2017 2nd International Conference for Convergence in Technology (I2CT)

Numerical Investigation of Structural Design of Torsion

links in a Landing gear retraction mechanism
Krishnan.A*, V. Thejus, BM Arjun
Department of Mechanical Engineering
Amrita School of Engineering, Amritapuri
Amrita Vishwa Vidyapeetham, Amrita University, India
*aravindkrishnan.47@gmail.com

Abstract-Torsion links are components which couple inner and outer highly stressed regions in the components at a particular time
cylinders of a landing gear mechanism. They prevent torsion or step and to modify the shape if necessary. It is done using
twisting between the cylinder and piston components in the MotionView and HyperMesh tools.
structure. The integrity of this component is essential to prevent
shimmy vibrations in an aircraft landing gear. Failure of this
component is thus catastrophic for the aircraft. The typical
geometry of torsion link of a medium sized aircraft is shown in
figure 1. In this paper, an attempt is made to study stress contours
and maximum bending stress on the torsion link and the effect of
changing the holegeometry and hole patterns on it, from this the
best optimised model is selected and modal analysis is done on it. The
upper torsion link is the component on which the analysis is done.
This component behaves as a simply supported beam with a
moment on one end. It is seen that maximum bending stresses
occur on the top surface of the upper torsion link on the ends near
the flanges. This analysis is done on MotionView and HyperMesh
tools. It is required to minimize the stresses by changing hole
positions and geometry. It is to be noted that changing hole
geometry affects total volume of the component and that volume of
aircraft components is to be maintained at minimum. So a balance
between reducing stress and volume needs to be achieved. These
stresses increase with increasing distance between holes. The effects Fig. 1 Landing gear mechanism
of giving multiple holes and combining holes are also studied. The
optimised model from stress analysis is further analysed through The geometry is altered keeping in mind the objective of
modal analysis. The model is analysed in free – free run using
ANSYS. From this six natural frequencies of the optimised model is
reducing weight and maximum stress. It is tried to obtain a
obtained and further discussion on the mode shape is done. general trend in the variation of stresses, also trying to keep the
volume at minimum value. A modal analysis is also performed
Keywords-Torsion links, landing gear, Stress analysis, shape on the optimized models. Landing gears are subject to large
optimisation, modal analysis sources of vibration during take-off and landing.Phanikrishna,
Bernad and Mark[1] did a study on the onset of shimmy
I. INTRODUCTION oscillation of an aircraft landing gear to find the regions of stable
The landing gear system is one of the most important systems in torsional and shimmy oscillations. It is essential to identify the
an aircraft. The basic model used for analysis is shown in Fig. 1. natural frequencies and mode shapes of torsion links to avoid
The retraction of landing gear is a mechanism which is simulated resonance. Available literature on study of torsion links is very
with utmost care so that the components within the systems are less in number. In the work done by FatmaKocer [2], topology
not over-constrained. Equally important is the requirement of optimisation of torsion links are done to reduce overall weight.
conducting a stress analysis at no load condition on the system. It The initial model has two holes as shown in Fig. 2. This study
is a dynamic system and so, the stress analysis gives the stresses will help engineers reduce weight of components by removing
at each time step of the mechanism. This dynamic analysis gives material from the main component. The study by Arravind,
the effect of inertial stresses on the torsion link. In other words Saravanan and Rijuvan [3] uses carbon fibre in composite
the self weight of components causes stress on torsion links and materials for the design, optimisation and analysis of composite
they have to be analysed. This analysis helps to identify the torsion links. In this study, the positions of holes and geometry

