Gian Luca Ghiringhelli, Stefania Gualdi

Dipartimento di Ingegneria Aerospaziale Politecnico di Milano

Mario Boschetto, Riccardo Bianco-Mengotti
Aermacchi S.p.A.

Analysis of Landing Gear Behaviour for Trainer Aircraft

Paper presented at
15th European ADAMS Users Conference
Rome, 15-16 November 2000

Analysis of Landing Gear Behaviour for Trainer Aircraft

Analysis of Landing Gear Behaviour

for Trainer Aircraft
Gian Luca Ghiringhelli, Stefania Gualdi
Dipartimento di Ingegneria Aerospaziale
Politecnico di Milano
Mario Boschetto, Riccardo Bianco Mengotti
Aermacchi S.p.A.

Abstract. In the field of aircraft landing analysis, the use of a multibody code such as ADAMS
has enormous potential: it allows the designer to create as complex a model as desired, with the
possibility of integrating subsystems and of accounting, for example, for aerodynamics and
structural flexibility. In this paper, an approach to the application of ADAMS in the analysis of
landing gear behaviour and ground manoeuvring characteristics of a trainer aircraft is
presented. The models developed are to be used during the design phase for the evaluation of
the loads transferred to the fuselage.
The first step taken towards the implementation of the aircraft model was that of separately recreating the behaviour of the auxiliary and main landing gear during drop-testing. In this phase,
a rigid-body model was examined, along with built-in ADAMS tire models and external, userwritten tire models. System components such as shock absorbers, for example, were modelled
using the know-how previously acquired with a dedicated code.
After having validated the ADAMS model using the results obtained in a parallel study
conducted with the dedicated code, the auxiliary and main landing gear were assembled and a
complete aircraft rigid-body model was built. The model was then used to simulate the
behaviour of the landing system during ground manoeuvring, using linear aerodynamics.
The model is planned to evolve, integrating flight mechanics for the approach and touch-down
phase, and modelling the hydraulic subsystem.

Introduction
In the context of the development of a new advanced trainer aircraft, Aermacchi and
Politecnico di Milano are co-operating on a project aimed at the evaluation of the loads
transferred to the fuselage and the structural behaviour of the landing gear during
impact and ground manoeuvring using dynamic and kinematic simulations. The need
for integrated models as early as in the design phase, able to cater for the requirements
of the different project design areas, led, during a similar cooperation in the 1980s, to
the development of a dedicated code, GRAALL (Ground Roll Air And Landing Loads),
used for such aircraft as AMX, S211, SF260 and MB339 [1]. As with most dedicated
codes, it was relatively simple to construct and tune a model, obtaining very precise
results, as long as one remained within the limits of its development: any attempt to go
beyond encountered the necessity of additional, burdensome programming. In order to
overcome this intrinsic, time consuming limitation, a new approach, based on the use of
an industrial, multi-purpose, multi-disciplinary code was prospected. The choice fell

Analysis of Landing Gear Behaviour for Trainer Aircraft

upon ADAMS, a tool in continuous evolution, supported by a community of developers

and users oriented towards similar design problems, which offers significant advantages
in terms of flexibility in modelling phenomena, parameterisation, optimisation, and, last
but not least, ready integration with widely used, general purpose programs.
The experience acquired with the previous code has proved to be very useful for the
construction of landing gear multi-body models in ADAMS. The versatility of the latter
has in fact allowed an easy implementation of the dedicated software equations,
reducing the conversion time between the two calculation procedures. For example, the
oleo-pneumatic shock absorber model was developed following the method already
implemented in the dedicated code, which allows for both conventional orifice and
variable diameter metering pins. Analogously, the FIALA tire used for the simulations
with ADAMS was tuned using both experimental data and simulations with the
GRAALL tire model applied in ADAMS as a user-defined subroutine. Furthermore, the
first validations on the multi-body model were carried out by comparing the results
obtained on the corresponding models built and simulated with the previous method.

