Jenkins, S.F.N.
Landing Gear Design and Development
Institution of Mechanical Engineers, Proceedings, Part GI- Journal of Aerospace Engineering
Vol. 203, No. GI, Pag- 67-73, 1989
5 Witt, W.W.
Air Transport Landing Gear Maintenance United Airline, Inc.
Nationai Aeronautic and Space Engineers and Manufactoring Meeting
Los Angeles, Calif., 1965
SAE-Paper 650842
7 1 H. Huth, D. Schiitz
The Collection and Analysis of Fatigue Damage Occuring in Aircraft Service
RAE Paper Library Translation 1934, 1977
8 1 FightIntenatinalSept1198
[26 1 M. Stone
Airworthiness Phislceophy Developed From Full-Scale Testing
Biannual Meeting of tke ICAF
London, UK, 1973
Douglas Paper 6099
[32 1 V. Ladda
Drag Loads at Nose Landing Gear During High-Speed Towing Operations
in: Review of Investigations on Aeronautical Fatigue in the Federal Republic of Germany
Pusblished by 0. Buxbaum and H. Huth, LDF, Darmstadt
ICAF Conference 1985 - LBF Report S-173
Wl ... ....... .
Landing Gear Improveent.-
for
Transport Aircraft
by
J. Greer MClain
Senior Mechanical Engineer
Wright Research and Development Center
Wright-Patterson AFB, (Chio 454 33 USA
and
B. M. Crenshaw
Staff Engineer
Lockheed Aeronautical Systems Company
86 S. Cobb Drive, Marietta, Georgia 30063 USA
SUNARY
This paper discusses development, testing, and analysis of retrofit nose and main landing gears designed
for improvement in C-130 transport aircraft rough field capabilities, concentrating primarily upon
analytical model prediction aspects which might be applicable to other gear designs. Two levels of
improvement were examined, with the second level resulting in a new longer stroke strut designed to retract
into existing stowage volume. All improved gears were qualified by laboratory drop tests and also
evaluated for rough field performance by laboratory shaker testing. Extensive analytical model studies
were compared with test results. Model improvements were developed where necessary, with particular
attention to representing strut friction, predicting strut rebound damping, and modeling transition from
isothermal toward adiabatic conditions in the strut inflation gas. Rough field capability estimates for
the C-130 equipped with these gears have been made. Recommendations for further analytical model
imp ovements are included.
1. INTRODUCTION
Concepts of future warfare envision combat operations from and air delivery of cargo to austere airfields,
repaired bomb damaged runways, or in some cases onto hastily prepared unpaved surfaces, all expected to be
quite rough and very short. If additional runway roughness tolerance can be achieved through landing gear
improvements, bomb damage repair times could be reduced, or alternatively taxiways or grass strips might be
utilized, thus greatly diminishing the potential effectiveness of an airfield attack. One tactical
transport, the C-130, could potentially reap enormous benefits in this scenario through improvements in
landing gear design.
One way to accommodate short runways is to reduce forward speed and the kinetic energy to be dissipated
during rollout by using higher approach glide slopes. Higher glide slopes generally mean higher landing
impact sink speeds. Design sink rate for Short Takeoff and Landing (STOL) aircraft is usually assumed to
be 15 FPS (Reference 1.) Rates higher than 15 FPS would likely provide diminishing benefits because of
higher landing gear weight to achieve necessary gear stroke. Current land based aircraft are designed for
10 FPS sink rate, or on some older models, 9 FPS.
The C-130 transport is a 9 FPS aircraft and considerable attention hat been directed to methods for
improving its sink rate and ground roughness capability. The desirability of landing gear modifications
became increasingly apparent during the joint US/UK HAVE BOUNCE program (Reference 2.) Frequent bottoming
of the nose gear and the lack of ability to maintain stiut positions set during service first directed
attention to the strut behavior. Fortunately, the availability of good instrumentation made it possible to
compare operational pressures with theoretical predictions. Under high compressive loads, significant
pressure loss frequently occurred. Occasionally under decreasing loads, as during takeoff, pressure
recovered.
Subsequently, the instrLnented nose and main (one strut) gears were installed by thf UK on another C-130
for unpaved surface testing (Reference 3.) During short flights, dramatic pressure variations were found
in the main gear from liftoff to the next touchdown. Figure 1 shows results from a series of three takeoff
and landing runs from a paved surface where extended strut pressure recovery varied between 140 psi and 60
psi wile in the air.
Upon completion of aircraft testing, the instrumented gears were shipped t', the U. S. Air Force Wright
Aeronautical Laboratories for further study. Tests of the nose gear showed that duplication of aircraft
measured dynamic strut position resulted in pressure variations similar to the previously collected
aircraft data.
There have been sporadic reports of gear servicing problems (i. e. flat strut3 although following
recommended procedures.) It is believed that these incidents are a manifestation of the same pressure
loss/recovery phenomenon observed during the test programs.
Figure 2 illustrates the dilema faced in servicing struts with variable pressure behavior. If servicing
is done with the aircraft on jacks, any subsequent gas loss under load will lower the alrspring curve and
may even. allow the strut to bottom, forcing the aircraft to taxi with only the tire spring. Of course
bottoming is undesirable as there is no damping by the strut. If servicing is done under load with soe
gas already in solution, over charging could result. Pressure recovery could cause excess gas pressure
adding proportionately to the orifice load at landing impact, and increasing peak load at all rates of
sink.
I2-I
If the orifice peak load happens to set the design limit load at design rate of sink, overload could occur
should this sink rate be reached. Even though the landing gears might be strong enough to withstand the
dded loads, restrictions on allowable wing fuel might be required to offset the additional aircraft load.
Shortly aft-r the HAVEBOUNCE program, the Lockheed preliminary design organization began a program to
investigate ways to improve C-130 gear performance. The study covered a number of concepts and two of
these resulted in experimental hardware.
To increase surface roughness and sink rate capability, both improved and STOL landing gears have been
built and tested. This report focuses primarily upon the surface roughness capabilities wiLn these gears
as predicted by analysis. The extended stroke gears are now being flown on Lockheed's High Technology Test
Bed aircraft which is shown in Figure 3; however, testing at high sink rates and over test runway profiles
has not yet been accomplished.
A new upper bearing assembly contains orifices and a flap valve to control extension rate by restricting
flow from the chamber between the piston and outer cylinder. The rebound orifices were sized
experimentally to achieve the recommended maximum extension rate of approximately one third the landing
impact closure rate (Reference 4.) Originally, a flap valve was installed on the main orifice, but tests
showed cavitation in the main oil chamber, consequently this flap valve ws removed. A replacment lower
bearing with a chamfered lower edge is used to minimize friction during wheel spin-up.
These modifications are designed for retrofit. Maximum use is made of existing landing gear parts
including all existing forgings. The changeover can be accomplished on later series one piece forging
struts by mechanical disassembly and parts replacement. On earlier 2-piece forging/welded struts,
machining of the orifice plate is necessary to allow clearance around the new inner cylinder.
Figure 5 compares existing and improved main gear struts. Again, the goal was to provide a kit that could
be installed by parts substitution without machine shop re-work of existing gears. As shown, the existing
oil chamber bulkhead and the orifice plate are removed and replaced with a bulkhead supporting a metering
pin having an oil spray deflector counter piston mounted on its top end. The counter piston minimizes oil
spray during strut rebound. A new upper bearing assembly has orifices and a control flap valve. By
shortening the bearing length, stroke is increased from 10.5 in. to 11 in.
One each nose and main gear modification kits were constructed for laboratory testing and installed in the
instrumented nose and main struts previously used in the HAVE BOUNCE program. No installations have yet
been made on an aircraft.
b. Extended Stroke Landing Gears
During the time improved gears were being developed, a test aircraft was obtained for evaluating
aerodynamic devices, flight controls, and electronic equipment related to STOL operation. New landing
gears developed for this test aircraft, in addition to providing longer stroke, incorporated features built
into the improved gears. STOL operation gears are referred to as extended stroke gears.
Figure 6 shows sketches of extended stroke gears. One each nose and main gear struts for laboratory
testing and one aircraft ship set for flight testing were built by M-asco to T-21heed drawings. These
struts are similar to the improved nose and main gears. They incorporate gas/oil separation, rebound
daping, and for the main gear, a metering pin. In addition, the nose gear stroke is increased from 10.5
in. to 18 in. and the main gear stroke is increased from 10.5 in. to 24 in.
Since these struts will not fit into the existing wheel well when fully extended, a compression feature has
been built into both nose and main gears, wherein ship's hydraulic pressire is used to compress the struts
prior to retraction. Insufficient door clearances on the test aircraft required the use of smaller tires
on a temporary basis. Should these gears be installed on production aircraft, modifications to the
compression system will be made to accommodate current tire sizes.
Figure 7 shows a comparison of the standard and extended stroke main landing gears. The additional energy
absorption potential of the longer stroke gear is readily apparent in this figure.
Both types of new gears may be retrofitted to existing aircraft or incorporated into new production
C-130's.
2-3
Guidance for development of these landing gear modifications and replacements have been and continue to be
dependent upon extensive testing.
a. Aircraft Testing
There have been three recent opportunities to collect flight test results to evaluate C-130 landing gears.
These were the programs described in References 2 and 3, and current testing on Lockheed's High Technology
Test Bed aircraft. Prior to the HAVE BOUNCE test program, most C-130 testing utilized either
oscillographic recording or FM analog techniques that were not readily usable to the level of detail
required to determine gear performance. Main ccnc--ns during earlier test- were the magnitudes of peak
loads and only a limited amount of strut performance data was available for analysis in time history or
cross plot format. Internal pressures were not usually measured, and string potentiometers used for stroke
measurements were not very accurate.
Figure 8 shows an example of the high quality time history data collected in the HAVEBOUNCE program. For
the first time, it became practical via computer to cross plot measured pressures vs. stroke and to
superimpose theoretical pressure/strokes with test results as shown in Figure 9. Having this cap ... y,
particular attention could be directed toward gear response as was the case during the grass st,rface trials
program.
Flight tests have also emphasized the significance of main gear telescoping friction. Figure 10 shows main
gear strut friction during a landing rollout. It is seen that strut motion occurs in a serie cf steps as
load changes, indicating only limited shock absorption. All four main struts on the High Technology Test
Bed aircraft have been instrumented. The data from four main struts has also revealed interesting main
gear friction behavior as shown in Figure 11. The differences in the amount of telescoping friction
between the left and right landing gears is believed to illustrate a lateral drift (left) landing where
friction is increased due to outboard loading of the right pair of struts and decreased due to inboard
loading of the left pair of struts.
Only limited landing gear data has been obtained from the test aircraft. STOL evaluation testing is now
underway.
b. Laboratory Testing
Both the improved and extended stroke landing gears have gone through extensive laboratory testing as
individual struts. All teats were performed by the Air Force Wright Aeronautical Laboratories at
Wright-Patterson Air Force Base, Ohio, using Numbers 3 and 4 drop towers, as sketched in Figure 12.
For drop testing, the gears were suspended beneath a ballast container and allowed to free fall onto an
instrumented platform. Backward pre-rotation of the tires simulated spin-up. At the position where the
tirp juLt touched the platform, further downward motion of the ballast container was resisted by air
cylinders at a pressure calculated to equal the ballast weight, thus simulating aircraft lift.
Instrumentation consisted of strain gage measurement of vertical load, main hydraulic chamber oil pressure,
gas inflation pressure, rebound chamber pressure, and in some cases inflation gas temperature. Vertical
displacement and acceleration of the ballast container were recorded. For shaker testing, hydraulic shaker
ram pressure and displacement were added. For drop tests, drop platform vertical and drag loads were
included.
Figure 13 illustrates a fitting used to gain access for rebound chamber pressure measurement directly
through the strut sidewall. This technique was for laboratory use only and was not considered for any gear
installed on an aircraft.
Since no increase in sink rate or aircraft landing weights were planned for the improved gears, they were
drop tested to the same conditions as originally specified for the standard struts. Any gains for landing
impact are thus realized by lower ianding loads.
References 5 and 6 report drop test results for the improved landing gears and References 7 and 8 report
drop test results for the extended stroke gears. For both improved and extended stroke gear sets,
preliminary drops were made for sizing rebound damper orifices and main gear metering pins. The extended
stroke gears, being essentially a new design were drop tested to the specifications of MIL-T-6053. They
are fully qualified for flight.
Figure 14 compares drop tests for 3, 5, 6, 9, and 10 FP'S sink rates of the improved nose gear with Le drop
tests of the standard C-130 nose gear. These comparisons show that initial load build up is more rapid in
the improved gear as a result of deletion of the inverted cone orifice, that significantly less stroke is
used at each sink speed, that no bottoming occurs, and that rebound is slower. Lower loads and higher
strut efficiencies could be achieved with a larger main orifice; however, it was felt that the increased
damping of the orifice Size chosen would be beneficial in reducing aircraft porpoising over runway
roughness.
For shaker testing, the instrumented drop platform was replaced by a platform attached to a large hydraulic
shaker capable of producing programmed vertical displacements equivalent to the excitation seen by the tire
when passing over an actual obstacle at various speeds. All operations except rolling tire effects and
tire wraparound of obstacle sharp corners are accounted for. Of course, since individual nose and main
struts were tested separately, aircraft pitching, pilot response, engine thrust and aerodynamic effects are
not included.
12-4
During specific rough surface simulation the ballast container was unrestrained in vertical motion. In
order to study response to a particular motion such as step closure or extension, and to run slow rate
load/stroke curves, the ballast container was restrained to prevent vertical movement.
Figure 15 il.!;strates typical shaker tests of the extended stroke main gear. Gear vertical load is shown
for 50, 75, 100, and 125 ft./see. simulated aircraft speeds over a 7.2 in. (1-cosine) shaped dip, also
shown in Figure 15. Because of ballast container and support structure weight, these tests were run at
50,000 lb. strut load, corresponding to an aircraft weight of over 200,000 lb. substantially greater than
current aircraft weight limits. This capability illustrates the robust design of these struts.
Prior to the development of powerful desktop computers in the early to mid 1980's, the cost of caputer
time limited to a large degree the number and complexity of simulation studies that could be carried out.
Computational cost for non-linear landing gear studies can no longer be considered to be a limiting factor
within even very modest size organizations, as the newer 32-bit cpu micro computers have the capacity and
speed to run extensive landing gear simulations in combination with a flexible aircraft model. Prospects
are good for achieving even greater coputational capability over the next several years.
While not as spectacular as computer hardware improvements, software techniques have also progressed.
Reference 9 presents a variable step size integration technique that is several times faster than the
popular Runge-Kutta fixed step size methods. Data presentation methods have also evolved, as cheap and
easy to use graphics devices have proliferated.
In order to more accurately simulate performance of improved and extended stroke gears, several changes to
the usual C-130 modeling techniques were made. These included polytropic gas compression, revised strut
friction, and representation of rebound damping. These changes have resulted in an acceptable level of
correlation. Simulation and test results comparisons, based on individual struts, increases confidence in
the predictions of loads in the whole aircraft simulation.
The modeling techniques developed for these gears are expected to be beneficial for other landing gears
since they resulted in increased understanding of several previously troubling areas in gear simulation.
A first task in validating a landing gear simulation model is to establish a correct pressure vs. stroke
relationship, as this determines the gear spring characteristics and primarily the response frequency.
Accurately measured pressures and strokes are vital to establish this relation. Once the correct
pressure/stroke is defined, oLner model features such as strut friction can be more easily established.
Usually, analyses are made on the basis of an assumed constant polytropic gas compression/expansion
coefrneient (Reference 1), perhaps 1. *.r struts without gas/oil separation and 1.3 - 1.4 for struts with
gas/oil separation. In the case of taxi with the strut starting from isothermal equilibrium, there have
been few ways to transition to the higher coefficients without encountering a fictitious pressure step.
Reference 10 indicated that transition might be accounted for during compression by making the
compression/expansion exponent an exponential function of stroking velocity. This approach was tried in an
experimental simulation and found to work satisfactorily during compression; however, at maximum stroke
when strut velocity goes to zero the exponent drops back to the isothermal value of 1.0 resulting in a
sudden drop in computed pressure.
Reference 11 points out that the polytropic exponent is a variable and is dependent upon the ratio of heat
transfer to work done. It was decided to experimentally model the compression/expansion coefficient as a
function of the work done on the gas. This method was found to work quite satisfactorily and accounts for
both heating and cooling of the strut. Figure 16 shows the exponential functions which were fitted for the
nose and main landing gears. Additionally provision is made for an initial exponent which may differ from
1.0 due to prior heating or cooling of the strut.
Also, tests at slow rates of loading produced pressures less than those calculated by a simple isothermal
process, even when the gas and oil were separated. Approximation of the effects of material expansion and
hydraulic oil compressibility as a function of pressure were required. For the extended stroke gears these
3 3
quantities were determined to be .001587 in. /pi for the nose gear and .0070588 in. /pi for the main
gear. With this modification the pressure becomes an implicit function which must be solved by iteration
at eachi time step.
Figure 17 illustrates that the variable thermodynamic coefficient and the campression/expansion modeling
functions result in a close match of the pressure vs. stroke relations for both gears. There is a
substantial improvement over methods which do not allow for these effects.
The level of strut friction in a gear can usually be readily seen from test results in a cross plot of
strut load and stroke. Changes in load while the strut position stays fixed is an indication of friction.
Figure 18 shows examples of slow and rapid rates of loading for the high sink rate main gear. The
stairstep characteristic at the slow rate is indicative Of a stick/slip situation. As stroking velocity
increases, the load/stroke curve becomes smooth; however, when the strut reverses direction a large change
in load occurs before the strut moves. Since damping is zero and air pressure remains constant while the
strut is stationary, friction is the remaining unknown force.
12-5
A calculation of normal force in the strut shows that usually assumed friction coefficients for teflon on
steel are much too low to generate the levels of friction observed. It is theorized that bending of the
strut causes the piston/cylinder/bearing combination to became oval in shape as illustrated in Figure 19.
Binding between the bearing and cylinder at the upper bearing and between the bearing and piston at the
lower bearing are believed to cause the large friction forces observed.
Figures 20 and 21 illustrate the importance of representing strut friction in the simulation model. Test
results illustrate the standard C-130 main landing gear loaded in the laboratory by a simulated 5 in.
amplitude (1-cosine) bump at a simulated speed of 50 ft./sec. Since there were significant differences in
measured pressure/stroke vs. theoretical values, an isothermal curve fit was made to minimize errors from
the airspring. Figure 20 shows results of the simulation and test comparisons with the fitted airapring
plus friction, with good correlation indicated. Figure 21 shows similar results with the friction model
removed, and comparisons are poor. It is assumed that discrepancies resulting from poor friction modeling
have not previously been so apparent because airspring errors have partially compensated for friction
errors.