978-1-5090-4307-1/17/$31.00 ©2017 IEEE 382

are changed, to obtain an optimal design. Further, a modal
analysis is performed on optimized designs and results are
discussed. Finally an alternative design is proposed in this paper.
II. THEORY
The component in hand is a torsion link. From the mechanism
motion, we can see that it behaves as a simply supported beam.
There is a motion given on the one end of the component in the
mechanism. This causes a moment. Effectively, the component Fig.4: cross sections of (a) Shorter end (b) Wider end
acts as shown in the figure 4 below. Since the shape of the given
link is non-uniform, it is required to make reasonable From the structure and loading, we see that this link behaves as a
assumptions to propose a theory for the stresses on the link. simply supported beam with a moment at the wider end. The
flanges have uniform dimensions. The diameter of the smaller
A. Analytical Model: hole is 15.2mm and the larger one is 34.2mm. The distance
Fertis and Demeter in their book Nonlinear Structural between the centres of holes is 57.2mm.
Engineering [4] describe a theory for solving nonlinear We need to formulate this as two parts for x<95 and x>=95mm.
beams, with all loads and boundary conditions. Let us The width varies according to the formulae
ͳǤͲͷš
assume that both the ends of the link do not participate in the ܾሺ‫ݔ‬ሻ ൌ ቐ
ʹͲሾͳ ൅
‫ܮ‬
ሿǡ ‫ ݔ‬൏ ͻͷ

bending. These ends are where the joints are present. So it ʹͲ ൅ ͻͷ –ƒሺͳʹǤͷሻ ൅ ʹ‫݊ܽݐݔ‬ሺͳʹǤͷሻǡ ‫ ݔ‬൒ ͻͷ
behaves as an H-section non prismatic non uniform beam as (1)
ܽ൅ܾ
shown in figure 3. The width of the web differs at both ends. Area of trapezium is given by h ቀ ቁ (2)
ʹ
Our second assumption is that the width varies uniformly Hence area of web section varies as
from one end to the other. This greatly helps us to formulate š
ሾʹͲ ൅ ʹͲሾͳ ൅
ͳǤͲͷš
ሿǡ ሿǡ ‫ ݔ‬൏ ͻͷ
the variation of cross-section with respect to the distance and ܾሺ‫ݔ‬ሻ ൌ ቐ ʹ ‫ܮ‬
(3)
ͳͻͲ –ƒ ሺͳʹǤͷሻ൅ʹ‫ ݊ܽݐݔ‬ሺͳʹǤͷሻ
hence, calculate area, moment of inertia and other ‫ݔ‬ሾ ሿǡ ‫ ݔ‬൒ ͻͷ
ʹ
parameters. Where L=210mm
From the equation of moment of inertia of an I section, I is
Iflanges+ Iweb and
͵Ǥͺ ͵ ͵Ǥͺ ͵
Iweb varies as .b(x) (4)
ͳʹ ͳʹ

So, moment varies as

͵Ǥͺ͵
I(x) = 2Iflanges + b(x). (5)
ͳʹ

For a uniform simply supported beam, with a moment at one end

Fig. 2: Top view with two holes; the ends are assumed to be inactive in bending refer fig. 5, deflection at any point x is given by
As shown in fig. 3 and fig. 4, the length of the link is 210mm. y=
‫ݔ݈ܯ‬
ቂͳ െ
‫ʹݔ‬
ቃ (6)
The flanges are 7.6mm thick. The web is 3.8 mm thick, and ͸‫ܫܧ‬ሺ‫ݔ‬ሻ ‫ʹܮ‬
width varies uniformly from 20 to 92 mm. and the maximum deflection is
‫ʹܮܯ‬ ‫ܮ‬
δmax = at x= (7)
ͻξ͵‫ܫܧ‬ ξ͵

Fig. 5: The torsion link is an effective simply supported beam with the above
Fig. 3: Top view of web boundary condition

383
Now, stress at any point x can be calculated using Euler’s hole. Stress concentration factor K is defined as ratio between
equation max and normal of σh.
› ɐŠሺƒš ሻ
σx = (8) K= (15)
‫ܫ‬ሺ‫ݔ‬ሻ ɐŠሺ‘‹ƒŽ ሻ
WhereMxis the moment at the point x and y is the distance from
neutral axis and I(x) is the moment of inertia at that section.