Model construction
The first step taken towards the construction of the complete model for the new
advanced trainer aircraft was the separate implementation in ADAMS of the main and
auxiliary landing gear with rigid body models, designed on the basis of an existing
undercarriage with similar performance requirements. Experimental drop test results
and previously developed models were therefore available for validation of the basic
model. Successive modifications, either geometric or functional, corresponding, for
example, to variations of the shock absorber metering pin geometry or initial pressure
value, could then be carried out on the basis of the validated model.
The basic parts considered for each gear were the attachment braces (the actual interface
towards the fuselage), the structural cylinder, the shock absorber, the fork, and the
wheel. For the nose gear, an additional part representing the steering cylinder was
introduced.
Various tire models were examined and in particular two were chosen for comparison:
the built-in FIALA tire and the GRAALL tire, implemented as a GFOSUB. The
principal difference between these models resides in the calculation of the vertical force:
the FIALA model treats the tire like a beam on elastic foundation, with a linear vertical
stiffness coefficient, while the GRAALL model adopts a geometric approach, with the
vertical force calculated using a polytropic compression based on the intersection
volume of a torus with the ground, thus equivalent to a non-linear vertical stiffness. The
FIALA model implies an elaboration of test data prior to the implementation in order to
determine the stiffness coefficient, while the GRAALL tire parameters correspond to
physical quantities, such as inflation pressure, polytropic exponent, tire initial internal
volume.

Analysis of Landing Gear Behaviour for Trainer Aircraft

Once all the parts had been defined and their mass characteristics assigned, the singleacting oleo-pneumatic shock absorber with its force components was implemented,
using the model developed for GRAALL. The hypotheses behind this model are of gas
polytropic compression and fully turbulent oil flow through the orifices, which lead to
the following well-known force expressions [2]:
1.)

elastic force:

p0
Agas
l
cint
l

2.)

viscous force:

Atr
Cd

frictional force:

cfriction
vref

Fviscous =

A3 s s
1
s
2
2
( Atr Cd )

hydraulic oil density

oil chamber reference area
shock absorber stroke velocity
orifice area
flow coefficient

s

s

initial pressure
gas chamber reference area
gas chamber initial length
oil-gas interaction coefficient
shock absorber stroke
polytropic exponent

As

3.)

Felastic

l
= p0 Agas
l cint l

s
F friction = c friction Felastic Tanh
vref

friction coefficient
shock absorber stroke velocity
reference velocity

An IMPACT force was implemented in order to limit the shock absorber fully extended
length.
A series of simulations aimed at reproducing drop test experimental results brought to
the proper tuning of the model components, namely tire and shock absorber parameters.
The existing simulation results, tuned on the basis of experimental data, were used as
validation material. Comparative testing of the two tire models brought to the
conclusion that, at least for this particular case, the FIALA tire model could be used for
the development of the landing gear ground manoeuvring model with an acceptable
margin of error. In fact, under the loads examined, the tire does not present an
exasperated non-linear behaviour, but can be reputed fairly linear, as can be seen in
figure 1 graphic 4. The hysteresis loop is somewhat different, but in the case under
examination, as can be seen in the other graphics of figure 1, the repercussions on the
gear behaviour are minor.

Analysis of Landing Gear Behaviour for Trainer Aircraft

1) Shock absorber stroke vs. time

4) Tire vertical force vs. deflection

2) Shock absorber force vs. time

5) Tire longitudinal force vs. time

3) Shock absorber force vs. stroke

6) Tire vertical force vs. time

Figure 1: Simulation results for the MLG drop tests: GRAALL continuous line, ADAMS dotted line

Once the separate components were tuned, the rigid body model of the complete
undercarriage was assembled and the fuselage mass characteristics were assigned to a
part reproducing the aircraft centre of gravity, constrained to the main and nose gear via
spherical joints and very stiff bushings.