Pressure -cacurements from shaker tests were used to determine effective flow coefficients from the rebound
chambers. Figure 22 shows rebound chamber and main oil chamber pressures for the extended stroke main
gear. It is seen that each time the gear extends the flap valve closes, producing a pressure drop across
the rebound orifices in the upper bearing, limiting the extension rate of the strut. Two E-bvmp profiles
were used for the gear simulation.
As indicated in Figure 2', the extended stroke nose gear flap valve does not close promptly as the strut
starts to extend. The scme 2 E-bump profile is used so the difference in behavior of the main and nose
struts in attributable to mechanical causes. Figure 24 shows a plot of the mathematically calculated
derivative of strut displacement along with the rebound chamber pressure. Flap valve closure occurs in the
neighborhood of 30 in./sc. extension velocity. For simulation purposes, it was necessary to set flap
valve closure criteria as a function of both rebound chamber pressure and strut extension rate. The values
selected were; rebound chamber pressure dropping below 600 psi and strut extension rate exceeding 27
in./sec. Using this criteria, fairly good rebound chamber pressure calculations were obtained as
illustrated in Figure 25. For a range of other speeds, simulated flap valve closure was reasonably
consistent with test results.
Successful correlation for the extended stroke nose gear would have been unlikely without benefit of the
rebound chamber pressure test measurements, as flap valve sticking is not directly detectable from load and
stroke measurements.
With the previously discussed modifications to the polytropic gas exponent, derived rebound chamber orifice
coefficients, flap valve closure criteria, and a modified friction model for the main gear, overall model
accuracy can be evaluated by comparing strut load with test results. Figures 26 and 27 show comparisons
for the extended stroke nose and main gears. Overall agreement between simulation and test is seen to be
good. These models are now considered suitable for analytical studies of aircraft capability.
Since no analytical model is yet available to adequately represent gas solubility and subsequent
dissolution, it is necessary to resort to approximations to generate adequate comparisons between analysis
and test results. For a particular strut, the best approximation can be generated by plotting test
generated pressure vs. stroke and then fitting an analytical curve to the resulting data. For portions of
the time history where the fitted curve is close to actual values, overall agreement between test and
simulation is usually good. For the other portions of the time history, results are usually poor.
For the HAVEBOUNCEprogram, fitted data from a number of tests were averaged to produce a composite curve
fit. Comparisons with test data usirg the averaged fit data were less accurate than the comparisons using
fits for individual runs; however, the averaged values were considered more suitable for parametric
studies, and have been used on an interim basis until a better model is developed.
A good analytical model of gas solubility would be useful for a number of current gear designs, and also
for future gears in which gas/fluid separation might not be feasible.
An aircraft's response to runway roughness is a non-linear resonant system having highest loads at certain
excitation frequencies. From an operational standpoint, this translates into a question of how fast the
aircraft is going over a particular rough spot on the runway, and does that speed represent a resonant
situation.
It is not possible to pre-define the distribution of roughness wavelengths along a runway. Although many
attempts have been made to develop generalized prediction methods, none have proven to be totally
acceptable. Reference 12 is probably the latest examination of the roughness problem. It is concerned
with loading from runway bomb damage repairs and was directed primarily toward occurrences of multiple
bumps.
In addition to considering candidate levels of roughness for landing gear design, Reference 12 went into
considerable detail about methods of analysis and presentation of results. One approach was to select
12-6
worst case combinations of speed, wavelength, and bump amplitude, and use this recognizing the resulting
severe restrictions on allowable runway bump amplitudes. Another approach was to recognize that some
additional capability could be obtained if certain speed and roughness wavelength combinations could be
avoided. Here the term wavelength could be used with reference to a single bump or to the periodic
excitation for multiple bumps.
To illustrate critical combinations of speed and wavelength from this second approach, a data presentation
method, called the load contouring approach, was developed in Reference 12. To generate gear load
information, a matrix of simulations are made with speed and wavelength varying in a prescribed manner. By
using interpolation, contours of load amplitude can be extracted fran a matrix of peak loads from these
computer simulations. For most purposes, limit load would be used. Unless some part of the aircraft is
known to always be the first to reach its limit, loads for a number of locations must be monitored. These
load contours can be generated for different conditions, i.e. taxi, takeoff, rollout, etc. and for
specified roughness.
The loads contouring methods from Reference 12 are general and can also be applied to roughness
specifications other than bomb damage repairs. In this report they are used to study roughness capability
with respect to the specification roughness of MIL-A-8862 as shown in Figure 28. These curves were derived
from measurements of potential runway sites and are thus at least semi-realistic. Three levels of
roughness classifications are contained in this specification.
For evaluating this aircraft, the semi-prepared surface was chosen since there is little question of the
C-130's ability to operate from paved surfaces, and the unprepared surface roughness is so severe few
aircraft could operate at that level, except at very restricted taxi speeds.
Figure 29 shows estimated capability with respect to a semi-prepared surface for a C-130 at design landing
weight for encountering a single dip having the amplitude and wavelengths relationships specified. It is
assumed that the aircraft is operating during rollout with full reverse thrust, with full up elevator to
minimize nose gear loads, and with no wheel braking. Figure 29(a) shows estimated capability of the
aircraft as equipped with a standard landing gear. Figure 29(b) shows estimated capability of the aircraft
equipped with the improved nose landing gear along with standard mains. Since the improved main gears do
not have full gas/oil separation and only a minor increase in stroke, aircraft capability would not be
significantly changed with their inclusion. Figure 29(c) shows estimated capability of the aircraft when
equipped with the extended stroke nose and main landing gears.
These results show an unrestricted capability of approximately 70 percent of the semi-prepared roughness
level with the standard gears, approximately 80 percent capability with the improved nose gear, and
approximately 100 percent capability with the extended stroke gears. In all cases nose gear vertical load
limits established these capabilities.
6. LESSONS LEARNED
Experience with several configurations of gears for a single type of aircraft, along with extensive
laboratory and flight testing, has provided guidance for better application of the load regulating
techniques incorporated into these gears as well as providing data for improved landing gear modeling.
Sone of the lessons learned were:
o There should be generous allowance for flap valve clearance in order to prevent erratic damping
behavior and provide for prompt valve closure at the start of gear extension.
o Although there may be cases where separation of hydraulic oil and inflation gas may not be practical
or desirable, laboratory results from these gears would indicate that this is a useful means of
achieving better control over strut internal pressures and aids servicing reliability. Further
flight measurements should reveal any operational problems that might not be seen from laboratory
tests.
" Although simple floating separator pistons should not have significant pressure drops, there is a
possibility for leakage. Evidence of gas or fluid transfer past separators in these gears was not
seen in laboratory tests of C-130 gears; however, conversations with individuals experienced with
other gears indicate that chamber purging capability needs to be provided, as well as attention to
good sealing.
" Possible binding of internal floating pistons fran cylinder deformation under load should be
considered.
O The effects of strut deformation in contributing to high friction are believed to be significant.
Possible contributions of high binding loads to strut fatigue damage should be considered.
o For struts which must be partially compressed for retraction, adequate hydraulic supply pressure and
pressure area should be provided to ensure closure during specified gear retraction time and to
ensure that sufficient pressure is available to overcome adiabatic compression pressure inside the
strut.
b. Landing Gear Testing and Computer odeling
mcng the lessons learned from testing and simulation correlation efforts are:
12-7
Although much progress has been made over the past several years in improving transport landing gear design
and analysis, there is opportunity for further improvement. The major items which have become apparent
during the development and testing of C-130 transport landing gears are:
a. Strut Thermodymmics
Despite the success achieved in modeling isothermal to adiabatic transition, the technique remains
empirical in that it must be fitted to each strut. A more satisfactory and accurate model should result
from a thorough investigation of heat transfer between the inflation gas and strut walls under dynamic
conditions. Initial efforts have been made as illustrated in Figure 30; however, the accuracy level
obtained in simulation does not yet equal that of the empirical method. With additional work a good
thermodynamic model can probably be developed.
b. Structural Deformation and Strut Binding
The significance of strut binding upon the response of struts has been illustrated in this report.
Reference 13 indicates that high friction may be serious enough to induce strut cracking. It is therefore
advantageous to understand the strut binding problen. as thoroughly as possible.
Research into strut deformation could include analytical (perhaps finite element) modeling of deformed
struts to calculate localized pressure loaoing at the bearing/cylinder Interface. Design efforts should be
directed toward minimizing deformations consistent with reascnable weight and manufacturing costs.
Statistics on the severity of friction induced cracking would be useful. If some struts are found to be
more immune to cracking than others, then the reasons for the differences should be investigated.
There is no general method yet available to predict the degree and timing of inflation gas solubility and
the occasional sudden release from solution. Solubility might be modeled on the basis of work (and rate)
input to the gas; but detailed internal gear design may result in variations from gear to gear. Careful
laboratory experimenting will be required to develop a generally applicable model. Perhaps experiments
could be carried out with a simplified strut made from hydraulic cylinders where orifices, fluid to gas
ratios and inflation pressures could be easily varied. Results might be more useful than those aimed
toward a particular landing gear.
Development of improved retrofit landing gears for the C-130 has demonstrated the practicality of modifying
existing hardware to improve rough airfield performance. The unique opportunity to evaluate both minor
internal modifications as well as substantial stroke length changes provides valuable guidance for making
tradeoffs in cost vs. level of improvement. The roughness tolerance increase for each strut modification
has been quantified for comparison and technique demonstration purposes, utilizing one selected surface
roughness criteria.
Of equal importance, especially for future landing gear design, was the experience gained from the several
cycles of testing, analysis, and performance comparisons of these gears. Efforts to refine the analytical
models for better representation of gear response has produced lower cost techniques with higher accuracy
than previously available. Significant improvements included dynamic variations in polytropic gas
exponent, a fairly realistic representation of strut binding, and rebound chamber hydraulic damping with
allowances for flap valve sticking. Analytical models of the C-130 gears are considered to be adequately
validated for parametric studies over other roughness criteria.
Although these studies have produced quite satisfactory results for the C-130, it is recognized that more
effort is needed to achieve a sound theoretical basis for application to new gears. It is recommended that
work continue in the areas of gas exponent modeling, thermodynamic equations, binding friction from
structural deformation, and rebound chamber fluid flow characteristics with a goal of calculating these
quantities from strut geametry and volume without relying exclusively on experimental data to provide the
necessary relationships. This work should encompass continued analysis of existing data for these landing
gears as well as plans for future general purpose landing gear testing.
9. REFERENCES
1. Currie, Norman S., "Aircraft Landing Gear Design: Principles and Practices," AIAA Education Series,
American Institute of Aeronautics and Astronautics, Inc., Washington, D. C., 1988.
2. Crenshaw, B. M., and Owen, M. M., "C-130 Response to Bomb Damage Repaired Runways," Volumes 1 and 2,
ESL-TR-82-23, Engineering and Services Laboratory, Air Force Engineering Services Center, Tyndall
Air Force Base, Florida 32403, December 1983.
3. Crenshaw, B. M., "C-130 MK3 Hercules Grass Trials," LG82ER0082, Published Under University of Dayton
P. 0. RI-17781, August, 1981.
4. Conway, H. G., "Landing Gear Design," Chapman and Hall Ltd., London, 1958.
5. Crenshaw, B. M., "Summary of Standard and Improved C-130 Nose Landing Gear Performance Testing,"
AFWAL-TR-3093, Air Force Wright Aeronautical Laboratories, Wright-Patterson Air Force Base, Ohio,
March, 1988.
6. Crenshaw, B. M. "Suammaryof Standard and Improved C-130 Main Landing Gear Performance Testing,"
AWAL-TR-87-3094, Air Force Wright Aeronautical Laboratories, Wright-Patterson Air Force Base, Ohio,
March 1988.
7. Vennel, W. W., "C-130 Extended Stroke Nose Landing Ger Drop Test Results, "LG86ER0085, Lockheed
Georgia Company, Marietta, Georgia, July 1986.
8. Vennel, W. W., "C-130 Extended Stroke Main Landing Gear Drop Test Results, "LG86ER0O86, Lockheed
Georgia Company, Marietta, Georgia, July 1986.
9. Sampine, Lawrenie F., and Gordon, Marylin K., "Computer Solution of Ordinary Differential
Equations," W. H. Freeman and Company, 1975.
10. Anon, "Service Experience of Shock Absorbers with Gas Separator Pistons," Dowty Presentation to
SAE-A5.
11. Wahi, M. K., -Oil Compressibility and Polytropic Air Compression Analysis for Oleopneumatic Shock
Struts,", Journal of Aircraft, Vol 13, No. 7, July 1976.
12. Anon," Aircraft Operation on Repaired Runways," Report of Working (;.oup 22 of the Structures and
Materials Panel, NATO/AGARD, Paris France, (In Publication).
13. Fewtrell, H. E., "Consideration of Mechanical, Physical, and Chemical Properties in Bearing Selection
for Landing Gears of Large Transport Aircraft," Journal of the American Society of Lubrication
Engineers, March 1983 pp 153-157.
12-9
Pressure RecO.vry
Isothermal Airload
Leve_1)
iv
ioStatic Curve d l :::
after..Overcharge:::::::::
-
a Pressure
Recovery
4 Isothernal Air'oad
otential Stroke
pj
Lossfor rax....
STROKE STROKE_____
(a) Servicing on Jacks (b) Servicing Under Load
I ORIFICE
III
OIL (DELETED) FLAP VALVE - IrAO
REBOUNDDAMPERSEATO
(ADDADDED)
BULK(HEAD
(DELETED) PJRCYLINDER
PISTON
(SAME)
STANDPIPE
ORIFICESUPPORTTUBE (OIID
GA P (SAME) CONTR-ISO
PEIBARN (ADDED)
-( (DELETED) ~1
ORIFICEPLATE FLPVLE EON
(DELETED) UPPER ADDED)VREON
(II
OIL BLGADBEARING OIL ___METERING PIN
(DELETED) J NEA)
-OUTER CYLINDER
I (SAME)
PISTON (SM4E)
UPPER BEARING
V41THREB3OUND
MAIN DAMPING DAMPERFLAP VALVE
ORIFICE
MAINDAM4PING UPPERBEARING
ORIFICE NITHREBOUND METERINGPIN
DAMPERFLAP VALVE
OIIL
GASIOIL
SEPARATORPISTON STRUT CONTRACT1ON
L.OWER LOWER
BSEARNG
MSRG STRUTCONTRACniON GAS
PISTON
GAS SIL
GM PISTON
SEPARATOR
WIFILTIONGAS
FORT 11*LATlO
(a) None Strut POR (b) Main Strut
ULjmte 6. schematics Of C-130 Extended Stroke Nose and Main
Figure 7. Comparison oi1 En, Standard and Extended SrroXn MAIn
Landing Gear
12-
20
f 4--
E- Note Pressure
02 Recovery After Liftoff-
0 25 5.0 7.5 10.0 12.5 15.0 17.5 20.0 0 2.5 5.0 7 .5 10.0125i. i., 0.
TIME - SECONDS TIME - SECONDS
35si
30 Pressure Decrease
30 Isothernu Attributed to __
o Compressi tO4 -Cooig
15
(15
0 1
1) 10 20 30 40 50 60
o 2 4 6 a 10
STRUT STROKE I N. TIME - SECONDS
FradStrut oeoo.E )
fie I j.-CYUNOER
0 3 6 9 12 15 18 21Cu.S
TIME SECONDS
Figure 11. Exampileof Strut Friction Figure 13. Example of a Rebound
during a Lateral Drift Landi ng. Chamber Pressure Tap
Fitti ng.
Hyraulic kctuated
Shaker Table
Piqure 12. Drop Test end Shaker Test setups at WPAFD Ohio.
10 tP
SrtInitial Condition
lI~D t. 'Se.
,Loading)
75.Ft./Sen.
50~ Ft./Sen. Main Strut
Il( ft./SecI
24 - __ 24 -
21 - Tst21
T
016 - -
155
IS
9 simulat ion
P4 Test
0 .4 .8 1.2 1.6 2.0 2.4 2.8 3.2 0 .4 .8 1.2 1.6 2.0 2.4 2.8 1,2
TIME - SECONDS TIME - SECONDS
(a) Extended Stroke Nose Gear (h) Extended Stroke Main Dear
10 in C
pesio 2 E-Bump Simulation
100Mm Iopeso I8
10 Min Extension ____
Load Increment
o60- 070 - - Due to Strut
N Load Increment to
60 Static Friction 60Overcome B0din
S40.- Ss-
50 _
-. 4 -
LOADING
030 00
% 20 -
U,_ AS UNLOA35
C
20- - -O O,
a:.
00 1_ -m 10---
0 2 4 6 5 10 12 0 2 4 6 8 20 12
STRUT STROKE -IN. STRUT STROKE - IN.
(a) Slow Rate of Loading (b) Rapid Rate of Loading
Figure 18. Iliust ration of Main Gear Strut Friction during
Various Rates of Loading.
a m- - 90 90 Simulation
q9 ---- Teat
80 -
__
-
- -
- Simulation
-- Teat
05--
w7 -- - A 1- -
140 40-: -
-g30
- - 10- A- _
510 -- - -
0 .4 .8 1.2 1.6 2.0 2.-: -.8 3.2 0 .4 .8 1.2 1.6 2.0 2.4 2.8 3.2
TIME - SECONDS
Figure 20. Strut Load Comparisons for a Figure 21. Strut Load Comparisons for a
Standard Main Gear with Strut Standard Main Dear with Strut
rictiton ModelI Incorporated. Friction ModelIRemoved.
12-15
36 Extended Stroke Main Gear Tests Extended Stroke Nose Gear Tests
Rebound Valve Closed I Rebound Valve Closed
Pressure Rebound
Z28 28 Pressure
24 I24.
20 020.
12 12
4. 1 1 .
.4 .8 1 2 1.6 2.0 2.4 2.8 3.2 0 .4 .8 1.2 I.6 2.0 2.4 2.8 3 .2
TIME - SECONDS TIME - SECONDS
Figure 22. Rebound Chamber Damping Figure 23. Rebound Chamber Damping
without Valve Binding. with Valve Binding.