B. Governing Differential Equation

Using energy methods, a beam with a moment on one side
may be analysed to obtain the governing equations.
Consider the element of area dA , y from the neutral axis,
then (V My / I ) , and using (8):
Fig. 6: Stress variation near a hole.
2
2
V 1 § My ·
¨ ¸ dV
U i ³ 2E
dV ³ ¨
2E © I (x)
¸ Regina Hermelin[5], created finite element models to investigate
V V ¹ (9)
stress distributions in beams with holes in the web to calculate
The volume dV can be written as dV dAdx , so: where fracture would occur and how it would grow. According
L 2
M 2 to formulae provided by Finnish manufacturer FinnForest, the
U i ³ 2 EI ( x ) ³ y 2
dAdx
reduced characteristic shear force capacity Vk,holeisthen calculated
0 A
(10) using(16) [6].
Will give the strain energy in the member, and since ³ 2
y dA I: Vk,hole= 1.1kholeVk≤ Vk (16)
A WhereVk is the characteristic shear capacity for the beam
L
M
2 without the additional hole. The reduction factor khole is
U i ³ 2 EI ( x )
dx calculated using(17) [7].
0
(11) ݄‫ ݓ‬൅݄݂Ȃ݇‫ ݈݁݋݄݄ ݁݌݄ܽݏ‬Ȃ͵ͺ݇
If a moment is applied to a body with linear-elastic material Khole  for0 ≤ khole≤ 1 (17)
݇‫ ݂݂݁ݓ‬െ͵ͺ
behaviour such that the magnitude of the couple increases from 0 Wherehwis the depth of the web and hfis the flange depth. The
at I 0 to M at I T then: factor kshapeis1 for circular holes and 1.23 for rectangular holes
T and hholeis the diameter in caseof a circular hole and the largest
M 1 side of the hole in case of a rectangular hole. The factorhw,effis
We ³ IdI MT
T 2 calculated with (18) and the factor k with (19)or (20). The
0
(12) thickness of the web is denoted bwand H is the total heightof the
Potential energy is given by strain energy – work done. beam.
The angle and length may be related as follows θ = 2δ/L ͵ͷ„™
݀ܲ hw,eff = (hw + hf ) ≤ hw + hf (18)
Using energy conservation = 0 Š™
݀Ɂ ʹͷͲ Ȃ‫ܪ‬Ȃ݄݄‫݈݁݋‬
This helps us to obtain the values of deflections and reach the For H ≤ 212 m:k= ,0≤k≤1 (19)
͹͸
stresses ‫ܪ‬െ݄݄‫ ݈݁݋‬Ȃͳ͹Ͷ 
For H ≥ 212 m:k= ,0≤k≤1 (20)
͹͸
C. Effect of hole and stress concentration bending and shear
It is important to determine the natural frequency of the
For discontinuity such as a hole in this case, drilled at a distance
component so as to avoid any fatal accidents. This is done by
x from an end of the components, stresses will vary. At the
studying the dynamic properties of the torsion link under
section with hole, nominal stress based on area of section is
given by excitation through modal analysis.The work done by Kruger and
› Morandini[8] gives a general outline is of how vibration
σh(nominal) = (13) problems in landing gears can be treated by numerical analysis
‫ܫ‬ሺ݄ሻ
where I is now given as Iflanges+ Iholesection and methods .
Iholesection= (b(x) – dh)
͵Ǥͺ ͵
(14) Natural frequency of an object is that frequency with which any
ͳʹ object will vibrate if disturbed and allowed to vibrate on its own
where dhis the hole diameter. The stress around the hole varies without any external force.
due to the stress concentration effect as shown in fig. 6 (refer ωn = (k/m)1/2 (21)
appendix). σh increases from the edge of the link to the hole. The where k and m are the stiffness and mass of the object
maximum stress σh(max) is at a point very close to the circular respectively. An object can have infinite natural frequency.To