Figure 2: Complete aircraft model

Analysis of Landing Gear Behaviour for Trainer Aircraft

Linearised aerodynamic forces, dependent upon the aircraft pitch and yaw angles, were
then applied at the aerodynamic centre. This represents a large step forward with respect
to the possibilities offered by the previously developed dedicated code and allows a
simulation of the aircraft behaviour which is far more adherent to reality, as will be
discussed further on. Some difficulties were encountered in the correct determination of
the aircraft attitudes, that is model pitch and yaw angles relative to the trajectory,
necessary for a correct evaluation of the aerodynamic forces (relative wind direction):
mathematical models were used to overcome the problem.
At this point, the model was ready to be used in the simulation of ground manoeuvres.
To start with, two basic cases were approached: symmetrical braking and steering.

Ground manoeuvre simulations: symmetrical braking

For the symmetrical braking manoeuvre model, the hypothesis of an ideal ABS system,
able to constantly maintain the maximum allowable tire-ground friction coefficient, was
made, thus guaranteeing the maximum braking effect and consequently maximising the
longitudinal load. Under this assumption, the FIALA tire was used with built-in tireground friction disabled while the braking forces were implemented as forces and
couples acting directly on the axle:
Fbraking=(v)Fztire
Mbraking=RrollingFbraking
(v)
Fztire
Rrolling

ground friction coefficient (varies with aircraft forward speed)

tire vertical force (FIALA model)
tire rolling radius

Being a symmetrical manoeuvre, no lateral tire forces were taken into account. This was
possible thanks to the particular choice of landing gear geometry, in which the axles
only undergo vertical and longitudinal movement during gear deflection.
Various mass distributions and initial velocities were examined. An example of the
output obtained can be seen in figure 3, where the presence of the aerodynamic forces
yields a realistic evolution of the manoeuvre: as the aircraft speed decreases, the vertical
load on the landing gear increases, thus augmenting the available braking force.

Analysis of Landing Gear Behaviour for Trainer Aircraft

1) CG vertical and longitudinal acceleration

2) Tire vertical forces and braking force

3) MLG fuselage attachment vertical and longitudinal forces

4) NLG fuselage attachment vertical and longitudinal forces

Figure 3: Braking simulation results

The results obtained in the different configurations were then compared to those
obtained with a simplified program based on the MIL specifications and to those
obtained through theoretical studies. The outcome was deemed satisfactory, especially
considering the fact that the MIL-based program does not include the effects of
aerodynamics or of a varying tire-ground friction coefficient [3], and, more important, it
does not consider the transient behaviour. The simplicity of the model leads, in fact, to a
conservative envelope for steady loads; there is no guarantee that transient loads are
encompassed, as can be seen in the case of the nose landing gear in figure 3 graphic 4.
In the specific case, a cosinusoidal braking transient was applied, bringing the braking
force to full value in 0.2 seconds.
Having included aerodynamics and modelled the transient certainly results in a more
realistic model behaviour, but this is costly in terms of complexity and requires the
tuning of a greater number of design variables. This can render the application of the
detailed model impractical in the early stages of the design.
A non-symmetrical braking model is planned to be implemented in the near future; of
course, in this case the lateral tire force becomes an important issue, thus the FIALA
model will be used with tire-ground friction enabled and an ideal ABS system will be
implemented as a PID controller on the braking torque, this time applied directly on the
tire body.

Analysis of Landing Gear Behaviour for Trainer Aircraft

Ground manoeuvre simulations: steering

In the steering manoeuvre model, the FIALA tire was used with the tire-ground friction
enabled. The steering motion was applied as a PID controlled torque and not as an
imposed joint motion, thus allowing a more realistic approach to the actual steer-bywire mechanism. Moreover, the over-constraining imposition of the joint motion would
have altered the lateral stability of the system.
A first series of simulations, conducted with constant velocity on the trajectory,
guaranteed by a PID controller traction force applied in the centre of gravity, allowed
the determination of the limit steering angles and turning radii for each rolling speed.