10 -r-----
- -40 - - _
Cal Rebound Valve Clos.ed Simulation
S0 . - ..--- Test
z 0 2.-5- -
0
50 :_ x30r-
10 2
l- .I L; 25-i -
W 25
;. 20' . Al1
0.8 1.2 1.6 2.0 2.4 2.8 3.2 3.6 4.0 .4 .8 1.2 1.6 2.0 2.4 2.8 3.2
TIME - SECONDS TIME - SECONDS
Figure 24. Correlation Between Strut Extension Figure 25. Comparisons of Measured Rebound
Velocity and Flap Valve Seating. Chamber Pressures with Pressures
Calculated in Simulation Model.
21 --
oil
f-~ 0-
12
10
~
"IME
EODST
. .: ~ ~ - 5
- SECND
1 2'
Figure 26. OndividuaI Noe Strut Load Figure 27. Individual Maim Strut Load
Comparisons etwee.n S imulation Comparisons Between imu latio0
and Test Results, and Test Results.
12-16
r
-. Amplitude vs. Wavelength
Pavd SurfaceAmltdvsWaengh
10 20 30 40 50 60 70 80 90 100 Semi-Prepared
WAVELENGTH - FT. Roughness Levels
100 -
-----
Amplitude "%
5- Amplitude 80% Roughness-
-Amplitude
S40 -. Roughness- I I
Roughness
n 30 haded Area
Restricted Operations
20 Clear Area [
(a) Standard is Unrestricted
Landing Gear 10- - - " -
20 30 40 50 60 70 80 90 -15 -i0 -5 0
90-- o/I // I.
I"80%
N60-
Amplitude Amplitude -
0 Roughness Roughness
0--
90 ---- A11
80- -
70-s-- 4
Amplitude Amplitude
. 50-- - - Roughness-- - Roughness
0
zShaded Area Operations
0= 30- R-estricted 10
0M20- Clear Ae
(c) Extended Stroke i nrestricted
Landing Gears 10 - -
20 30 40 50 60 70 80 90 -15 -10 -5 0
VELOCITY AT ROUGHNESS ENCOUNTER - KNOTS ROUGHNESS
AMPLITUDE - IN.
Figure 29. Speed/WavOlength Limitations Over MIL-A-8862 Roughness
for Standard, Improved Nose/Standard Main, and High
Sink Rat: Landing Gea r.
- > C-hgaa -- - CLahgale,.12,, 1l78, S_, S,
I'S
78 [2(T-J S, S
Z'oe 8 070
lb Slw on
Io50,08 6.
F 2,oCS0(oi(J [~~2 a
I1 S-n,,
A,
Ar
Thermodynamic Analysis.
13-1
ABSTRACT
During the design phase, at pre-project step, there is an increasing need for a
comprehensive, integrated numerical simulation tool, capable of providing the design
engineer with the complete answer to his analysis problems: kinematic, thermal,
mechanical, linear or non-linear (static, dynamic, fatigue).
Numerical s-mulation of dynamic behaviour is a priority area for landing gear, as
there are many problems to be solved (landing impact, taxiing shimmy, retraction
and extension, active control, crash landing, etc...) and they are far from simple (non-
linearity, sometimes acute, due to large rotations, shock absorber friction, hydraulic
damping laws,...)
The aim of the paper is to describe the Messier-Bugatti analysis and numerical
simulation system used for solving landing gear dynamics problems, which is designed
around a finite elements software package for the analysis of mechanisms and
flexible bodies.
The following examples of applications are examined
Simulation of landing
Simulation of taxiing on rough fields or repaired runways
Simulation of extension and retraction
Simulation of catapulting
Analysis of shimmy stability.
RESUME
1 - INTRODUCTION
2 -L'OUTIL DE SIMULATION
Par ailleurs. pour les besoins specifiques du train d'atterrissage, ont kt,
developp46s des 6l6ments speciaux tels q ue le v~lrin hydraulique ainsi qu'un
616ment de liaison entre un pneumatique flexible et le sol avec les proprie-tes de
mise en rotation. de ripei et de freinage.
Messier-Bugatti a dC-velo '6 pour ses propres besoins un 616ment utilisateur
d~crivant I amortisseur o leo-pneumatique monochambre ou bichambre, prenant
en compte les lois de laminage de I'huile en Ky2 , les lois polytropiques de
compression de l'azote, les frottements de paliers et garnitures de l'amortisseur
Ia compressibilit6 de l'huile, etc... La programmation de cet 6element ayant 4ta
faite au sein de notre soci~t6, elle inclut de ce fait tout notre savoir-faire en
matiLre d'amortisseurs ; de plus, cet 616ment peut ais~ment @tre developpcs
pour prendre en compte de nouvelles technologies d'amortisseurs.
Ce logiciel offre, comme fonctionnalitk suppifmentaire, l'utilisation de super-
64lsments i modes composants g4n4r~s dans le module d'analyse dynamique
lineaire, ce qul permet de rkdu ire considdrablement le nombre de degr~s de
libertd du mod ~le tout en conservant une tris bonne precision pour le
comportement dynamique des sous-structures correspond antes.
On peut tralter des syst~mes pouvant comporter plusleurs centaines de degr~s
de libert6.
Enfin, le schdma d'int~gration temporelle est une m~thode implicite du type
NEWMARK modifl# par HILBERT, HUGHES et TAYLOR pour am4l1iorer Ia
stabllt6 de Ia proc~dure ltdrative par l'intermdaire d'un coefficient
d'amortlssement num~rique des composantes hautes fr6quences (r~f. 1). Sa
resolution est men~e A bien grace A Ia m~thode lt~rative de NEWTON-RAPHSON.
3 -APPLICATIONS AU CALCUL DE PERFORMANCES DYNAMIQUES D'UN
ATTERRISSEuR
4 - CONCLUSION
INDEX BIBLIOGRAPIIIQUE
SUPER-ELEMENT
Tube toumant
rig.
oulssante Compas
Rous ,Pneumatique
(A
CC
322
. .
.
ACCRlHAE EN OL AVION EPMRQUE
ASSTETTE TETAO
- VITESSE DECHUTEY2O - VITESSE HOIRTALE YX
..
........
............ . ............ ...............
. ...... ....
... .....
......-------U.......7....
..... ....... ..
r.......
.....
.....
0.700.. ................. ..
0.30
0.2100- ...- ...-... .......--- --- -- - - - - - ........
--.............
. ...atio
Figure. ... ..Simu .I d* p nt g r..u.t..s amortsseu de.crosse ... . . ... ....
3-11
Tube tournant
Bielle
Barre de catapultage
Tige coulissante
Essieu I
Waems 2
M iTlm .q ndn \.
CT*PJTFE MMv2BOWG
DCLfl*IM BELA M WAI4FMT A LA W
3.......... .
40...............
Figure 3d - Catapultage avion embarqu6 inclinaison de la barre par rapport Ala RHF
13-17
cdo
do.
~~o.L~bas "Csvoitoo
prui~Lp~t,
jm~ukL
Figure 4a - Simulation de relevage d'un atterrisseur vue du modele en position train bas
13-18
Figure 4b Figure 4c
2. 2 -.
2. ~ ~ ~
____ ~ ~ H
____ 28
1.48E 7 - _ _
1.88E67
4.8eE6 .-
2.88E6
3 19 8 4 8 80 5 . 8 80 6 88 7 8 85
,1a8 8 2 88
-4.8 ____
-6.as[
8.308 -TEtS...._____________________
8.298
e______
8.2401-
a.228..-__
T
1.88 2.08 3.88 4.88 5.08 6.88 7.88
-1.58E 3 ___ _ __ -
-2.58E 3 ~
-3.881E 3 ___-___ ______
-4.96E 3 ___
-4.S8E 3 ______.___
-S.58E 3 ______
T-0
0-
C-
5a - S m lton d
Figure~ ~o lg-ae ur pse s m aren n
13-22
5U~-C Siuato
Figure ~ ~ deruae
DT (7Mitsmarmntaea
L-12,6 m Repair spacing S. 10M
Z-H~l1-cos 2 ix)
2 L
z
H-50Omm
L-40 m
F yire 6a
AITEMIEI.UR FIL0IRE - SHII94- K' : OCO. 5.1.
WtW IE 1. KG- Y 22.016 N PM SEC
FXKEEXTERRE
IN FWNI(P J TM~'
"aoil 3 COM)OW1E 6
404
ITIE1IME'R AM1LI
I0ER SH - K' : 00000. S. 1.
WMWIf I 1 KG- :22.016 M4POPSEC
JfMK96T ENFOKUII JEWSM
m128m 3 C0OSMIE 6i
U
2.50 .... .. .. .. .
-. 50 .........
1.5. . . . .. ... . ... . .
Figure 6b
ArrEMISEUM AIL1ALRE - SILFWY
- K, - loom. S. 1.
KLIRI If 1. XG- V1 22.016 MPM SEC
RKE EIE Em FaIC Ill UJ ENS
"(181 3 CI4o"WT! 6
4O.w
200..
...
....
43m
0. 5O ..
0.0 . ....
-06200.
00
Figure 6c
KOI1I I 1.G - Y :27. 143 MP~ SEC.
WE EXTEURE
IN FaIW NJ rEM'S
moonI a CKSANIE 6
50w.
100.
-3m.
-400
0.~~.R
~....
~.....
~ *G~ ~..........
EI ....
6GDR t ?N R
.............. ...........
S ....
F0gue00
13-28
6K .......
3........7.............7------
.........
.. . .. .. . .. . . .. . ... .. . .. . . - - - -- .. .. . .. .. .. . . . .. . ... . . . ..
100 ............. ... .................. ....... ............
0 . 0E.
0 .. . .. ..I
. .. . .. .. . .. . .. . ... .. . .. . . .. .. . .. . .. . . .. .. . .. .. .
Figure 6e
14-1
par
Yves MARTIN-SIEGFRIED
DASSAULT AVIATION
ABSTRACT:
1- INTRODUCTION:
3- PRINCIPES THEORIQUES:
d2 X dx
2
Mat + [C~) 51 + C K, ) X, = Fa + Fsys/avion
Iq az x
adt a dt C3 Jx a sys/avlon
avec:
M.~~=~~PW2 ~~KM, (a
f. ~ PT Fa ; fs/avion =[ Wa Id TFsysavfon
avec:
3-3- COUPLAGE:
'q=i
14-6
q i:D.D.L. inddpendants de x
Par ailleurs:
=SS2vo
- X' jT
11 f ,aon/sys
d2 Y dY aq
avec Y= i
-Coup1acie en flexibilit6:
aq
aq a
avec [ G 1= E-d] C Kadyn aT
on 6limine favion/sys
MI q+ ( Cf]q+[K q f a
dt' + K
aq4
+~~~8 [ - dyn + f.)
14-7
3-4- INTEGRATION:
Le syst~rne d'4quations A intdgrer est du second
4 degrd:
2
+ ~~~~-[ Y](k~) Y = F
Y = N1 Y1
W(M
J~
kIY ~
IYdt 2+[C
~~N k IY, dt +~EKE] Y, N1 )dt=O0
Kdyn ) AY = F -(V,, ) F=
avec:
F =F 0 + Fdyn
[V,,. ] tENF;, V )T AY
Venfa~ = a, ENF,,j + b
Formulation variptionnelle:
P(AENF,,) = I EF 7 [B] EF
- ENF,T B f' ENF.Q
2 NFT B 11 ENF
.1AEFT [ F +EF
2 Kdyn F1+-'
Mdthode de rdsolution:
On utilise une m~thode itdrative avec techniqye de
line-search trbs perfornante qui conprend A chaque itdration:
aP(AENFa.1- + Pv 1 ) = 0
V = (x, Z, U, V, W, T
avec:
Equations d'dcauilibre:
avec:
s l'abscisse curviligne
= v uy + ut
v ,y= +
.y ,t
F = V W ~
as =p + x. 2 +z 2,Y
ay ,
p~~I+ . (1)+j2y)(iU~
2
~~~~~~~2 ~vv +v x. 2 + z. 2 2
y (V V. + V.+,
2
1 +
+ 2 + j )vw )= 2
14-l1
Conservation de la Masse:
avec
2 2
2
= ( + X .y + z .y) (dy)
,ds,)
+ V.ydt)2 + W.ydt) 2 ]
2
(dst+dt) = [(i (xy + u ydt) 2+ (Zy + (dy)2
Elongation du cable:
Par definition:
= ( (ds,)2 - (dy) 2
2 (dy)
En differenciant on obtient:
Comportement elastique:
6W, bs 6s T dy T 6y
6y J T/ 1 +2yb 6y dy
T = P 4- 1 + 2y (6)
w
=v- v Z. (8)
V1t = A V + ,3 ; T = f(
nl A11 - x -
*~y (aK)
x -- n)
at) = - ., - ,. Ay
-.
4- EXEMPLES D'APPLICATION:
4-2- Impact:
5- CONCLUSION:
BIBLIOGRAPIIIE:
McGRAW-HI LL.
14-15
SFELAND
ORGANIZATION CHART
- rigid
or Lagrange discretization
- flexible in finite element basis
or in general load reduced basis
COUPING
Planche 1
14-1t
RAFALE
Planche 2
14-17
V)c
CDC
C Lu
LOo,
CDC
AW L
X.
Ur
Ct z
MIRAGE III
PERFORMANCE ON REPAIRED RUNWAYS
MAIN GEAR
SHOCK STRUT AXIAL LOAD (%OF STATIC LOAD) F(VX, S)
S METERS L E .
707
Je1.::
Planche 4
14-19
MIRAGE 2000
MAIN GEAR -DROP TEST
COMPARISON OF CALCULATION AND TEST RESULTS
9.
8
7 ...
-2
Planche 5
14
SUPER ETENDARD
DECK LANDING
NOSE GEAR
K!LONEWTONS
200......... . .- .........
. .--- >1...
.
.. . .i. .0
- C.ALCULATION
TEST
Planche 6
ETENDARD IV
CATAPULTIIG
MAIN GEAR
100
50
- -- - CALCULATION
Planche 7
14-"
DL-
.-
Planche 8
14-23
FFLAND
RAFALE CATAPULTING SIMULATION
F
OIrI DF [I A TGL Iii 1 [11ATi,4
N I EF T1
6.
4.
3.
by
INTRODUCT IaN.
Landing gears, back-up structures and other major components of the Fokker 100 aircraft,
are designed in agreement with the loading conditions as laid down in JAR and FAR
requirements.(1,5]
The limit load conditions during landing a. based on a vertical velocity of descent of
10 fps combined with a maximum landing weight, a critical centre of gravity position and
a critical forward velocity, the ultimate load conditions are derived by applying a
factor of 1.5 to the limit load conditions.
The introduction of automatic landing systems on aircraft has led to additional
airworthiness requirements, which apply a statistical approach to the loads as developed
during automatic landing.
The additional requirements are laid down in the JAR-AWO.(4)
The following results and requirements are set forth:
* The safety level in automatic landings may not be less than
that achieved in normal landings.
* The probability of exceedance of the sink rate for which the aircraft has been
certified for structural loads,
- must be:
1 10 in average conditions
-s
S10 in limiting conditions
The probability of exceedance of the lateral velocity or slip-
angle associated with limit structural loads, must be:
-
10 in average conditions
S 10-6 in limiting conditions
(A limiting condition means that from all input distributions one distribution is set at
a maximum value. In this case the air turbulence distribution is set at a maximum value.
ref. (41)
The JAR-AWO approach suggests that, for instance for vertical loading conditions, the
descent velocity of the aircraft is the determining quantity.
That implies that if it can be shown that the probability of exceedance, in the sink rate
-
distribution, of the 10 fps point is l then 10 (average), the limit loads in
vertical direction are exceeded with a probability which is lower than once every 10.
landings. So on the base of the sinkrate distribution judgement is passed about the load
cases.
This is also applicable for lateral load conditions with respect to the aircraft slip-
angle (or lateral velocity) distribution.
From the physical and statistical points of view it is not true , however, that if the
sink-rate distribution meets the requirements, also the "corresponding" load (case)
distributions meet automatically these requirements. Therefore in this paper a more
direct approach is given based on the loads themselves; in this context the additional
requirements are reinterpreted as follows:
the probability of exceedance of
- a limit load must be a 10-- in average conditions.
-
- an ultimate load must be 1 10 4 in limiting conditions.
The static strength justification in this paper is based on three load cases, which are
assumed to be representative for the landing gear and the aircraft,
a) the main and nose landing gear bending moments at the lower bearing.
b) shear force jump and bending moment jump between front and rear spar
of the fuselage.
c) the side-stay load (main landing gear only).
In the future, however, the static strength justification could be based on more load
cases.
From these three load cases only the bending moments at the lownr bearing and the side-
stay loads shall be dealt with extensively.
For the sake of completeness the distributions of loads of the regular FAR/JAR
requirements, collected from the simulation runs, are also presented. (TABLES XV, XVI ard
XVII ).
Statistical information about loads (and stresses) in any section of the landing gear or
aircraft can be collected by means of Monte Carlo simulation of the landing process.
For this purpose the aircraft with full aerodynamics, undercarriages and tyres, is
modelled with six degrees of freedom. Control laws are included, representing the
behaviour of an automatic pilot.
Windehear is ignored during the landing process.
Throughout this paper flapsettingq of respectively 42" and 25' are assumed.
For a detailed derivation of equations of motion reference is made to [8].
15-2
One of the most convenient methods for solving problems involving input data with known
or assumed probability distributions is the Monte Carlo simulation.
This method involves a repeated simulation process using a randomly collected set of
input data, from the probability distributions, in each simulation (run).
A sample from a Monte Carlo simulation is similar to a sample of experimental
observations, therefore the results obtained by Monte Carlo simulation may be treated
statistica!ly.
The first question is, what number of runs is required for validity? Basically the
number of runs depends on the risk level one wishes to cover. That means, if acceptable
risk levels are in the order of l:lO a minimum of 10 runs is required.
But now a practical problem arises.
For this landing simulation one run requires about 5s CPU time, thus 10- runs require
roughly 1400 CPU hours, which is impractically long.
So the high number of runs must be reduced to a practible number of runs.
A widely accepted method is to assume normally distributed output data. Then, by means
of the X-distribution, it can be shown that a number of runs of 1500 is acceptable.[7]
Of course with 1500 runs , only a statement about a risk level of 1:1500 can be
substantiated.
So if the desired risk level is in the order of 1:10- one must try to find a possibility
to extend the probability of exceedance curves down to this level.
A simple method is to plot the probability of exceedance, on for instance normal paper,
and extend the curve by "fitting by eye" down to the desired risk level.
In applying this method one extrapolates beyond the validity range of the data set and
(possibly serious) errors can occur.
A better method is to find out which probability distribution is followed by the output
quantity.