384
find the natural frequency and mode shapes we start with the TABLE 1: PROPERTIES OF ALLOY
equation of motion in matrix form. We have Property TiASM4911
[M]{X″} + [C]{X’} + [K]{X] = {F} (22)
Where [M], [C], and [K] are the mass, damping and stiffness Young’s modulus(MPa) 113800
matrices. {X} is the displacement vector and {F} is the overall
Shear Modulus(MPa) 44000
force acting on the system. For free, undamped response we can
set [M], and {F} to zero[9]. So we have
Density(g/cc) 4.43
[M]{X″} + [K]{X] = 0 (23)
Above equation is ordinary homogeneous linear second order
Poisson’s ratio 0.33
differential equation with constant coefficients. We are looking
for a solution of the form {X} = est{u}. Where s = iω and {u}
represents the mode shape. Substituting this in the equation of
The mechanism was run using MotionView [12] tool. The
motion we get
torsion link component in the mechanism was converted to a flex
([K] – ω2[M]) {u} = 0 (24)
body in MotionView using theFlexPrep tool. The HyperMesh
Here we have an eigenvalue problem from which a non-zero
[13] file created earlier was accessed by FlexPrep tool. The
solution for {u} can be found from
multi-body analysis simulation analysis is run again using the
det([K] – ω2[M]) = 0 (25)
flexible torsion link. The results are viewed using HyperView
If [K] and [M] are n×n matrices then we will have n values of
[14] post processing tool. The stress contours are studied.
natural frequency (ω). Corresponding to each natural frequency
Similarly for the second case, geometry is altered in HyperMesh
we can get a mode shape {u} for the system.
and the above process is repeated. The first set is done by
III. MODELLING AND ANALYSIS moving the smaller hole away from the bigger hole by increasing
centre distances.
The CAD model of the torsion link show for bending analysis in A free-free modal analysis is done using ANSYS. At first the
fig. 7 was used in the mechanism for multi-body simulation. This torsion link model to be analysed is imported and the properties
model is meshed (tetra mesh of element size 5) and the material are specified. The model is then meshed using an element size of
properties were given using HyperMesh. A solver deck must be 5mm, and its natural frequencies are found out. The first six
exported and saved to be used later by FlexPrep. The meshed natural frequencies are not considered. The different modes of
model is shown below in figure 7. The material used for torsion vibration for different frequencies are noted. The vibrations due
links are a Titanium alloy Ti6Al4V AMS 4911[10]. The to external excitations are approximated to 10 kHz, so the natural
properties of this alloy are shown in Table 1. frequencies of the body should not be around this value to avoid
resonance.
The best optimised CAD model of the torsion link, obtained after
the stress analysis, is used for modal analysis. This modal is IV. RESULTS
meshed (tetra mesh of element size 5).Meshing is an important
A. Bending Stress Results
and critical step in finite element analysis and it directly affects
the outcome of the complete process [11]. The material and The first set of results is obtained by changing the distance
properties of this model is same as that of the model used for between the holes, increasing it in 4 steps. The distance is
stress analysis. increased from 55.7 to 105.74mm. It is increased as 4 steps of
16.667mm. The stress contours are as seen in the figures below.
We see that the stress has a positive correlation with the distance
between holes. We see that the maximum stresses occur on the
flanges near the larger hole. The reason for this is very obvious
since the area is smaller there and that section is at a higher level
from the neutral axis as described in theory section.

As seen from the fig. 8 and fig. 9, maximum stress occurs on the
flanges and the values are seen to increase from 477.2MPa to
Fig. 7: Meshed model of torsion links 481.0MPa in small steps of close to 1MPa. The stresses on the
section increased, which is because the distance to the flange
section increased and so the moment increased. Increase in
moment caused the bending stresses to increase. In this case the
volume remains the same.