Minimum turning radius

1

0,9

0,9

0,8

0,8

turning radius

steering angle

Maximum steering angle

1

0,7
0,6
0,5
0,4
0,3

0,7
0,6
0,5
0,4
0,3

0,2

0,2

0,1

0,1
0

0
0

0,1

0,2

0,3

0,4

0,5

0,6

0,7

0,8

0,9

speed

0,1

0,2

0,3

0,4

0,5

0,6

0,7

0,8

0,9

speed

Figure 4: Example of steering model output

A second series of simulations was conducted at each limit steering angle, without
traction forces applied, thus allowing the speed along the trajectory to decay. An
example of the simulation results can be seen in figure 5, where the importance of the
transient behaviour in determining the maximum loads can be appreciated.
Once again the results were compared to those obtained with the simplified MIL-based
program, which considers steady state loads without aerodynamic effects and without
modelling the transient. The same considerations expressed earlier for the braking
model apply: in order to correctly define the transient, the model complexity increases,
thus making it difficult to use for preliminary calculations.

Analysis of Landing Gear Behaviour for Trainer Aircraft

1) CG trajectory in the plane of motion

2) aircraft attitude vs. time

3) Tire vertical force vs. time

4) Tire lateral force vs. time

Figure 5: Steering simulation results

Conclusions
The outcome of this experience can be summarised as follows: the models available at
present are far more adherent to the actual operating conditions than those previously
developed with the dedicated code. ADAMS in fact has the advantage of allowing the
user to properly take into account most of the major factors influencing the physical
behaviour with a relatively low modelling time expenditure, if compared to the
elaboration of a dedicated code. This does not mean that additional programming is not
necessary: some of the applied forces have to be defined through external, user-written
subroutines, since their implementation using built-in ADAMS functions is extremely
complex if not nearly impossible in some cases. The complexity of the model itself, thus
the large amount of work required in order to implement it in ADAMS, corresponds
though to a much greater versatility and ease in defining different simulation conditions,
resulting in a more efficient design procedure, able to follow all of the typical designphase evolutions, but, due to its complexity, being less efficient in the preliminary steps.
The project is planned to evolve with the introduction of flexible elements, for example
the attachment braces, which are the primary source of dynamic oscillations present in

Analysis of Landing Gear Behaviour for Trainer Aircraft

the landing gear-fuselage attachments. A more detailed aerodynamic force description

will be added to the model and ADAMS will also be linked with an external subroutine
that will generate the flight mechanics during approach, in order to trim the aircraft and
simulate the whole landing manoeuvre in various flight conditions. This will for
example permit the realistic simulation of non-symmetric touch-down configurations.
Another aspect that will be developed is the parallel study of the hydraulic system using
ADAMS Hydraulics and linking ADAMS with AMESim, which will allow the
simulation, for example, of the retraction kinematics, of the braking system and of the
steering system.
This evolution will eventually lead to the integration and synergy with other projects
under development both at Aermacchi and at Politecnico di Milano: the reconstruction
and simulation of accidents involving aircraft landing gear failure, the verification of the
kinematic behaviour of aircraft flight command systems in various operating conditions,
nose gear shimmy simulation, just to name a few.

References
[1] M. Boschetto, G.L. Ghiringhelli, "Design Loads Evaluation by Dynamic
Simulation of Flexible Aircraft", AGARD Landing Gear Design Loads Specialists'
Meeting, 1992
[2] B. Milwitzky, F.E. Cook, Analysis of Landing Gear Behavior, NACA Report
1154, 1954
[3] ESDU, Frictional and retarding forces on aircraft tyres, Part 2: estimation of
braking force (friction data updated 1981), ESDU Data Item N 71026,
October 1971 (with Amendments A and B, August 1981)
[4] N. S. Currey, Aircraft Landing Gear Design: Principles and Practices, AIAA
Education Series, 1988