A wide scala of statistical methods is available to test the hypothesis wether or not
data sets are normally, log-normally or otherwise distributed.J2] If one is able to
identify a statistical distribution which fits the data well, it is possible to read the
probability of exceedance plot at the desired (low) risk level. If one is no able to
find such a probability distribution, there are possibilities to transform the data to
new data which hopefully will fit a known distribution.C31
Another method which works reasonable well under certain circumstances is: try to fit
only the upper part of the distribution by a known distribution.
If all the methods fail, only one possibility remains: "fit-by-eye".
If it is known that the load (or stress) under consideration posseses a physically upper
boundary below the limit (ultimate) load, there is no need for extrapolation to the low
risk level. The load can simply not exceed this value.
So a high number of runs has the advantage to make the estimation of variables at low
risk levels more certain.
As allready mentioned a number of 1500 runs is used in this paper.
In order to reduce the amount of input data for the landing simulations, some realistic
assumptions are made:
- the aircraft lateral velocity as well as the rotational velocity
are zero at the moment of touch down It-C).
- the runway friction coefficient, the airfield height and the ISA-deviation
are not varied because no realistic distributions could be found.
However, a p,,.=w.8, an airfield height zero and a ISA are
probably conseriative with respect to the output distributions.
Taking into account these assumptions, the landing impact simulations only require
probability distributions of the following set of input data:
- mass of the aircraft (GRAPH I, TABLE I)
- moments of inertia I., I, and I.. (TABLE I)
- centre of mass. (TABLE 1)
- aircraft ground speed.
- aircraft airspeed.
- aircraft descent velocity.
- aircraft pitch, yaw and roll angles.
Detailed information about the input distributions is given in the next chapter.
KNPUT DATA
In GRAPH I a scatter plot of the aircraft mass versus the aircraft centre of mass is
given. The boundaries of the official mass and centre of mass diagram for this aircraft
configuration are indicated.
The graph shows that in 4.3% (65 out of 1500) of the landings an "overweight" landing is
assumed.
In TABLE I relevant statistical parameters of mass data are summarized for different
flapsettings and in average and limiting conditions.
In TABLES It, II, IV and V statistical parameters of airspeed, groundspeed, pitch, yaw
and roll angles are presented.
15-3
The mathematical model of the aircraft landing process used to get the results given in
this paper, is implemented in the computer program LANDAU.EB) This model is validated by
examining the vertical and the lateral behaviour of the aircraft.
The vertical behaviour of the aircraft during the landing process is calculated by means
of a set of equations which are in agreement with the equations used for the SIulation
of droptests and the determination of the deterministic landing loads.
The lateral behaviour can be checked by comparing measured loads and accelerations with
calculated loads and accelerations.
Comparison of measured and calcula- 4 loads show good agreement.
BendinQ moments at the main and nose landina pear lower bearings.
In GRAPH It a scatter plot of left versus right hand main landing gear bending moments
°
is presented for 8,42 in the limiting condition.
By this scatterplot it is suggested that the probability distributions of left and right
hand landing gear bending moments are "similar".
A distribution-free statistical test confirms this hypothesis indeed.(2]
This is equally applicable to the 8-=256 (average, limiting) simulations.
For this reason the events of left and right hand main landing gear are taken together.
In TABLES VI and VII statistical parameters of the bending moment distributions are
presented.
Probability of exceedance plots on normal (Gaussian) paper (see GRAPH 1I1) show that in
none of the four cases the fit against the normal distribution is acceptable. (Compare
the Anderson-Darling statistic (AD) with the AD-90% value. If AD>AD-90% the hypothesis
of normally distributed bending moments is rejected).
So "extrapolation" procedures (as mentioned already in the chapter Monte Carlo
Simulation) are examined .
Cce it is decided which "extrapolation" procedure is the most valid one, it is possible
to read the bending moment values, at their desired risk levelsfrom the probability
distribution plots. (see GRAPH IV as an example).
These bending moment values can then be compared with the limit and ultimate bending
moments available from deterministic load calculations.
The results are summarized in TABLE VIII.
- a
25 average i:0 upper-half/{nor 109000 130560
-
25 limiting I:l0- by eye/lnor <100000 195841
42 average L:IO upper-half/Inor 106600 130560
-
42 limiting 1:10 bye eye/Inor <150000 195841
A more satisfactory approach is to define an ultimate or limit load (or stress) by means
of its value at the desired risk level. This approach meets 'ar more real world
experience of people.
For the nose landing gear bending moment the same methods and procedures are applicable.
A summary of the results is presented in TABLE VIIIA. Statistical parameters of the
distributions of the nose landing gear bending moments are presented in TABLE IX. (see
also GRAPH V). (Although the teststatistic AD>AD-90%, these distributions are accepted
because it is the best obtainable result).
25 average 1:1O-'
-
by eye/weib. <15000 29078
25 limiting 110 by eye/weib. <15000 43616
-
42 average 1:10 bye eye/weib. <20000 29078
-
42 limiting liO upper-half/weib 17320 i43616
The side-stay loads are examined in the same way as was done before for the bending
moments.
It appears that no known probability distribution fits well.
The results of the "extrapolation" procedures are given in TABLE X.
A summary of statistical parameters is presented in T SLES XI up to XIV inclusive.
Conventional load cases distinguish between push and pull loads. So for every set of
°
simulations ()8,.25x,42 and average, limiting) two load distributions must be
investigated.
TABLE X Main landing gear side-stay loads.
For the sake of completeness the parameters of the probability distributions of all
landing loadcases for S.,42' (average only) are presented in the TABLES XV up to XVII
inclusive.
In TABLE XVIII the corresponding deterministic JAR/FAR loadcases are given for
comparison.
If one tries to compare the simulated loadcases with the deterministic loadcases a
problem arises.
A loadcase is not the load itself but a combination of loads (and shockabsorber
deflection) which define a certain situation as asked for by JAR/FAR (the spin-up or
maximum vertical reaction loadcase for instance).
To compare loadcases one wishes to have the multidimensional correlated probability
distributions of the relevant parameters by which the loadcase is defined.
For instance the spin-up loadcase is defined by the spin-up (drag) load together with the
corresponding vertical load and shockabsorber deflection. Each of these three parameters
has its own probability distribution and in most cases these distributions are also
correlated. So a three dimensional, correlated, probability distribution must be found
which describes the probability distribution of the loadcase.
In most practical situations it is very difficult to find such kind of distributions and
one is forced to use approximation procedures to be able to extrapolate the
probability distribution to the desired risk level.
The whole proces is a very complicated one and in most cases no probability distribution
can be found which fits the loadcase well. Also extrapolation by means of "fitting by
eye" can not be performed due to the three dimensional character of the distribution.
On the other hand it is sufficient to determine loads or stresses directly for any
particular section where stress calculations are made. This makes at least the
probability distribution of the determining load for this section one dimensional.
A more direct approach is thus to calculate the loads directly which are needed to make
a stress calculation for a certain section of the construction.
From this point of view the whole concept of loadcases is superflous (apart from
preliminary design)..
Another point is that nowadays the FAR/JAR loadcases define the limit or ultimate loads
against the structure is designed.
It is, however, more conform the daily reality to define an ultimate or limit load
(stress) by means of its values at the desired and accepted risk level.
It is known that the fatigue-damage of a varying load depends primarily on the amplitude,
the variation in load, rather than the absolute load. level reached. (6]
That means that maxima (or minima) of load time histories, as collected from Monte Carlo
simulated runs, can not be used for fatigue purposes. From the fatigue point of view the
whole time history of loads is of importance and must be collected and stored rather than
only the maKima (minima) of the time history.
If the time history of a varying load in a section is known, this time history can be
analyzed by means of, for instance, a rainflow counting method [63.
Due to the computer storage capacity it is, however, not possible to store the whole time
history of every relevant load. An acceptable burdening of computer memory is obtained
if only a few relevant Fourier components of each time history are stored.
Fourier analysis of the time history is then performed by means of a Fast Fourier
Transform (FFT). By means of the inverse Fourier transform the time history can be
reconstructed (approximately).
All kinds of (rainflow) counting methods can be applied afterwards to these reconstructed
time histories.
Unfortunately at the moment this paper was written, no calculated data were available so
no example can be presented.
15-5
REFEREEs.
3. Hahn 6.J. and Shapiro S.S., Statistical models in engineering, John Wiley and Sons,
Inc. 1967.
MASS CS lI Ii
t
(kg) (% MAC) (kgmt) (kgm2) (kgm )
TABLE I
TABLE III
°
LIMITING 6..42 V.. Vm -OU I VI.m.. PITCH ROLL YAW
(m/s) (M/s) (fps) (degree) (degree) (degree)
TABLE IV
°
AVERAGE 8..-25 V.m Vmou. Vl... - PITCH ROLL YAW
(m/s) (m/sI (fps) (degree) (degree) (degree)
TABLE V
8S. - 42* M.. (Nm) M.. (Nm) M,.. (Nm) M.... (Nm)
8,- = 25" AVERAGE PULL (N) PUSH (N) PUSH (N) PULL (N)
8.. - 25* LIMITING PULL (N) PUSH (N) PUSH (N) PULL (N)
°
6., 42 AVERAGE PULL (N) PUSH (N) PUSH (N) PULL (N)
6- = 42' LIMITING PULL (N) PUSH (N) PUSH (N) PULL (N)
Legend.: MN denotes that the loads are applicable to the main repectively
nose landing gear.
MVR denotes Maximum Vertical Reaction.
Lateral drift for the nose landing gear is no regular loadcase.
16
S 120 'VHRAG TA21K AS
- I2
( 300 .I 1'..
130 .S 0., 7'771 7 9
1 0 IfII 1i 0 9-8, 1. -
9 qII 136 - I -)
IA I 1
d1
0h k a bE f o~ /Ao r.
0 kA
at er-b d Ifctit 41/~~
0 ~(N) (N)
~ 1
( N)
.0S2R/~~,A~
)
A 2M' 41V
(n-)
MV
O
(mr)
~ /3V
N/N NE!l o in
-220 1
20 2017 276 l.A4 25800S~ 38213
32 1720 7.2 3 0 (i 4 180
-'.....,,,~ II.i~~
1 R 1 F,
I CA 7,,
71 111-
~0NVENIEALRAT1O
AVE RAG F A II XV II
F ()'y(N)
482, A A2 F, N A I~ ~I ~
~m
1~~ f r Iod an v t Irsc
n I
_I A 1 I . ,
509Mn- 10 A 0A~A c
5e860
25-5 - IN
KvMR
TAU--
I fain -x t~t LI - ol 1oc
o .s
'tlI:~~, 14
id~lLt,A lIr we r Ia rt.
2 .r I
5-o
sao
&0
A 'V Q AA *N :-Ax
95.0~- A.O-MS
9.901
Sax
kove 610
si 2104
N A':
~99.99 at a.966
Bet. 8.2677
C 5.29
98.8~.
.177
8CO
-. -------
.8-A
~*, ____________
______________________ ____A
99.
.,:oal Q. t.
-t-, a. int..
or
16-1
by
G.L.GHIRINGHELLI
Dipartimenco di Ingegneria Aerospaziale,
Politecnico di Milano, Milano, ITALY
and
M.BOSCHETTO
Air Vehicle Technology Department,
AERMACCHI S.p.A., Varese, ITALY
SUMMARY
This paper presents some significant applications of the integrated system GRAALL (Ground
Roll Air And Landing Loads) to the analytical prediction of aircraft landing loads
carried out at Aermacchi.
The capabilities of the system, able to treat both rigid and flexible models, make it a
tool that can be profitably used during different phases of the design process.
The results reported herein describe the whole development of an actual design applica-
tion; comparisons between analytical and experimental data are also provided.
LIST OF SYMBOLS
Mlan design landing weigl at gear
Wlan design landing mass at gear
Vv design vertical velocity
FVstatic vertical static load
FVmax maximum vertical load
Sshabs shock absorber stroke
Stire tire deflection
Nshabs shock absorber efficiency
Ntire tire energy absorption efficiency
Whose design mass at nose landing gear
Wmain design mass at main landing gear
W aircraft design landing mass
U friction coefficient
Ki touch-down load factor
K design load factor
Ngear landing gear efficiency
Sgear wheel axle stroke
INTRODUCTION
The need to satisfy more and more demanding performance requirements in all operational
conditions makes integration of all design phases a must for the designers of modern and,
in particular, of innovative aircraft.
Landing gear design is no exception in this respect. While still giving due consideration
to the general operating chnracteristics of the system, the landing gear, like any other
aircraft system, should n, be designed merely to match the airframe, but together
with it, with the aim of achieving the optimal solution, whi'e the tendency is often to
neglect the landing systems in the early design process.
The assessment of the ground loads and gear structural design are interacting activities.
each generating a mutual feed-back. The interactions between aircraft structure and
landing gear are also very significant, and the loads applied by the gear are much
influenced by aircraft elasticity and gear configuration. It 2.s therefore essential to
consider all the effects of structure elasticity and to keep updated the load
computational model.
The solution found at Aermacchi, in cooperation with the Department of Aerospace Engineer
ing of the Politecnico di Milano, consists of the system of integrated programs calleu
GRAALL (Ground Roll Air And Landing Loads). This system permits an interc isciplinary
exploitation of a single data base so that each engineering area (flight mechanics, flut-
ter, landing gear, etc.) may profitably and timely use the most up-to-date and/or suited
structural r,.o
del.
The first results obtained and a detailed description of the analytical model are given
in l].
This paper provides a model outline limited to the simulation of landing and ground
handling, and emphatises the central role of GRAALL as effective working tool for
integrated landing gear design.
16-2
A set of compari3ons between simulation and experimental data from drop tests and landing
tests is also provided.
DESIGN PROBLEMS
A possible flow chart of the activities concurring to the full development of design is
shown in figure 1. The flow lines are solid for operations and dashed for design loops.
The preliminary phase of the study involves the integration of the data from the
most important applicable standards and specifications (MIL, AVP, AIR), and the study of
the energies involved at touch-down according to the well known relation (6] :
Repeat.ag this computation for different design assumptions permits the ground reaction
behavior for changing shock absorber strokes and sink speeds to be assessed.
The obtained charts also enable a first evaluation of the paramete-s that concur in the
definition of the design landing load. Figure 2 depicts a few examples of how the
maximum vertical load changes with varying shock absorber strokes and sink speeds.
Two methods may be used to calculate the design load in this phase:
A) the energy balance method; in this case the energy absorbed by each landing gear is:
ENERGY ABSORBED = 1/2*Wlan*Vv**2+(Mlan-lift)*(Sshabs+Stire)
and the maximum vertical reaction is given by the product of static reaction by the
design load factor:
FVmax = FVstatic*K
at this Foint, if the tire and shock absorber absorption efficiency, drawn from
literature or previous experience, is used, it is possible to write:
K = i/2*Vv**2+((Mlan-Lift)/Wlan)*(Sshabs+Stire)
g*(Nshabs*Sshabs+Ntire*Stire)
B) gear pertaining mass method; based on energy considerations and directions provided
by MIL-A-8862, mass can be computed as follows (see fig.3 for geometric parameters):
The maximum vertical reaction will, therefore, be given by the product of the above
mass by the touch-down load factor:
FVmax = Ku*(Want or. Wpri)
where
Ki = Vv**2/(2*g*Ngear*Sgear)
After calculating the necessary terms, the design load factor can be defined:
K = FVmax/FVstatic
The experience acquired so far has demonstrated that method B is better suited to this
phase of design, in which the type of the landing gear is still undefined.
At this stage, there starts an iterative cycle including in the loop draftsmen, system
and stress engineers; this aims at producing the first hypothesis of a structural and
kinematic solution.
After the system geometry and the order of magnitude of the loads the structure is
required to absorb are known, the parameters needed to define the shock absorber model
are available. Obviously, the type of shock absorber depends not only on the energy to be
dissipated, but also on the landing gear configuration (conventional design, leveled
suspension, triangulated), and the choice is very frequently dictated by the need to
design a retraction system compatible with the aircraft structural configuration.
In this phase, however, the analysis should not be confined to touch-down, but should
encompass also the aspects related to ground handling and bump traversal. This is why
there is needed a tool allowing several hypotheses from the dif' rent engineering
areas (system and stress engineers, etc.) to be analyzed, and able to help design the
shock absorbers most suited to the intended configuration. Moreover, if it is considered
that many of these parameters are interdependent, it becomes evident that the problem
to handle features many variables and calls for a flexible computational procedure, i.e.
a procedure that adapts to all interwoven problems and ensures the largest possible
integration of the different engineering areas. Failure to integrate would in fact give
rise to subsequent design problems able to be removed only through compromise solutions
impairing aircraft optimization.
GRAALL system enables the determination of the loads acting on an aircraft for many types
of maneuvers, taking into due account the flexibility of the structure. It is comprised
16-3
of several computer programs and of the interfaces with some other programs used at
Aermacchi [2, 3].
A few characteristics of this method are described hereafter and refer to touch-down,
ground roll and taxiing only.
GRAALL system can be divided into three main portions (fig. 4):
[KI 0
IM] = R
M W MVS K W K VS
1i I- t* I* I
V S
T (2)
[M I- (o I ( M I (J
SK1- (01 IK K 0 1 10 1
a
The data base contains also the information related to the distribution of displacements
and internal forces and/or stresses, corresponding to the modes inl* I . These matrices
16-4
-
- motion due to shock-absorber stroke
steeering motion
/3
- tire rotation -
Rotations 0 ,'Y, and 9,ewhich describe these motions are dealt with as independent
degrees of freedom of the mechanical system, and are integrated with respect to time.
It is worth pointing out that as this description is made with rotation axes and arm
dimensions, it can describe also very general landing gear configurations.
The shock-absorber is oleo-pneumatic, and details such as:
- displacement of the air/oil separator piston;
- independent variable flow orifice for extension and retraction strokes;
- variation of the viscous coefficient with cross-section
are considered [5, 6, 7].
The degree of freedom related to the rotation of the wheel may be integrated starting
from different initial conditions so that drop tests with pre-spun wheel or wheel spin-up
in flight, if required, can be simulated.
The computation model of the forces deriving from tire deflection includes the pressure
variation effects and realistic models for the evaluation of longitudinal and side
friction [8].
The possibility of moving the steering through an open-loop control is provided, and so
is the simulation of the control linkage chain through data representative of its
compliance and damping.
the drag brace. From the plot, it is possible to observe that the load on the drag brace
is higher in the case of an elastic attachment.
Figure 12 shows some comparisons between the axial force on a structural element and
vertical load, obtained from the calculations, and the experimental data from the drop
tests of the nose landing gear. Figure 13 in turn plots the simulated actual behavior of
vertical loads versus shock absorber stroke at two different sinking speeds, for the main
landing gear.