385
 Fig.11: centre distance is 105.734mm; Hole sizes unchanged

Fig. 8 Centre distance is 57.2mm; Hole sizes unchanged

Fig.12: Hole radius 30mm

Fig. 9 Centre distance is 72.4mm; Hole sizes unchanged

Fig.13: Hole radius 26mm

The second set of tests is done by decreasing diameter of the
larger hole (See fig. 10 to fig. 15). The radii were reduced from
34.2 to 24 to 26 and finally to 22mm. The between holes was
maintained at 105.7 mm. This gives a reduction in stresses as
area on which stress acts increases as hole diameter decreases,
according to the formula specified in theory. Case 4 from the
previous set is considered as case 1 here, where hole radius was
34.2 mm.

Fig.14: Hole radius 22mm

Fig.10: centre distance is 89.067mm; Hole sizes unchanged

Fig.15: Smaller hole radius increased to 16.2

From this set, we see that decreasing hole radius decreases

stresses. This is not a practical solution since it causes increase in
weight. The stress was 481MPa, and it got reduced to 418 to 408
and finally to 402MPa.

386
Similarly, the smaller circle radius is increased from15.2 to Case Smaller Larger Distance Volume Maximum
16.2mm at distance of 107.4mm from the larger hole, and we see hole radius hole between (mm3) Von Mises
(mm) radius centres stress
that stresses increased. It is shown in Fig. 15. Stress in Fig. 11 (mm) (mm) (MPa)
was 481MPa and it increased to 483MPa. 1 15.2 34.2 57.2 479195 477.2
Tests with multiple holes: 2 15.2 34.2 72.4 479195 478.0
The next test was done by putting 4 holes; the bigger hole 3 15.2 34.2 89.067 479195 479.8
remains the same, and 3 smaller holes, 2 of 15.2 mm radius and a 4 15.2 34.2 105.734 479195 481.0
5 15.2 30 105.734 480690 414
smaller one of 7.5mm radius. The distance between centres of 6 15.2 26 105.734 483356 408.7
holes is 66.66 mm. This helps in further increase in volume 7 15.2 22 105.734 485985 402
reduction. Volume reduction is always considered favourable. 8 16.2 34.2 105.734 477430 483.0
Here we see the following result: stress is just 421MPa which is 9 7.5, 15.2, 34.2 66.6, 66.6, 474477.5 420
15.2 66.6
much lesser than all the cases shown above. The volume too is 10 16.2 34.2 105.734 465960 420
lower. Hence it is a desirable design. This is shown as Fig. 16.
Tests with new design: The results are plotted below in fig. 18 and fig. 19.
The final test was done by combining two holes of radiuses 15.2
and 34.2 with a centre distance of 107.2mm. This has maximum
120
reduction in volume and also, the stresses are same as in Fig. 16.
We see now that, this is a better design than the previous one. It

Distnace between holes

100
is shown in Fig. 17 below.
The 10 cases have been tabulated and shown in the table 2. 80

(mm)
60

40

20
larger hole radius center distance
0
476 478 480 482
Maximmum Stress Values (MPa)
Fig. 18 Variation in stresses with centre distance and hole radius

Fig.16: Multiple holes 10 stress volume

9
8
Case Numbers

7
6
5
4
3
2
1

Fig. 17: Combined holes 400 450 500

Maximum Stress (MPa)
TABLE 2: TABULATED RESULTS Fig.19 Variation of stresses and volume in each case

B. Modal Analysis results

The different mode shapes obtained for higher modes (rigid
modes are avoided) are shown below in fig. 20to fig. 22 and their
respective frequencies and planes of deformation are tabulated in
Table 3. The natural frequencies obtained are tabulated. The
mode shapes show bending in most cases and 1 torsion mode.
The values of first modes are below 1 kHz. It is seen to rise and