An elastic model of the landing gear and an elastic model of the complete fuselage
(supplied by the structural analysis engineers) are available at this point; this allows
a flexible model of the complete aircraft to be prepared, which can be used to analyze
the aircraft behavior during different maneuvers, and to evaluate the final loads.
Thence, not only the time history of the loads applied to the landing gear, but also the
stresses acting on all the elements of the complete aircraft model can be obtained.
Figure 14 shows some of the stresses affecting elements of the complete aircraft
(fuselage wings and empennages) at touch-down and during the first instants of ground
roll.
At this stage, a comparison of the final loads obtained with the complete aircraft
elastic model with those assumed in the preliminary phase, may highlight the margins
available for possible structural optimizations or improvement of aircraft landing and
ground handling performance (Fig. 15).
An experimental test carried out on an aircraft with nose landing gear fitted with strain
gages, permitted the loads generated during and after landing with brake application to
be acquired; then, through the computer simulation of this landing it was possible to
perform a comparison between the theoretical and experimental data, and to evaluate the
simulation reliability.
The results of the comparison are given in figures 16 through 18, that show a
satisfactory agreement between the numerical and experimental data.
The availability of the aircraft elastic model allows the simulation of entire landing
maneuvers, even of those which would be difficult to perform experimentally and a careful
evaluation of the extreme operational possibilities without taking up on the burden to
carry out experimental tests that can be very hard to set up;moreover the animation of
the structural response can be very useful in the analysis of a large amount of data.
Figure 19 displays some frames of the animation developed by using GRAALL; the
structural model has been paneled in such a way to pictorially reproduce the actual shape
of the aircraft; the figure delineates the maximum stroke condition, the deformation is
magnified to highlight its behavior.
At this point, the designer can verify whether there exist problems of taxiing/ground
roll instability, and modify the stiffness of some components or study additional
damping devices, still in the early design phase.
Fig. 20 shows the conditions for possible shimmy initiation on a conventional desig;1 land
ing gear due to brake application; presently this condition cannot be fully investigated
with GRAALL because the available tire model is still unsuited to describe this
phenomenon correctly.
Eventually, the flexibility of GRAALL allows collateral problems to be investigated, such
as the presence of overswing at touch down or during ground roll/taxiing on bumpy or wavy
runways in the case of aircraft with high aspect ratio wings and/or carrying underwing
stores: figures 21t22 shows the time histories of the displacement of the extremities of
an underwing store at touch down and bump traversal and the animation of ground
clearances during asymmetric landing.
DEVELOPMENTS
It is at present envisaged that the developments of this procedure will occur along
three main guidelines:
- description of tire strain through independent parameters, to correctly simulate shimmy
phenomena;
- integration in the structural optimization procedure already in use at Aermacchi (3);
- implementation of an active shock absorber and utilization of the integration procedure
as an analysis tool in the design of the control system.
CONCLUSIONS
The experience gained with the use of GRAALL for the evaluation of the landing and ground
handling loads proved that this system is easy to use, reliable, and is a flexible tool
in the hands of the designer. The use of physical quantities in the model contributes to
make this system particularly "user-friendly". The designer has in fact in this way an
immediate perception of the actual values involved, and only minimal tuning is required
to achieve satisfactory results.
The unified approach typical of this method permits the same model to be utilized for the
simulation of all types of in-flight and ground maneuvers.
On the other hand, a few critical aspects have emerged:
the estimate of structural damping to assign to modes: if it is underestimated, the
internal forces show increases (decreases are noted if, conversely, it is
overestimated), while the kinematic terms and ground forces in the response do not
change significantly;
16-6
- the selection of the modes to keep in the analysis set: their evaluation as far as the
participation coefficients are concerned, is very important for the results it is
desired to achieve; these can, in fact, be required in terms of constant response
(both kinematic or ground reaction forces), or in terms of internal forces.
Eventually, GRAALL proved to be not only a suitable tool for the determination of the
ground loads, but also a simulation procedure fit for many other applications and provi-
ding a solution to many of the problems related to the study and development of an air-
craft as far as landing, take-off and ground handling are concerned.
REFERENCES
ACKNOWLEDGMENTS
The authors wish to thank prof. V.GIAVOTTO and prof. P.MANTEGAZZA of the Dipartimento di
Ingegneria Aerospaziale, Politecnico di Milano, for their precious help, and Mrs. R.
Fischer of the Aernacchi Translation Unit for preparing the English text.
FYMAX
MILSPEC 'REQIREM
PSEUM.tL--
- .. ....
...
S.ABS -,.30 M
I<
e t
SSN~s.u25 (fl
535185 .50 (PI)
FIG. 1I Landing system design flow FIG. 2- Parametric diagrams for gear vertical
loads evaluation
16-7
C.G. POSITION
A) B)
FIG. 5- F.E. dynamic model: A)complete aircraft, B) gear model for drop test
16-8
'y V )N 11.1U-
.. - .... ..
RIGIDRODEL
GEAR
NOSE RAINGEARR6O MOW.
VERTICAL
VERTICAL LOAD
'GAD 1.5
o.. L ' 6 0.
. . ' g
TIFF (SEC)
B) TIFF (SEC)
A)
I AXIAL MODEL
ELASTIC
FORCE
ATTACH~MT
RIGID1
I )~ - -ELASTIC ATTACHENT
AXIALo VICA
ELASTICM MODEECAT
NONDIMESHCICBSOBL
I. AXISTRLOA
FORCE A)
~
FIG. ~12-Cmaio
ewe xeietldt
and/ nueiaIeut
frado et
~~ ~
A)~~~~~~~~
B)~~~etclla
nndmdabrc
W]frevtim
hc /o-l
bobrsrk
16-10
SIWING SPEEDLIMIT
VERTICAL501 SINKING
SPEED
LIMIT
LOAD
VERLOAD
LOAD EXPERIMETAL
DATA
-- ELASTIC MDEL DATA
DATA
E__XPERIMENTAL
STROKE STROKF
ELASTIC
MOEL
UEAIELMN
SHEAR ELEMENT
WIN-AGM
PLANE
I . ENPENNAGE
ELEMENT ILANDING ELEMENT
GEAR
REND
ING
MOMENTDESIGN LIMITLOAD
.1 . INE(SEE)
LOAD
TiIIN LiMTT
IT 1,T i
RI - PANE I
TIME(SEC)
SHOCK ABSORBER
VERTICAL - ELASTIC MODELDATA 1.STROKE
1, LOAD --- EXPERIMENTAL DATA
I I~4
L
I I- I
ELASTIC MODEL
DAT
0.
0. k
2.5 5.IM 7.5 C 2.0.iM SC
5 AM(S)
GEAR TIE(E)25
NOSE
NOSE
GEAR
FIG. 1le Comparison between numerical and FIG. 17 - Comparison between numerical and
experimental non-dimensional loads experimental non-dimensional nose
shock absorber stroke
SHOCK
ABSORDEN
STROKE
.5
0.
2.5 5. 7.5
TIME(SEC)
ELASTIC
MODEL
. VRIA
ICAL
3. VERT
DISPLACEMENT DSLCMN
2. 2.
2.TF2.E)7 IM SC
FIG. 21
Grundclarace f te ndewi stre
-PINT C LIN
GROUND
TIME(3qC) LIN
AROUND
A) B)
and
SUMMARY
Aircraft dynamic loads and vibrations resulting from landing impact and from runway and
taiway unevenness are recognizd a, significant factors in causing fatigue damage, dynamic strcss on
the airframe, crew and passenger discomfort, and reduction of the pilot's ability to control the aircraflt
during ground operations. One potential method for improving operational characteristics ot aircratt
on the ground is the application of active-control technology to the landing gears to reduce ground
loads applied to the airframe.
An experien tal investigation was conducted on series-hydraulic active control nose c ear Ilie
experiments involvcd testing the gear in both passive and active control modes. Results of thi,
invrstigation show that a series-hydraulic active-control gear is feasible and that such a gear t,
effective in reducing the loads transmitted by the gear to the anrframe during ground operation,.
INTRODUCTION
Aircraft dynamic loads and vibrations resulting from landing impact and from runwav irid
taxiway unevenness are recognized as significant factors in causing fatigue damage. dynamic stresI on
the airframe. crew and passenger disconfort. and rcducti,.i f 0.c
i. _il'o , arility to control the aircraft
during ground operations. The ground-induced structural vibrations on large, flexible airplanes can
reduce the pilot'% capability to control the airplane during high-speed ground operations. These
ground-induced dynamic loads and vibrations are magnified for supersonic-cruise aircraft because of
the increased structural flexibility inherent in these slender-body, thin-wing designs. Such
operational problems with supersonic--cruise airplanes have occurred at high take-off and landing
speeds on some runways which are only marginally acceptable for itiost subsonic coririercial
airplanes One potential method for improving operational characteristics of such airplanes on the
ground is the application of active-control technology to the landing gears to reduce the ground load,
applied to the airframe.
Previous analytical studies ireferences I and 2) have been conducted to determine the
feasib;lity and potential benefits of applying active load control to the airplane main landing gear to
limit the ground loads applied to the airframe. The results reported in reference 2 indicate that a
shock strut incorporating a hydaulically controlled actuator in series with the passive elements of a
conventional shock strut have acceptable properties and would be quite fea'ble to implement. Based
on the results of reference 2. a modified version of the series-hydraulic active gear which eliminated
the actuator and effected control by using a servovalve to remove or add hydraulic fluid to the shock-
strut piston (lower cylinder) was analytically and experimentally investigated in references 3 through
6. Based on the results described in these references, tle gear from a F-106B was modified for drop
tests. 'The purpose of this paper is to present the results of passive and active drop tests of the
F-106H nose gear.
SERIES-tlYDRAULIC AC'TIVE-CONTRO GEAR
Control Concept
The series-hydraulic control concept limits the gear force applied to the airframe by regulating
the damping force (hydraulic pressure) in the piston of the olco-pneumatic shock strut. To
incorporate this active control concept into a conventional gear requires a modification to the gear to
control the flow of fluid in or out of the shock-strut with a servovalve. A schematic drawing of a
series-hydraulic landing gear that has been fabricated to permit experimental verification of the
concept is shown in figure 1. The gear represented is a simple generic oleo-pneumatic shock strut
without a metering pin. The control concept is designated series-hydraulic because the control
servovalve is in series with the shock-strut piston and hydraulic fluid is removed from or added to
the piston to provide force regulation.
The actual gear selected for inclusion of the active control concept was the nose gear of the
F-106B with no meteuing pin. The gear was modified to accommodate the control by adding a three
tube arrangement to the orifice as shown in figure 2. A collection chamber at the top of the 3 tube,
connects the fluid in the shock-strut piston to one side of the secondary piston. The other side of the
secondary piston is connected to the servovalve. The purpose of the secondary piston is to
mechanically limit the amount of fluid that can be taken out or added to the shock strut for flight
safety.
The control hardware required for the active gear test program included a 200 GPM (0 76
m3/min) servovalve, a low-pressure (atmospheric) reservoir, a 9 GPM (0.04 m 3 /tnin) hydraulic putip,
a high-pressure (3000 psi (20.7 MPa)) accumulator, an electronic contioller, and feedback transducer,.
The isolation valve allowed isolation of the gear from the control hardware to permit passive gcat
testing.
System Operation
System operation is briefly described as follows. The electronic controller deterinei, the
operational mode (take-off or landing), and implements the control laws. The control loss
programmed into the controller are based on the following logic. At touchdown. the cotutollcr
receives a signal from a transducer to measure the instantaneous sink rate. Assutting a cotlstan
mass, the present energy is then calculated, An integration of tile acceleration is also begun at this
time so that the gear upper mass velocity is known at all subsequent tines. As the gear coitpicsses .
the remaining work capability of the shock strut is calculated using the instantaneous vAues ot
acceleration (or force) and stroke remaining. This remaining work capability is then compared swith
the present energy of the upper mass calculated using the instantaneous upper mass selocity. When
the remaining work capability equals or exceecs the present energy of the upper mass the controllcr
stores in memory the instantaneous value of the scaled acceleration (wing gear iite rface force) for u c
as the impact limit force and activates the servovalve control loop. The controller attempts to
maintain this force by removal or addition of hydraulic fluid from or to the oleo-pneumatic shock
strut lower chamber. Feedback from the acelerometer provides the controller %s ith a ineans it
determining the difference between the present and the desired force. The slope of the acceleriotctcr
outnut is also used for razc feedback in the control laws. so that if :he force is not at the proper lesel
but is tending to return to it on its own, the magnitude of the servo command would be reduced by
some amount. Likewise, force trends away front the desired le.el provoke servo commands laiger
than world be generated if using force difference alone in the control laws. When the upper rias,
energy has been dissipated and the sink velocity is nearly zero, the controller linearly transitions the
impact limit force to a value of zero for rollout control. During rollout and taxi the controller maintains
the wing-gear interface force within a designed tolerance (deadban of ±1750 Ib (±7.8 kN) for these
tests) about the static normal force. After control initiation at touchdown, the controller continuously
operates with a long-time constant (5 seconds) control to return the gear stroke to the designed static
equilibrium position
EXPERIMENTAL INVESTiGA'F I
Landing simulation tests (passive and active) with the nose gear from a F-106B fighter
interceptor airplane (fig. 3) were conducted at the NASA Langley Research Center to demonstrate the
feasibility ano the potential of the active gear for reducing ground loads transmitted to the airframe.
The vertical drop tests simulated to-chdown impact with and without lift.
Drop Tests
A photograph of the test apparatus for conducting the vertical drop tests of the nose gear is
shown in figure 4. Additional dtails of the gear and apparatus are shown in figure 5. Using the drop
test apparL.,us, the nose gear was dropped vertically with simulated lift at 4.5 fps (1.37 m/s) in both
the pasFive and active modes. A 1-g lift simulation was obtained by using crushable aluminum
honeycomb to stop the drop carriage (upper mass) vertical acceleration. The chosen test condition is
representative of the airplane being derotated at a high pitch rate. A second test of the gear was also
conducted at a vertical speed of 2.5 fps (0.76 m/s) without lift. Wituout lift applied, vertical speeds
higher than about 2.5 fps (0.76 m/s) would cause the gear to bottom out. Such a drop test is
representi.tive of losing pitch control during derotation.
A comparison of the measured upper mass acceleration for the active versus passive gear
without lift is shown in figure 6. Sigr 'ficant events such as drop carriage release, free fall, tire impact.
and control .,.:i ation are indicated in the figure. A 47% decrease in upper mass acceleration was
obtained with the active control gear. The decrease in acceleration translates to a 47% decrease in the
amplitude of forces transmitted to the airframe. For the 2.5 fps (0.76 m/s) vertical drop without lift,
the passive gear stroke shown in figure 7 nearly bottomed out: consequently, the active gear stroke
w.as essentially the same as for the passive gear case. Upper mass acceleration data for a 4.5 fps (1.37
m/s) drop with lift are shown in figure 8. A 36% decrease in the transmitted force was obtained with
the active gear. As shown in figure 9, there was a 10% increase in the strut stroke associated with the
active control.
CONCLUDING REMARKS
A potential method for improving the operational characteristics of aircraft on the ground by the
application of active-control technology to the landing gears to reduce ground loads applied to the
airframe has been investigated. An experimental program was conducted on a series-hydraulic
active-control nose landing gear from - F-106B fighter interceptor aircraft involving both passive and
active control modes. Results of the investigation show: (al That such a concept can be achieved
through modification of existing hardware, and (b) that the concept is effective in significantly
reducing the loads transmitted by the gear to the airframe during landing and ground operations.
REFERENCES
1. Wignot, Jack E.;Dump, Paul C.;and Gamon, Max A.: Design Formulation and Analysis of an
Active Landing Gear. Volume I, Analysis. AFFDL-TR 71-80, Vol. 1,U.S. Air Force, August 1971.
(Available from DDC as AD 887 127L.)
2. Bender E. K.; Berkman, E. F.;and Bieber, M.: A Feasibility Study of Active Landing Gear.
AFDL-TR-70-126, Air Force, July 1971. (Available from DDC as AD 887 451L.)
3. McGehee, John R.; and Dreher, Robert C.: Experimental Investigation of Active Loads Control
for Aircraft Landing Gear. NASA TP-2042, 1982.
4. Morris, David L.: Active Landing Gear Respo,. Testing. ADWAL-TM-82-177-FIEM, U.S. Air
Force, April 1982.
Loh
19-4
5. McGehee, John R.;and Morris, David L.: active Control Landing Gear for Ground Load
Alleviation. AGARD-CP-384, pp. 18-1 through 18-12, March 1985.
6. Morris, David .: F-4 Active Control Landing Gear Dynamic Response Testing. AFWAL-TR-85-
3005, Air Force, August 1985. (Available from DDC as AD-B096481L.)
Isolation valve
Secondary
Accelerometer piston
-1
controller
Hydraulic I
pressure P
transducer P
Low-pressure High-pressure
reservoir accumulator
Collection chamber
Leaptetiomete
47% decrease -
Drop carriage
release ,, " Active
Upper mass V
acceleration, Control initiated
g-units
-1.0 - Passive stroke
(both tests)
Free- Tire impact
fall
I I I I I
0 0.1 0.2 0.3 0.4 0.5
Time, sec
9
8 Active.
No change ..--
7
6
Strut stroke, 5
in.
4
3
2
1
0
0 0.1 0.2 0.3 0.4 0.5
Time, sec
Drop carriage
release
Upper mass
acceleration,
g-units
Tire impact
Fall 1 g wing lift initiated
9
8 Active
7 10% increase
6
Strut stroke, 5
in.
4
3
2
1
0 J
0 0.1 0.2 0.3 0.4 0.5
Time, sec
Figure 9. Comparison of passive and active control strut stroke for
vertical drop of nose gear at 4.5 ft/sec (1.37 tt/scc( witht lift.
20-1
by
Abstract
An active control undercarriage for the alleviation of aircraft landing gear and
structural loads during operation on rough runway surfaces is described. For quantita-
tive determination of the improvements obtained with an active control undercarriage
compared with conventional landing gear systems, aircraft taxiing is realistically simu-
lated by means of a laboratory test set-up especially designed for this kind of testing.
1. Introduction
Since many years the NATO countries have been concentrating on the problem of how to
improve the capabilities of aircraft to operate from damaged/repaired runways. The back-
ground and the goal of the thus related work is to guarantee the readiness of aircraft
even in the event of an airfield attack.
The many investigations performed have indicated that among others a considerable
effort has to be placed in the development of advanced landing gear systems for improve-
ment of the aircraft structural dynamic response behavior when taxiing on rough runway
surfaces. With regard to the landing gear system this especially implies to prevent the
bottoming of the struts when encountering a series of discrete obstacles located at
worst spacing conditions (1]. It has repeatedly been shown - by tests (2j and analyses
[11 - that the dynamic landing gear and aircraft structural loads encountered during re-
alistic taxi phases on damaged/repaired runways can significantly exceed the specified
landing impact loads [3,4,51, thus becoming the most critical ultimate or fatigue land-
ing gear design loads.
Finally it must be explicitly mentioned that this paper does not concentrate on the
task to reduce landing gear loads at touchdown by making use of the "almost unlimited"
possibilii ,s -eu by acLive control systems.
The basis of the work described in this paper ha.;been elaborated by the author at
the DLR-Institute of Aeroelasticity in Gottingen (Germany) as well as at the Air Force
Flight Dynamics Laboratory at Wright-Patterson Air Force Base in Dayton (Ohio) and has
already been partly published in Refs. [8,9,101.
This chapter is intended to give the reader a clear understanding of the formation
of the dynamic response of aircraft taxiing on rough runway surfaces. It is not the in-
tention of the author to present here quantitative results from (complicated) dynamic
response analyses performed in the time domain on a nonlinear multi degree-of-freedom
rigid body/flexible aircraft structural system running over a runway surface with a
random shaped roughness profile. Detailed investigations with regard to the handling of
this problem have been described in (11 and (9]. But, as was shown in [11), quite simple
analytical considerations carried out on a linear one degree-of-freedom oscillator can
shed light on the special dynamic response problem we have to deal with here.
Thus, according to this simple mathematical model, the basic mathematical considera-
tions denoted below are based on the following assumptions:
d) The stiffness characteristic of the tire is neglected in the calculations. This sim-
plification is tolerable in so far that we are primarily focussing here on the low
frequency range related to the aircraft rigid body moti-n where the dynamic struc-
tural behavior of the tires is of minor importance.
In the time domain the equation of motion of the oscillator can be formulated as
follows:
In Equation (1), a is the oscillator mass, d the viscous damping and c the stiffness
ii"
20-3
of the landing gear strut; h(t), fi(t),h(t) designate the vertical displacement of the
oscillator mass, its velocity or acceleration, respectively; z(t), i(t) characterize the
runway roughness and its derivative with respect to time t. As shown in L11, this equa-
tion can be used as a basis for calculating the (acceleration) response of the oscilla-
tor when encountering successive discrete runway disturbances. The most interesting re-
sult from these investigations consists in the definition of a factor, the so-called
"bump multiplier" MA,
2 which is defined as follows:
(2= I. 2E CA
(2) MA V1 EA 2 A
with
(3) EA e
(4) CA W*sV
)SW
(5) w* w
A being the bump spacing according to Figure 1, V the aircraft taxi speed, < the damping
ratio (relative to the critical damping value Cc-l) and w0 the natural circular eigen-
frequency of the oscillator system. The bump multiplier defines the extent to which a
second discrete disturbance amplifies or attenuates the maximum dynamic response of the
oscillator to the first disturbance. Equation (2) is valid under the assumption that the
two successive disturbances are of a similar shape.
Experimental investigations (2,121 have shown that the definition of a bump multi-
plier is significant in that a conventional landing gear, primarily designed to absorb
the aircraft kinetic energy at touchdown, can pass discrete single obstacles with a sig-
nificant roughness amplitude at all taxi Speeds. But the tests also revealed that the
aircraft, due to bottoming of the landing gear struts, could not traverse a series of
successive discrete obstacles at well-defined "critical" speeds, even in the case that
the roughness level was rather small.
(6) A = 2k - (k = 1,2,3.... )
V
(7) A = (2k- 1)T - , (k = 1,2,3....)
0
II! .
20-4
G.i
L-_0 are characterizel by the fa-t that the
LAN
G£A N G =04 oscillator hits the second obstacle when
its upward speed h(t) reaches its -mxi-
Application of the results obtained from the example of the one degree-of-freedom
oscillator to the example of a complete aircraft structure, which generalized equations
of motion are denoted in [9], now leads to the following statements:
1. a high dynamic response in its (rigid body) degrees-of-freedom must be expected when
the runway roughness exhibits major obstacles which - in connection with the eigen-
frequencies of the various eigenmodes and a range of defined taxi speeds - are lo-
cated at critical spacings A as defined iy Equation (6),
2. the dynamic response in the various eigenmodes can be lessened by a higher damping
of the undercarriage.
Operating aircraft on unprepared runway surfaces entails that we do not have any influ-
ence on the distribution (e.g. critical or non-critical spacing) of major obstacles.
Thus, according to the definition of the bump multiplier, the only parameter we can in-
fluence to improve the operational capabilities of aircraft on rough runway surfaces is
the damping of the undercarriage.
20-5
can reach extreme values in the case of a 'too) large damping factor d. This also ap-
plies to the case when the aircraft, operating in its taxi mode, hits a high amplitude
obstacle which has a sharp shoulder, hence inducing a high roughness velocity z(t) (see
Eq.()). Due to this fact the damping of landing gear systems must be kept small at
least in the compression phase of the shock absorber. To provide, despite this require-
ment, a landing gear with good energy dissipation characteristics, the practical design
of an undercarriage is achieved so that its damping characteristics vary during the com-
pression and recoil phases. Real landing gear systems feature a ratio of about 20 to 30
between their recoil and compression damping factors dRECOIL and dCOMP.
BeaLl a in mind the different curves for the bump multiplier, depicted in Figure 2,
it is obvious tha, L.. small value of d is highly adverse to the dynamic response
behavior of aircraft during t-h runway opekdtions. In order to realize both good
touchdown and rough runway perfomance of aircraft undercarriage, a landing gear system
must be designed which features low damping characteristics in the compression phase
with regard to the touchdown impact and the encounter of single high amplitude tsno.(
wave) obstacles with a sharp shoulder, and, in accordance with the bump mulitpliar, high
damping characteristics with regard to the aircraft's taxi capabilities over multiple
(long wave) obstacles.
The next chapter will show the extent to which both of these requirements can be
fulfilled by the implementation of active-controlled systems into the undercarriage.
--
DAMPING VALVE
I.ETUR PIP
- --
ES
OIL PRESSURE/
RTR IE
Operation of the landing gear in a closed-loop control system allows a very selec-
tive damping control to be achieved. In order not to increase the transient (short wae)
landing gear loads at touchdown and upon encountering sharp-edged, high-amplitude ob-
stacles, which are characterized by a broad band amplitude spectrum, a low-pass filter-
ing element is included in the compensation network. This precaution guarantees that the
dynamic response of the aircraft in the higher frequency range is not fed back to the
servovalve and thus not affected by the active control system. To reduce the maximum
_ls of th, bump multiplier, extensive damping must be achieved in the (long wave)
frequency band adjacent to the rigid body heave frequency of the one degree-of-freedom
oscillator. By the implementation of a band-pass filtering element into the compensation
network, a pronounced control can be realized in this frequency range. Moreover the
characteristics of this element allow any long term (sensor) zeroing errors in the con-
trol loop to be suppressed.
Lip to this point the active control system for alleviation of the landing gear loads
has been dpscribed. The operation of the active shock absorber, however, requires the
implementation uf two additional loops with regard to the static control of the landing
gear working point. This reference point is characterized by the following two parame-
ters: the reference elongation of and the reference pressure in the strut. These refer-
ence values refer to the static landing gear state with the (oscillator) mass resting on
top of the strut. Typical values for real >'nding gear are in the range of 50 bar for
the pressure and 0,2 m for the elongation, if the fully collapsed strut position is !on-
sidered to be the state of zero-elongation.
The two loops for achievement of the static landing gear control are depicted in
<
Figure 5. A low-pass filtering element with a very low cut-off frequency (f. 0.5 F7)
is included in both of the loops. This element is required to avoid any feedback from
the static control loops in the higher frequency range, which is intolerable since it
would cause an enormous stiffening of the strut (as in the case of servocontrolled ac-
tuator systems (13]). Furthermore it is of primary importance to provide the strut po-
sitioning control loop with a priority on th, pressure feedback loop. This entails that,
during the take-off-phase, the landing gear will remain in a position adjacent to its
reference elongation, even in the case where the landing gear loading is significantly
reduced by the aircraft lifting forces.
Having explained the basic functioning of an active-controlled landing gear by means
cf the eyimnp of the one degree-of-freedom oscillator shown in Figure 5, we can oo.
concentrate on the more complicated system consisting of the entire aircraft resting on
its undercarriage. Figure 6 shows the way in which the overall rigid body dynamic re-
sponse behavior of aircraft operating on rough surfaces can be controlled and thus im-
proved by a feedback of the aircraft vertical, pitch and roll motions or of the corre-
sponding velocities or accelerations. Aircraft pitch control is achieved by a feedback
of the pitch motion to the active-controlled nose landing gear. The rigid body heave
control is realized by an in-phase activation of the active-controlled main landing gear
struts, using the aircraft CG heave motion as a feedback control signal. Roll control is
achieved by a feedback of the aircraft roll motion, the signal of which, fed through the
compensation network, activates the main landing gear struts by phase-opposed signals.
As in the case of the one degree-of-freedom system, the layout of the entire compensa-
tion network can be considered satisfactory, when the additional dynami. forces produced
by the various feedback control loops in the struts of the undercarriage are in phase
with the velocities of the respective aircraft motions they ire intendeo to control.
INERTIAL GUIDANCE //
CG HEAVE MOTION
The feedback loops, as depict, _n Figure 6, were drawn based on the assumption that
no major coupling exists between the rigid body heave and pitch modes. This applies to
aircraft loading configurations with a CG location "just" in front of the main undercar-
riage. In the case that this criterion is not fulfilled it would be beneficial to the
layout of the control system that the resulting structural coupling effects be accounted
for by an additional feedback of the pitch motion to the main undercarriage and of the
heave motion to the nose landing gear.
5. Analytical Investigations
h X.
With regard to the bump multiplier, the results indicated a reduction in the range
of 41% for the aircraft pitch motion and of 37% for the aircraft heave motion. A further
result is indicated in Figure 8, which depicts the dynamic response of the aircraft
pitch and heave motions to a harmonic vertical excitation at its nose and main gear
wheels, respectively. The plated curves clearly indicate the hgher damping of the un-
dercarriage in the active-controlled mode, which considerably reduces the peak response
in the eigenfrequency neighborhood of the aircraft pitch and heave degrees of freedom.
These very satisfactory results were obtained at rather small flow rates of the hy-
draulic oil supply. To actively control the entire undercarriage, only a flow rate in
the range of 40 I/min was required. It has to be mentioned, however, that the oil supply
flow rate is quite dependent on the basic (conventional) design of the undercarriage
considered.
6. Experimental Investigations
The results obtained from the analytical investigations, some of which were pre-
sented in the foregoing chapter, were so promising as to encourage further research in
this area. The decision was made to construct a laboratory test setup for simulation of
the dynamic (rigid body) behavior of aircraft when operating on rough surfaces. To de-
monstrate the efficiency of active-controlled landing gear, tests should be run in a
conventional (passive) as well as in an active-controlled landing gear configuration.
The following sections give a description of the test setup and of the results obtained
with it.
20-9
70
n len6n
_n .... ... - - anti lose and
p I ao Main leanding gs-
0.2
~ N~ 04
1.2 1.2
8 h 8
landingger
Ai ga 00
I kI Ro-1
.amlm
0.9 a
555L t
-eI OU 21 2
-I.71
Nose landirg gear control loop Main landing gear control loop
The aim of the experimental investigations was to demonstrate the improvement in the
rough field operational capabilities of aircraft equipped with an active-controlled in-
stead of conventional undercarriage. Due to the smaller size of nose landing gear, it
was decided to perform all experimental investigations on an actual sized passive and
active nose landing gear system. The test setup is shown in Figures 10 and 11. A beam-
shaped structure, idealizing the front part of an aircraft fuselage, is pitching around
a pivot point, considered as the (fixed) attachment point of the main undercarriage. On
the other end of the beam structure, a modified hydraulic cylinder acts as a nose land-
ing gear shock absorber. A solid rubber element, simulating the nose wheel tire stiff-
ness is fixed to the free end of the cylinder piston rod. The "tire" is placed on top of
a platform connected to the piston rod of a hydraulic shaker. This system allows arbi-
trary rough runway profiles to be simulated by feeding defined electric set value sig-
nals to the shaker control system. For the simulation of (different) aircraft inertia
properties, a set of masses is fixed to the beam-shaped (fuselage) structure. All hy-
draulic elements, such as damping and back-pressure valves, which are integrated inside
the strut of the real landing gear, are mounted here outside the cylinder on a steel
plate and are connected to its upper and lower oil chambers by large-size hydraulic
steel tubes. The orifice cross sections of the recoil and compression damping factors
are variable, thus allowing the recoil and compression damping factors to be independ-
ently adjusted to defined values. The compressed nitrogen volume, forming the stiffness
of the shock absorber, is realized by means of a commercially available high-pressure
gas reservoir with a well-defined volume.
'TRE' HYDRAULIC
As shown in Figure 11, two sensors, an extensiometer for measuring the positon of
the landing ge-r piston in the strut and a gauge to determine the pressure in the upper
oil chamber are used for the static control of the landing gear working point. As a
feedback to the dynamic load alleviation control system various sensor signals, detect-
ing either the pitch rate or the pitch acceleration, are available.
20-I
- RUBBER D-APHRAGMAIRCRAFTPITCNTACCELERATION
STATIC ELECTROHYDRAUDC
SERVOVALVE
To simulate various supply oil flow rates, a manually adjustable flow rate limiting
valve is implemented into the pressure oil supply pipe, allowing the flow rate to be
varied between 0 and 25 1/mmn. Furthermore a mechanical spring system is used to simu-
late aircraft lifting forces during high-speed taxi phases. This simple mechanism allows
the landing gear behavior to be determined in the case of a quasi-steady change of its
loading forces, e.g. during take off. Finally it has to be mentioned that the test setup
is designed to allow determination of the characteristics of conventional passive land-
ing gear when shutting off the oil pressure and return pipes.
The defining data of the investigated test setup configuration are as follows:
mass
F .ad
of 1000 kg on top of the landing gear,
static landing gear reference point characterized by a strut elon-ion ot 0.2 m and
a pressure of 50 bar.
To detect the influence of the supply oil flow rate on the performance of the dynam-
ic load alleviation control system, tests were performed at flow rates of 15 and 25 /
min.
A first test consisted of determining the frequency response of the landing gear
transfer function relating the strut pressure in the upper oil chamber to a (harmonic)
electric set point signal fed to the electrohydraulic servovalve. The test was performed
in a locked strut configuration, thus not allowing any displacement between the shock
20-12
absorber cylinder and its piston. The results revealed that, in the entire frequency
band of interest ranging from 0 to 5 Hz, the pressure signal is delayed by a phase angle
of 90" relative to the servovalve input signal. The reason for this delay is that an ap-
preciable volume of hydraulic oil has to be inserted into the upper oil chamber of the
landing gear in order to compress the nitrogen enclosed in the reservoir, before a sig-
nificant pressure is built up. It is of primary importance to realize and understand
this "effect" which is in contrast to the dynamic behavior of servocontrolled actuator
systems, the (differential) pressure/servovalve input frequeny response of which exhib-
its smaller phase lag characteristics in the low frequency range [13].
It is obvious that the 90" phase lag angle between the shock absorber pressure and
the servovalve input signal must be taken into account when designing the compensation
network of the load alleviation control system. Thus, to produce dynamic damping forces
in the strut, it is required that the transfer function of this control system exhibits
a phase lead angle of 180" at the eigenfrequency of the rigid body mode to be contrciled
in the case of a displacement feedback or a 90" phase lead angle in the case of a veloc-
ity feedback or a 0" phase angle in the case of an acceleration feedback of the rigid
body motion in question. This entails that, if we consider a displacement or velocity
feedback, we have to include correspondingly two or one (first-order) differentiating
elements into the compensation network. The simplest compensation network, as depicted
in Figure 10, can be realized by means of an acceleration feedback of the aircraft rigid
body motion. During the tests the three different feedback types described above were
investigated. In combination with well-adapted compensation networks their overall per-
formance could be considered equivalent.
Final testing of the landing gear system was related to the simulation of aircraft
taxi over a series of AM-2 mat repair profiles. A typical sequence of investigated ob-
stacles, consisting out of 3 successive AM-2 mats, is depicted in Figure 16. To deter-
mine the aircraft landing gear dynamic response in the entire range of reduced fre-
quencies of interest with regard to the first maximum of the oump multiplier at w, - 2n
(Figure 2), the simulated taxi speed was incrementally varied between 10 and 80 m/s.
The time plots obtained in the case of the worst condition (most critical taxi speed
for the given obstacle spacing) are depicted in Figure 17. The plots indicate a signifi-
cant allevation of the landing gear loads when the control system is activated. Another
interesting result is obtained when drawing a plot of the maximum landing gear load, en-
countered at any location during the full traverse of a sequence of discrete obstacles,
- 20-14
Fo G I IJ
IN-
.. F lrTlfr1I
ELOSECATION fR 1 [UII
INGLE
RA-
0H0ER 0 0s
rSURE
-, 4* A
I0 Ic 103't
.
sGNAt I......
40 Is 40 15 40
as a function of the (constant) taxi speed. The experimentally determined results in the
case of a triple AN-2 mat profile with a height of 0.075 m are depicted in Figure 18. It
can be seen that the peak load level is reduced by 331 or 50% for the active-controlled
landing gear in the case of an oil supply flow rate of 15 or 25 I/min. Figure 18 depicts
moreover the maximum load levels for the testing cases
- triple AM-2 mat profile with a height of 0.10 m,
- double AN-2 mat profile with a height of 0.075 m.
20-15
17-
r/ -;
F-
0 //
1 2 fjec ix_--
...F _\ If\'~~...J
PASSIVELG ACTl I
1/,-. ACTIVELG 25 1/rmn
It is interesting to note that the loads produced in the active strut are only iden-
tical at very high taxi speeds to the loads in a conventional undercarriage. This fact
indicates that the active-controlled landing gear system, as presented here, neither al-
leviates nor increases the loads induced by short-wavelength obstacles or at touchdown.
With regard to the discrete bump testing it has to be mentioned that the simulation of a
taxi run over 30 (1) successive AM-2 mat profiles, with a height of 0.075 m and placed
at worst spacing conditions, could be achieved without collapsing the actively damped
landing gear.