387
rate of increase keeps rising till mode 18.The last modal V. CONCLUSIONS
frequency is 2.2 kHz. These values are important parameters in It is seen that maximum stress occurs at the hole section where
deciding the resonance characteristics of the component. The the cross-section is thin. This stress is dangerous for the structure
forcing frequencies have to be free from these natural frequency and must be reduced. This stress occurs as a result of the bending
values. moment on one end of the torsion link, given to the mechanism.
This causes the bending stress. From these results of the above
tests it can be concluded that it is advantageous to use multiple
holes in the case of torsion links. This helps in reducing volume
without a considerable increase in stress. Stresses can be reduced
by decreasing larger hole diameter, but this leads to increased
volume. This is not a desirable change. So designers have to
balance the need to reduce stresses and reduce volume. Still a
Fig. 20 Mode shapes for modes 13 and 14
better design, the one used in case 10 can be achieved by
combining two or more holes. This case has least volume and a
relatively low value of stress may be used. Hence an alternative
model was proposed to suit the specific needs for the component.
As for modal analysis, the natural frequency of vibration of
different mode shapes obtained ranges from 0.7 kHz to 2.252
kHz.

Fig. 21 Mode shapes for modes 15 and 16 ACKNOWLEDGMENT

A part of the study for this article was conducted in Altair
Engineering, Bengaluru, India between January and May 2015.
The authors would like to thank the company and members of
HyperWorks core development group for the opportunity to
conduct the research and use all the tools required .

REFERENCES
Fig.22 Mode shapes for modes 17 and 18 [1] Phanikrishna Thota,Bernd Krauskopf, Mark Lowenberg, Interaction of
torsion and lateral bending in aircraft nose landing gear shimmy, Nonlinear
TABLE 3: MODAL ANALYSIS RESULTS Dynamics, Volume 57, Issue 3, pp 455–467,August 2009.
[2] Fatma Kocer, Multi-Disciplinary Design Optimization of an Aircraft
Component Natural Plane of Landing Gear, NFAEMS world conference Canada, 2007
frequency(kHz) deformation [3] Arravind R, Saravanan M, Rijuvan M. FEA Based Analysis of Composite
Torque Link for a Passenger Aircraft Landing Gear, International Journal
on Theoretical and Applied Research in Mechanical Engineering
Mode 13 0.751 Bending in XZ
(IJTARME), ISSN : 2319 – 3182, Volume-2, Issue-2, 2013
plane
[4] Fertis, Demeter G, Non-linear structural Engineering, Ch 1, ISBN 978-3-
540-32975-6
Mode 14 0.852 Bending in XY [5] Regina Hermelin, Strength Analysis of Wooden beams with a hole in the
plane web
[6] European Organisation for Technical Approvals. (2004) European
Mode 15 0.992 Twisting in YZ Technical Approval No. ETA-02/0026, VTT rakennus-ja yhdyskunta
plane tekniikka, Espoo.
[7] Swelite. (2005) Projekteringsanvisningar, edition 2005:1, Rundvik.
[8] Wolf R. Krüger, Marco Morandini, Recent developments at the numerical
Mode 16 1.743 Bending in XZ simulation of landing gear dynamics, CEAS Aeronautical
plane Journal, Volume 1, Issue 1, pp 55–68, September 2011
[9] .Nithin S G, Sanjay S D, Sanjeev V B, Anand N T, Practical Finite Element
Mode 17 2.176 Bending in XZ Analysis, Ch 14, ISBN 978-81-906195-0-9
plane [10] Jean- Pierre Serey, Trends in Landing gear material, Aerospace
Engineering, September 2005
Mode18 2.252 Bending in XZ [11] H Boroouchaki and P.L. George, Quality mesh generation, Comptes
plane Renchus de I’Academie des Sciences – Series IIB – Mechanics, vol.328, pp
505-518, 6//2000
[12] Altair HyperWorks help manual MotionView documentation
[13] Altair HyperWorks help manual HyperMesh documentation
[14] Altair HyperWorks help manual HyperView documentation

388