All of the results in the three subsections above clearly indicate that a consider-
able reduction in the landing gear (peak) loads and consequently in the aircraft struc-
tural loading during rough runway operations can be obtained by the implementation of
active control systems in the undercarriage.
First, the reference pressure couid be easily adjusted in accordance with a function
of the aircraft inertia parameters. This would allow the undercarriage to be adapted to
any aircraft loading configuration and would thus improve its overall landing, take-off
and taxi performance by modifying its pneumatic spring characteristics.
- 10- 16
20
kN
F
10
PA
SSIVELAWOAG WEAR
t6 -151/mW7
___ACTIVE
ACTIVE
£1W-251/mm
0' Y0 40 50 r/S 8
25-
20-- -32%
/7 '32
5 PASSIVE
LANDING GAR
ACTIVE
16 - 151/mn
ACTIVE
L6 -251/min
0 0 40 6o MA 6"
0
10/
5 PASSIVE
LANDING GEAR
ACTIVE
L6-151/min
- ACTIVE
16 -251/m*n
0 20 49 50 r/-S 60
Second, the lift produced on the aircraft lifting surfaces, when operating in its
high speed take-cff phase, could be significantly increased by elongating the nose land-
ing gear strut which entails a higher angle of attack of the entire aircraft. Such a
system would require the reference elongation value setting (of the nose landing gear)
to be changed in a transient form at a well-defined taxi speed during the take-off
phase. Rough estimations have indicated that, due to this type of control, the take-off
distance for fighter aircraft with a high power-to-weight ratio could be reduced by
about 100 m, which is in the range of 25% of the normal take-off distance.
Moreover the simple principle of active damping, as outlined here in detail for the
case of tle aircraft rigid body modes, could certainly be applied to achieve damping
control of defined "critical" aircraft flexible modes. A typical example could be the
contrci of the wing bending and/or torsional eigenmodes or of the fuselage vertical
bending mode in order t- prevent the structural damage of wing pylon, outer wing or
ftselage front structures when operating the aircraft on damaged runways (161. In this
;ase it would be beneficial to use several accelerometers, distributed on the wing tips
and/or at the fuselage nose, to generate convenient feedback signals to the landing gear
system control loops [17].
These reflections clearly indicate at least some of the many advantages to be attri-
buted to an active-controlled landing gear design. The question as to the extent to
which the various systems can and should be realized addresses above all problems re-
lated to the reliabiiity an'- .., extent of damage in the case of a failure inherent to
the various active systems under consideration.
8. Conclusion
The design of an active-controlled landing gear system was described. The layout of
the control system was designed to provide tha active undercarriage with improved damp-
ing characteristics in the frequency range related to the aircraft rigid body degrees-
of-freedom. This entails the alleviation of the landing gear and aircraft structural
loads due to long wave obstacles without influencing the dynamic response, in comparison
to a conventional landing gear design, to short wave impacts occurring at touchdown or
upon encountering sharp high-amplitude obstacles.
Moreover a detailed description of a test setup for the simulation of aircraft oper-
ation on rough runway surfaces was given. It was experimentally demonstrated that an ac-
tive-controlled undercarriage can significantly reduce the landing gear and consequently
the aircraft structural loads during aircraft taxi on unprepared rough runway surfaces.
For instance, the RMS-value of the dynamic landing gear forces could be lowered by 42%,
in comparison to passive landing gear, in the case of aircraft operation on surface,
with a randomly shaped roughness profile. Simulation of taxi tests over a runway profile
with three successive discrete obstacles (AM-2 mat profiles) yielded a reduction of 50%
in the peak landing gear loads in the case of worst spacing conditions.
The aim of this paper was to realistically demonstrate, in a laboratory test, the
possibilities offered by the implementation of active control systems to aircraft under-
carriage and to give a quantitative estimate of the improvements obtainable with such
systems. Further work will Stili be requsreo tor tne achievement of a pLaciial dei.1
of a real active-controlled undercarriage and its control system, especially with regard
to failure and reliability questions.
- 20-IS
9. References
[ GERARDI, A.G.
Status of Computer Simulations of USAF Aircraft and Alternative Simulation
Technique
AGARD-CP-326, Ref. 11, 1982
21 JOHNSON, W.P.
Comparison for the A-7D Aircraft Dynamic Response Using Experimental and Analyti-
cal Methods
AFWAL-TR-86-3084 (1986)
[ 31 Airplane Strength and Rigidity, Landplane Landing and Ground Handling Loads
MIL-A-8862 (ASG), May 1960
41 Airplane Strength and Rigidity, Ground Loads for Navy Procured Airplanes
MIL-A-8863, July 1974
1 91 FREYMANN, R.
An Experimental-Analytical Routine for the Dynamic Qualification of Aircraft
Operating on Rough Runway Surfaces
AGARD-R-739, Ref. 2, 1985
[101 FREYMANN, R.
An Active Control Landing Gear for the Alleviation of Aircraft Taxi Ground Loads
Zeitschrift fir Flugwissenschaften und Weltraumforschung (ZFW), Band 11, Heft 2
(1987), pp. 97-105
(121 FREYMANN, R.
Frequency Response Tests Performed on a YF-16 Aircraft Using the AGILE Test Faci-
lity
AFWAL-TM 14-')05 FIBES (1984)
20-19
(13] FREYNANN, R.
monlinear Ae~oelaSti Analyses Taking into Account Active Control Systems
AGARD-R-698, Ref.2, 1981
[14] FREYKANN, R.
Linearization Techniques
AFWAL-TM-84-229-FiBEB (1984)
(171 FSEYKANN, R.
Interactions Between an Aircraft Structure and Active Control Systems
Journal of Guidance, Control and Dynamics, Vol 10, No.5, pp. 447-452 (1987)
22-1
by
E.F. Wild (Mrs)
B.R. Morris
SAC Technology Ltd
Marlborough House
Churchfield Road
Walton-on-Thames
Surrey
KT12 2TJ
United Kingdom
A.E. Dudman
Assistant Chief Aerodynamicist (Dynamics Engineering)
Design Office No.8
British Aerospace Plc (Civil AirrafL Division)
Filton
Bristol
S99 7AR
United Kingdom
SEMARY
The application of standardised runway repair obstacles (SRO) for design purposes and
post-design capability determinations was a basic outcome of SMP Working Group e2
deliberations.
The WG 22 standardised shapes have been the basis of clearance work for a current
military aircraft. To verify the adequacy of the SRO profile for this work the response
to real repair profiles, obtained from practice repairs, has been related to the
response to the SRO. Some comments on the SRO as a representation of real repairs and
on some undesirable features of real repair profiles are presented.
1. INTRODUCTION
The clearance information, as requested by the UK MOD, will be provided in the for,,,of
contour plots. These plots define safe speeds and spacings for crossing double repairs
and are based on the calculations of the aircraft response to a specific height of SRO.
The objectives of this document are firstly to consider the adequacy of the SRO profile
shape as a representation of real repairs with respect to the aircraft response, and
hence the accuracy of the contour clearance procedure in the real repair situation.
Secondly to consider the Equivalent Obstacle concept as a useful measure of the
severity of the real repairs compared to the SRO.
To illustrate the discussion the response calculations of the heavy weight takeoff
configuration of the military aircraft are presented. This case models the aircraft
accelerating for takeoff from a specified initial speed. The BAe mathematical model of
this configuration has been validated against test data.
To irtroduce this discussion the aircraft response to single and double SRO in
time-history and contour plot format is outlined first. This is then followed by a
22-2
detailed comparison between the aircraft response to three real repairs and the SRO
response. A discussion on the adequacy of SRO contour plots to cover the real repair
variability is then presented.
The Equivalent Obstacle concept, and its inherent assumptions, are outlined as a way
of overcoming the infinite variability of the real repair profiles, and hence aircraft
response, and to provide a basis for determining the required magnitude of the SRO for
the contour plotting.
2. REPAIR OBSTACLES
A7EAAAEA
Re[l 0RAACLE REALRAPESR
p200121
COEREDBYCL60REPAIR
UT OVEE ByxM
AREPAR "AT
VTi'.RE1. A*XII
SR04ii22 00T,
OA.A1-0 h 17"0
FIG-- ka
Figure Figure 2
The 70mm profile is illustrated in Figure Ia. The maximum height of 70mm, 521,,m or 38mm
is achieved in 1.25m. The trailing edge ha,& the same slope a% the repair leading edge
0110 -,L Length, including the two ramps is zz.zm.
The profiles of the CL60 and AM2 repair mats are illustrated in Figures It, and lc
respectively. The CL60 mat has thickness 32mm and a ramp length 0.25m and the AP12 is of
thickness 38MM with a ramp length of 1.375m. For thiI mndelI-- *he total length of the
mats including the two ramps is 22m.
The repair mat, including the end ramps follows the surface of the repair profile
exactly. Direct additin, the two heights gives the full repair profile.
Figure 2 illustrates the three repairs covered by the CL60 mat and beside for
comparison the repairs covered by the AM2 mat. For the BAe modelling the repair mat was
fitted exactly over the crater in-fill and therefore the rise onto the repair is
distorted by the underlying fill.
The response to the SRO of height 70mm and length 22.5m is shown for the nose and main
ge-s on Figures 3-4 for initial speeds of 20,40,50,h0,70m/s at nose gear encounter
with the obstacle leading edge.
Takeo'f Takeof
4SE ESPOEOSE
TIP ii53O Asix o r TxirOUxSPEEDANGE
C xESxCNSE
-- - --- -- - -- - -- - -- - - - - - -- - - --
V1-----------------
Figure 3 Figure 4
Considering the nose gear first it is seen that for the lowest speed illustrated the
aircraft is sufficiently slow for the second peak to occur towards the repair trailing
edge and this is then the highest peak. For speeds above this, the first post-mat peak
is the maximum and remains so up to hOm/s while for speeds of 70m/s and higher the
maximum occurs during the rise onto the repair. Tuning in pitch over this repair occurs
at approximately 40m/s as indicated in Figure 5 where the maximum Percentage Available
Increment, PAl which occurs throughout each time history is plotted against speed, tor
the three SR0 heights 7Omm,52mm and 38mm. The lowest responses occur either side of the
peak at speeds of 30 and 60m/s.
The main gear response maxima occur during the rise onto the repair for every speed,
although the f-rst post-mat peak is not insignificant for any speed up to Om/s. These
maximum locds uo drop slightly as spe&d, ;-d therefore aerodynamic relief, increases.
Figure 6 illustrates this, showing maximum PAT against initial repair encounter speed.
22-4
IW
.5:--E '0'2? -
Figure 5 Figure 6
Response work hss been carried out to determine the aircraft capability over double
SRO. This capability over dcubles is oresented in the form of contour plots of PA1
against nose ws eel encounter speed at the leading edge of the first repair and repair
spacing. Repa)- spacing in this paper, to be consistent with some others, denotes the
distance from the trailing edge of the first repair to the leading edge of the second,
ie. the cap between the repairs rather than the more appropriate distance for
tuning of leading edge to leading edge. The maximum PAl at each speed/spacing
combination or this grid is c.oIculated and the 80,90 and 100% PAl contours mapped. This
method of defining the clearance gives an indication of the sensitivity of the responsc
at the criti-al speeds and spacings and makes use of the phasing off the repair to
define the maximum allowable clearance.
PAI COPATGIIE
PLOT
REPAIR SPACING EA.0NOUNtR2 PP
A'VAYWANE-OFF
22 1. SE. NOOSFOPAR
Fi.1
Figure 7
22-5
Figures 7-8 show the contour plots calculated for this configuration over the 22.5m SRO
for the nose and main gears respectively. The nose gear shows one critical area over
Lhe range of tuning speeds 40-50m/s; this lobe being formed from loads caused at
encounter with the leading edge of the second repair when placed in the region of the
first post-mat peak from the first repair.
6 5- LL I L .
55-
4 5-
2 5
7
ruI A I - A -I .
0 5- -x--
0 O -0I I L I
1 0 20 30 4 0 5 60 70 B 0
ENCOUNTERSPEED
Figure 8
The main gear contours show critical zones when the second repair is placed in the
region of the first post mat peak except at speeds close to 30m/s. In addition, areas
of some significance, 80%PAI, occur due to the combination of the second post-mat peak
and the leading edge of the second repair but only where the second peak after the
first repair is sufficiently large.
g ePLe
Fig-r 9 Figure 10
22-6
Figure 11
If the contour approach using SR0 is to be used as a guide to the aircraft clearance
over double repairs then, when translated to the real repair situation, the SRO must
show a response similar in both phasing and magnitude to the real repairs.
Similar phasing is essential to ensure that accuracy in spacing definition for leading
edge and post-mat loads is achieved. Magnitudes of the loads, at repair leading and
trailing edges, from response to the chosen height of SRG must be similar, or just
envelope, the loads from real repairs. Too high a level of SRO would result in an
unnecessarily restrictive clearance.
A comparison of the magnitudes and phasing between the chosen real repairs CB, C9 and
C2R, and the 70mm SRG, are given below.
Figures 12-13 show the response to CS the high, late-skewed repair at speeds 20,30,50
and 6dm/s. Figures 14-15 show the equivalent plots at leading edge ramp speeds of
20,40,50,60m/s, for C9, the repair with the central dip, and Figures 16-17, C2R the
fairly flat repair with the crater in-fill, which adds to the height and steepness of
the leading edge ramp.
The form of humped repair, illustrated by C8, is dominated by the fundamental component
of shape but, as it is skewed, the pitch response tunes at a lower speed than that due
to an SRG of equal length. Hence, for the nose gear, highest post-mat loads occur at
)Om/s over this repair, giving a PAT of only 2,4% less than the SRO, with a lag in the
peak load equivalent to 1.9m. At 5dm/s the first post-mat peak resulting from
response to C8 is tower than the SRO first post-mat pea by ld.7XPA. This gives, for
CS. an equivalent SRG height of only 44mm. The greatest mismatch in phasing occurs at
2dm/s with a lead equivalent to 4.3m, however the SAC response is .bPAI greater.
The form of dipped repair, represented by C9, tures is if the repair consists of a pair
of shorter repairs. Hence, for the nose gear, highest post -mat loads occur at 2 Gm's and
C9 shows exceedence over the SA0 of 2.9 PAl at 20ms with a la equixalent to 1.2m.
Through the rest of t' e speed range (greater than TOm/s' all post mat loads from C9 are
low compared with the SRO, for example at 50m/s C9 produues post-mat loads equivalent
to an SR of height 33mm.
The form of repair most similar to an SR is represented by C2C. The only deviation in
similarity to the SRO is a small section of crater in-fill which adds to the height and
steepness of the leading edge ramp. However, post-mat loads show no significant tuning
but otherwise the response to the SAO is similar, although of lower magnitude
22-7
throughout the speed range. Post mat loads give equivalent loads to a SRO of height
31mm at 50i/s.
Thus, at speeds at which the real repairs tune and the SR not, ie. 20m/s for repairs
with deep central dip, and 30m/s for skewed humped repairs exceedence of the 70mm SR
post-mat loads will be possible. At speeds where the SRO tunes and these repairs do
not, the 70mm SRO causes considerably higher loads.
Real repair loads at the second post-mat peak are lower than the SRO loads. Loads
closest to the SRO loads in magnitude occur in response to CB at the low speed end
20-30m/s with a PAI difference of only 3.8% at 2Om/s.
Due to the steepness of the CL60 repair mat leading edge ramp loads occurring during
the rise onto CL60 repairs are, in general, the maxima of the time history from speeds
as low as 20-30m/s regardless of the detail shape of the repair. Exceptions do occur,
for example C8 at the tuning speed of 5m/s, due to the relatively high drop off the
repair and low rise onto the repair so that the post-mat peak becomes greater.
Comparison with the equivalent SR0 leading edge loads shows magnitudes of the CL60
repairs at the lower speeds l0-30m/s, which are lower than the 70mm SRO, even for the
repair C2R. From speeds of greater than 30m/s the magnitude of these leading edge loads
is approximately equal to, and may exceed, those due to the 70mm SRO. This indicates
that a rise of nominally 32mm over 0.25m (CL60 ramp) gives rise to loads that can equal
a rise of 70mm over 1.25m (SR0 ramp). In the case of C2R, where crater fill augments
the rise these loads will exceed the SRO. An increase in PAZ of 5.1% is measured for
C2R at 50m/s, this being equivalent to an SFO of 79mm at the leading edge.
In addition, C2R has leading edge loads which are of equal magnitude to the SRO post-
mat loads at 60m/s.
The phasing of the nose gear leading edge loads over the CL60 repairs show a lead
equivalent to 1.3m due to the sharper rise and effect of crater-fill at this point.
Main gear loads resulting from crossing real repairs shows a similar form to the
response as for crossing the SF1. The highest loads throughout the time history occur
during the rise onto the repair.
Post-mat loads for real repairs may match and exceed those due to SF1 at tuning speeds.
CS shows exceedence of the first post-mat peak of the SRO at 30m/s by 2.7% PA!
(equivalent SF1 of 72mm) with a lead equivalent to 2.7m and the second post-mat peak by
2.9% PA0 (equivalent SF1 of 72mm) with a lead of 2.7m. For speeds greater than the
tuning speeds of the real repairs the SF1 post-mat loads exceed those due to the real
repairs, for example C6 at 5Om/s shows a PAl of 28.7% less than the SF1 and is
equivalent to an SRO0 of 43mm.
The leading edge loads for real repairs are always less than the SRO loads and are
equivalent to an SF1 of 48mm for C2R at 50m/sl this considered to be the worst repair
of the three for leading edge response.
Phasing of real repairs compared with SF1 at the leading edge shows a lead equivalent
to one metre.
To illustrate the detrimental effect of the CLA0 leading edge C2 was covered by a CL60
repair mat (32mm thick) with a leadirg edge vamp _t 1.25m, as of the SRO, placed clear
nf t- ater in-fill, and ':,e r-sonse calculated at 5Om's. Figures 16-19 illustrate
this for the nose and main gear loads respectively. The response is compared with
the response to a CL60 mat only and to C2R covered by the nominal CL60 mat (ramp
modified by crater in-fill). At this speed the CLAP mat alone gives nose gear leading
edge loads equal to the 70mm SRO and therefore C2R must exceed. The modified CL60 ramp
shows loads far lower, implying that a far less restrmctive clearance level would be
possible by:
2. Modifying the CL60 leading edge to be less severe and more similar to the SRO
ramp.
22-8
-oNMI 00 r IN NN 9 I 00D
Figu.re 12
Foqu-r 13
22-9
NN90 0 03 IN I4I Ip I
Fig., 14
F 1ur.
i1
10 10
22-10)
Oh 0
0000 0 0 00 0 00 000 00 00 00 0
000 000)0003 IN NM0 0001003 'N
Figure 16
0 0
Figure 17
22-il
N s - u ,
NN -PTU INI-
NNNd
22-12
The response for both nose and main gears are shown in Figures 20-21 for C8 covered by
the AM2 mat, Figures 22-23 for C9 and Figures 24-25 for C2R.
Comments on the tuning characteristics stated for the CL60 repairs apply to the AM2
repairs. At speeds at which the real repairs tune.and the SRO not, exceedence of the
70mm SRO post-mat loads will be possible. CB shows a slight exceedence at 30m/s of
0.IPAI with a lag equivalent to 1.9m and C9 at 20m/s shows an exceedence of 2.3%PAI
with a lag of one metre and giving an equivalent SRO height at this point of 75mm . At
speeds where the SRO tunes and these repairs do not, the 70mm SRO causes considerably
higher loads. C8 causes the greatest loads at these speeds which at 50m/s are
equivalent to an SRO of 49.6mm.
Phasing off the repairs is variable, the maximum difference for these -epairs being for
4
C8 at 20m/s showing a lead equivalent to .1m.
The leading edge ramp of the AM2 repair mat is less steep than the SRO, namely, 38mm
rise in 1.375m. Leading edge loads are far less severe accordingly. For example, at the
speed of 50m/s, C2R only reaches an equivalent SRO at this point of 45mm.
The phasing of the nose gear leading edge loads over the AM2 repairs show no
significant difference.
I IN
RI
00 H O09
00 00 0 1 00 09 00 0
.... :v NN, U3 it
Figure 20
F igure 21
S __
I- - -- - - I
- - -- -
I-
00 00. 0 S? 000
IM P-1 P' I 7IN NNV 21ooul IN
22-14
I N I I-P3 - p3 I
I0
11V - 11,1 IN r,
L(igure 22
22-15
I
NN 3 I N.-. - I N I
2, I2
NN 3 I N . . . I-[
Figure 25
22-16
The peak-to-peak comparison of the SRO and real repair responses certainly emphasises
the difficulty in trying to select a profile shape which gives a response for both nose
and main gear that is similar to, and envelopes, the responses to real repairs. The
infinite variaoility of real repair shapes, even though constrained within a
repair-build specification, can lead to a range of responses that cannot be matched
accurately by one shape alone.
Two major problems are evident. The first is the effect of the crater in-fill to the
repair shape. Whatever the resulting shape of the crater in-fill, be it humped or
dipped, the overall effect on the aircraft response is to cause the tuning speeds over
that repair to be lower than that over the Standard Repair Obstacle because real
repairs will always be effectively shorter than SRO. Tuning occurs strongly at 50m/s
over the 22.5m SRO profile and not at the 20-30m/s as for 22m real repairs. Also, due
to ter differing effective lengths post-mat phasing, especiall at the low speed end,
will vary. A modification to the flat SRO would be indeterminable, but an adjustment to
the length to ensure tuning occurs at speeds more relevant to the real repairs would
alleviate the problem.
The second problem lies with the use of repair mats which have leading edge ramps which
differ significantly from that of the SRO. The CL60 mat with a rise of 32mm in 0.25m
causes loads at the leading edge ramp that are greater than those occurring during the
rise onto the SRO at 5Om/s. So, in the case of the crater in-fill of a real repair
auqmenting the repair mat leading edge, as with C2R, the SRO leading edge loads can
only be exceeded further. SRO leading edge loads match more successfully to AM2 covered
repairs as the ramps are more similar. As previously stated there is the possibility of
reducing this leading edge problem by firstly ensuring that the repair mat ramps are
set sufficiently a-?r fr- the -- a* - - ;Hi sow r.--- " :!z- not augmeni-wj-
ramp and secondly serious reconsideration of the CL60 ramp design.
The differences for the CL60 repairs are illustrated in Figures 26-27. An envelope of
the maximum PAI resulting from the aircraft response to the full set of CLA0 covered
repairs analysed by BAe is overplotted against the SRC PAI "axima (reference Figures 5-
6). The aircraft response to the real repairs can create nose gear loads which are
equal to those resulting from crossing the 70mm SP). The e-eedence of the SRO response
at 60m/s actually results from the response to C2R. This is oue to the aircraft
response at the leading edge exceeding the resp-nse t, the SRO at both the leading edge
ramp and post-mat. The main gear shows a simila, .reno for both the real repairs and
the SRO because the response maxima for both geneoally occur cring the rise onto the
repair.
Figure 26 Fiqure 27
Figures 28-29 show the equivalent plots for the AM2 covered repairs. AM2 repairs show
no problems at the 60m/s region and similar to the CL60 repairs at the 3cm/s region.
4
22-17
* L
Figure 28 Figure 29
On the basis of the comparison of the calculated responses to the three real repairs
and to the SRG it is evident that the SRO based contour plots are not ideal. The height
of the SRO to be used must evidently be a compromise.
At the low speed end, speeds of 30m/s and less, there are problems with both phasing
and magnitude. Post mat loads of' real repair- for both nose and main gear give reason
to use the 70mm level. Leading edge loads which are equivalent to SRO of less than 70mm
at these -peeds will give allowance for the inevitable occurrence of post-mat loads
equivalent to an SRO of greater than 70mm. An increase in the 90% contour to encnmpass
the 30m/s region, and use of the 90% contour for" speeds up to 35m/s. which although
cautious, may cover the problems caused by phasing in thin region.
For speeds higher than 30m/s the three real repair responses do not exceed post-mat
loads which are equivalent to 50mm SRO for both nose and main gear. Leading edge loads
for the nose gear over this speed range cause loads equivalent to an SRG of 45mm for
the nose and 48mm for the main if the repair is covered by an AM2 mat and 79mm for the
nose and 48mm for the main if covered by a CL6 mat. Thus, at these speeds, for a pair
of AM2 repairs the 70mm SRO provides unnecessary restrictions on the aircraft. If
contours for a 50mm SRO were to be plotted the lobe on the nose gear contour would
probably not be there and the main gear lobe would be greatly reduced over this speed
range. In addition, no account need be taken of the areas of significance defined for
the second post-mat peak since they become further below criticality.
If a pair of CL6d repairs is to be represented at these speeds the leading edge loads
-Doe .. t-L
(-atlon tor using the clearance based on the 70mm SRO for the nose gear.
However, due to the lower post repair loads some over-representation of the severity
may still exist.
The above arguments suggest the possibility that two levels of contour plots could be
used for the aircraft at this configuration in order to produce a clearance that is not
excessively restrictive. The first set would be at 70mm and would be provided to cover
speeds of up to 30m/s or possibly 4dm/s for some contingency for both nose and main
gear. Modification to the plots may be considered to cover the low speed end variances
in magnitude and phasing. The second set would be provided for both nose and main gear
at the 50mm SRO level to cover combinations of AM2 covered repairs at the high speed
end. If no modification to the CL60 repair leading edge is foreseeable then the higher
level of contour would have to be applied over the full speed range for the nose gear.
Also when considering interpretation of the contours, some allowance should be made for
the following aspects adding to the aircraft loads:
* RJnway roughness.
* Repair asymmetry.
22-IS
7.1. Geeral
BAe studied the response uf the military aircraft to a large r ,mbr of real repair
profiles in order 'o determine the aircraft s capability. Lan ing and takeoff
configu-ations were aalys-d. For this work a more simplisti approach had to be
adopted in orter to reduce the quantity of work that would have been involved in
prouucing a clearanne by a peak to peak analysis such as has been used
above to illustrate the detail differences between the responses. The Equivalent
Obstacle zoncept had been envisaged by the AGARD WG27 as a possible means of assessing
this carahility anr it was decided that this should be applied to the military aircraft
clearance.
The Eqiivalent Obstacle , ED of a real repair is defined to be the height an SRO has
to be in order to give the same maximum load as that resulting from the response to the
real -epa r. To determine this height the maximum loads frot real repairs are compared
'n -imum loads from the response to t'e SRO of heights 70. 52 and '8mm regardless of
the nocct or. or off the repair at which the macimums occurred. Figure 30 illustrates
.he calculation of EQ for the repair CB covered by the CL6D mat. In this case this
method qiver a 'easonabie estimate for the main near hut is misleading for the nose
gear. he nose gear ED results from a comparison of real repair leading edge loads and
SRO post-mat bods. Thus nose gear based Ed must be considered with care.
14
"'Ci:a he
-a ,ecair nschwwO tI he a Sihfih then this evaluation at [ qis
a 45! h .. i 'e icai'- s sever i~, i rvi1it.'*hr- reponses'- are similar. Of course.
* -c,iie el-ac. Ir itda ccc'I ctads are h~Ot rrJ~ihu within a certain level ,say
Sa -w e 'r irer rt -,ica. roi.-hnessl then he actual ale of Eo for a single
' -
a . F; ha.. a , i- ation :r the utiIle epaitr -itatin.
-i-cfLa it"
" mr~l
- es--cu -C rmaiuus hi.w tat n c-- at tray mIe- -eat s
-i "C x , i . ar,1 'ueme -s s ,ti, iectl ;,c 1- iI he phasing at the
t;a-; c --
;
The peak-to-peak matching of the responses of CB. C9 and C2R show that this is not
always the case and that exceedences of the 70mm SRO response do occur at certain
locations through the time history although the Ed calculated for the repairs as
singles at these speeds would be less than 70mm. However, it has been shor that,
although there may be post mat exceedence of the 70mm SRO response it is unlikely to be
followed by a leading edge response equivalent to 70mm SRO that indeed phases
sufficiently accurately Lu cause a transgression into the regior bounded by the d00%PAl
contour.
A better estimate of EO would be to consider the leading edge ramp loads and the post
mat load first and second peaks and calculate an EQ at each of these points throughout
the speed range and select the most pessimistic. Small phasing differences would be
ignored. however this would involve a great deal of work that would not be fully
justified considering the variances in the real-life situation.
8. CONCLUSIONS
The differcg effective lengths of the real repairs compared with the SRO causes
tuning at different speeds and hence magnitude and phasing of the post-mat
response is dissimilar.
The differing leading edge ramp on the CL6O repair mat, and to a lesser extent toe
AM2 repair mat, compared to that for the SRO causes dissimilar resnnns in botn
magnitude and phasing during the rise onto the repair.
However a useful change would be to shorten the SRO length to one that would
produce aircraft tunirng at more similar speeds to the real repairs. Repairs to
smaller craters could need to he matrhed to even shorter SRO.
In order to make the response to the real repairs as similar as possible to the
SRO it is advisable if possible to avoid both excessive sag at 'he repair rctre
and overfilling the repair so as to cause a large hump.
In addition, ensure that the repair mat ramps are fixed clear of the crter -ra,
to avoid detrimental addition to the rise onto the repair.
The current SRO profile, when set at a -eight that reasnably ercomass-s ih ,
repair loads tnroiughot the speed range will lead to a clearance de- nition which
will be adequate at some speeds and overl restrictive for others.
* The peak-to-peak analysis to obtain equivalent SRO at the ritical points nf foe
resporse over a repair would lead to a more accurate theoretical estimate of the
severity of the repairs compared with the E0 calculation. The excessive wor-k
demanded by this technique in order to achieve the additional accuracy is nt
worth the effort considering all the variability ir. the real-life stuatio,.
The Ecivalent dhstacle values gives an estimate of the severity of the repair and
when onsidered with care will be sufficiently accurate to provide an estimate for
the height of SRO for the JK clearance procedure.
Modification to the C060 leading edge ramp to one more similar to that or the SRO
would bciade' the allowable clearaire for this configuration br allowing a lower
le-e of SRO to be used for the contour plots over a substantial speed range.
9. REFERENCES
by
D.Chaumette*
Dassault-Aviation - France
Specialist Meeting Chairman
It became e%ident during the Meeting that landing gear are a prime concern because they are the number two system causing
major airframe failure. But in most cases the origin of failure is fatigue, often in combination with corrosion. In this respect
selection of proper alloys with efficient corrosion protection coatings is very important.
Purely static failures (by exceeding ultimate strength of landing gear) seem to be very seldom, excluding some "rogue" cases
beyond any design possibility.
Speaking about regulations, existing military regulations are US MIL Specs 8862A, 8863A and 87221 A. and the French AIR
2004 D and E. The MIL Specs are outdated, not well understood and rarely used. The French AIR 21)04 E is in use, but may
also need improvement in the landing loads section. Concerning statics design cases, there are some divergences between both
sets of regulations, this being particularly noteworthy for the landing cases of carrier airplanes
As mentioned above, fatigue of landing gear structure is not properly accounted for in landing gear design.
It was generally agreed that a design guide based primarily on statistically compiled operational data is needed by the industrsN.
In the field of computer code, several papers presented extremely impressive computations. including the response of the
complete structure, taking into account non linearities typit al of landing gears. What remains to be dose are more comparisons
1ith actual measured cascs Also problems may remain fir such effects as gas-oil thermodynamics or friction forces in lco
struts.
The question of the use of bomb-damaged runways was the subject of sceral papers. The tentative conclusions are the
followine:
- Landing gears of olet design were optimized onl, for the landing impact cases. Examples were given thai modifications
u, the ilei struts absorbers may,,provide lots of improvements for the use of damaged runsways. this still isi ig paNssc
S) stcms
- hi may be ti)use ac)'.:, sstems. There was no clear consensus on the use
relpirid t i n ,,'rcesms'rc requiremeP! a soloi
of aitis(: ,sstems, with several concept, ranging between "adaptatic damping" and cinpletely piloted cost rut,
OJuestions on issues such as complexity. reliability ansi oil flow rates arc not resolved. And the last point is that tip to nw"
c.re requirements concerning bomb-damaved runway operations were nscer introduced in new airplane specs. so such
sssteiti is Iut necessary
par
D.Chaumette*
Dassault-Aviation - France
President du Specialists Meeting
11a tt evident dans le Meeting que les atterrisseurs sent un souci important car its sont en deuxi~me place pour les systemnes
causant des ruptures majeures de structures d'avion.
Mais dlans IaIplupart des cas lorigine de [a rupture est IaI fatigue, souvent en combinaison avec IaIcorrosion. Sur cc point, Ia
selection d'alliages approprices et de rev~tements protecteurs efficaces sont tres importants.
Les ruptures purement statiques (c'est dire par d~passement de IaI charge extreme de l'atterrisseur) semblent &tre quasi-
inexistants, si Ion exciut lea cas tellement hors norme qu'il serait irrealiste de vouloir les couvrir dlans la conception.
Parlant dleriglements, les specifications militairca existantes sont leaUS MILS SPECS 8862A, 886 3A et 87221 A et les normes
franqaises AIR 2004 D et E. Les MIL SPECS sent d~mod~es, pas tris bien comprises et rarement utilisees, La AIR 2004 E est
utilise mais pourrait aussi necessiter des ameliorations dans sa partie charges d'atterrissage.
Enl ce qui concerne les cas de dimensionnement staliques ii ya queiques divergences entre les deux groupes de reglements, ceci
6tant particulierement net pour lea cas d'atterrissage sur porte-avions. R~cduire ou expliquer de telles diffurences pourrait 6tre
profitable aux pays de lI'rAN.
Comme indiqu6 plus haut. IaIfatigue des atterrisseurs nest pas suffisamment prise en compte bors de leur conception.
11a ei g~rniralement admis qu'un guide dc conception, base principalement sur une compilation statistique de r~sultats en
operation, est un besoin pour l'industrie.
lDans bedomaine des programmes de calcub plusicurs papters ont pfu~sente des eabouls tres impressionnants. incboanit III reponsc
de IaI structure complete et prenant en compte lea non lin~arit6s typiques des atte'-;-s urs, Ce qui reste a faire est plus de
comparaisons aver des mesures de cas nuls. Aussi des probl~mes peuvent subsister lies it des effets idas que lea
thermodynamiques gaz-huile ou lea forces de firiction dlans lea amortisseurs.
l.a question de loutilisation de pistes endommag~es par bombes aene 'objet de plusieurs papiers
7. Presented at the 71 st Meeting of the AGARD Structures and Materials Panel, held in Povoa
de Varzim, Portugal on 8th to 12th October 1990.
8. Author(s)/Editor(s) 9. Date
Various June 1991
14. Abstract
- This publication reports the papers presented to a Specialists' Meeting organised by the Structures
and Materials Panel and held at its Fall 1990 Meeting.
The Specialists' Meeting provided a forum for the exchange of experiences betwtcLn the NATO
nations with the aim of advancing landing gear design criteria and methods of landing gear
analysis. The Meeting reviewed existing design practices and specifications, considered the
various methods for load measurement and data analysis and formulated guidelines for future
design procedures.
oU
~ y
5 nu0t1
~ ~ ~ ~ U~ ~ -
C -C.~CJU .CCU
I= z czL
m rC
CC
a rj
i C
--- u~ C.0C
t, LO b, ,- ZO-
~. c.E
,xv SEx-
E 9b
z E
NATO OTAN
7 RUE ANCELLE -92200 NEUILLY-SUR-SEINE DIFFUSION DES PUBLICATIONS
LAGARD ne detient paSde stocks de sespublications. dlansan but dledistribution getterale a l'adresse ci-dlessus. La diffusion initiate dles
publications de 'AGARD cst effectuee aupres des pays membres de cette organisation par I'iermcdtaire des Centres Nationaus de
Distribution suivants. A 1exception des Etats-Unis. cescentres disposent parfois dexemnplaires adaitionnels:cdans les cascontrairc. on pent
se procurer ces exemplaires sous furmne de microfiches ou de microcopies aupr~s des Agences dleVente dont la liste suite.
CENTRES DE DIFFUSION NATIONAUX
ALLEMAGNE ISLANDE
Fachintormationszentrum. Director of Aviatioii
Karlsruhe c o Flugrad
D-75 14 Eggenstein-Leopoldsbalen 2 Reykjavik
BELGIQUE ITALIE
Coordonnateur AGARD-VSL Aeronautica Militaire
Fiat-Malor de fit Force Aerienne Ufficio del lDelegato Nazioiale al'AjA RI)
Quartier ReinecElisabeth 3 Piazzale Adenanter
Rue d'Evere. 11410Bruxelles II0144 Roma Et IR
CANAD)A [LUXEMBOURG;
lDirecteur du Service desReuseigitements Scientifiques Voir Belgique
%linisiwrcde lit Defcns~eNationale NORVEGI>
O~ttawa,.On' rio K I A llK2 Norwegian Defece Researich Establishment