Proceedings of the ASME 2012 International Design Engineering Technical Conferences &
Computers and Information in Engineering Conference
MODELING GEOMETRIC NON-LINEARITIES IN THE FREE VIBRATION OF A
PLANAR BEAM FLEXURE WITH A TIP MASS
Hamid Moeenfard1,2 and Shorya Awtar1
1
University of Michigan, Ann Arbor MI, USA
2
Sharif University of Technology, Tehran, IRAN
ABSTRACT
The objective of this work is to create an analytical
framework to study the non-linear dynamics of beam flexures
with a tip mass undergoing large deflections. Hamilton’s
principal is utilized to derive the equations governing the non-
linear vibrations of the cantilever beam and the associated
boundary conditions. Then, using a single mode approximation,
these non-linear partial differential equations are reduced to two
coupled non-linear ordinary differential equations. These
equations are solved analytically using combination of the
method of multiple time scales and homotopy perturbation
analysis. Closed-form, parametric analytical expressions are
presented for the time domain response of the beam around and
far from its internal resonance state. These analytical results are
compared with numerical ones to validate the accuracy of the
proposed closed-form model. We expect that the qualitative and
quantitative knowledge resulting from this effort will ultimately
allow the analysis, optimization, and synthesis of flexure
mechanisms for improved dynamic performance.
1. INTRODUCTION
Beams are important building blocks in many engineering
structures ranging from micro/nano devices to macro-scale
airplane wings, flexible satellites, and long span bridges [1].
Flexible beams are also one of the most important constitutive
elements in flexure mechanisms. Flexure mechanisms provide
guided motion via elastic deformation, instead of employing
sliding or rolling joints, and are used in a variety of applications
that demand high precision, minimal assembly, long operating
life, or design simplicity [2]. Since they exhibit motion
guidance as well as elastic behavior, flexure mechanisms are
also ideally suited for single- and multi- axis resonators, energy
harvesting devices, and high-speed scanners. However, large
motion range in flexure mechanisms implies large elastic
deflections of the constituent beams, which in turn give rise to
geometric non-linearities. Even though commonly ignored,
these non-linearities critically influence the dynamic
IDETC/CIE 2012
August 12-15, 2012, Chicago, IL, USA
DETC2012-71498
Downloaded from https://asmedigitalcollection.asme.org/IDETC-CIE/proceedings-pdf/IDETC-CIE2012/45035/363/4257462/363_1.pdf by University of Michigan user on 29 August 2019
characteristics of a flexure mechanism [3]. Depending on the
application, the relevant dynamic characteristics could include
vibrational mode shapes, flow of energy between modes,
bandwidth or speed of response, dynamic range, command
tracking, noise and disturbance sensitivity, closed-loop stability
and robustness, etc. As a result, investigating the non-linear
dynamical behavior of flexure mechanisms is of primary
importance in their design.
In a flexure mechanism, the elastic motion provided via
flexure beams is transferred to one or more moving stages,
which can initially be modeled as concentrated masses. In fact,
many flexure mechanisms can be represented as a system of
interconnected beams with point masses. Therefore, a logical
first step in investigating the non-linear dynamics of flexure
mechanisms is to consider and understand the vibrational
behavior of a simple Euler beam, with a tip mass at its end.
Such a study is the focus of this paper.
In general, the sources of non-linearities may be geometric
or material. The geometric non-linearity arises from arc-length
conservation of the beam and large deformation curvatures at
which the linear relationship between displacement field and
strains no longer holds. Material non-linearity occurs when the
stresses are non-linear functions of strains [4].
Because of its long, slender geometry, a typical beam
flexure may be modeled using the Euler-Bernoulli beam theory.
This theory assumes that plane cross-sections continue to
remain plane and normal to the neutral axis after deformation
[5], and has been successfully utilized to study the static,
dynamic, and vibrational behavior of beams. In particular, large
amplitude vibrations of beams have been extensively
investigated both theoretically and experimentally in the
literature. Crespo daSilva [6] formulated the non-linear
differential equations of motion for Euler-Bernoulli beams
experiencing flexure along two principal directions, along with
torsion and extension. He presented a reduced-order analytical
model for the non-linear dynamics of a class of flexible multi
beam structures [7]. Nayfeh modeled the non-linear transverse
vibration of beams with properties that vary along the length
1 Copyright © 2012 by ASME
[8]. Zaretzky et al. [9] experimentally investigated the non- strain expression is used for calculating its strain energy. For a
linear modal coupling in the response of cantilever beams. differential element on the centerline of the beam, the non-
The presence of a tip mass on the beam changes the linear strain  xx is as follows [13]:
differential equations governing its deflection. This is because 2
the inertial force exerted on the beam due to the presence of a u 1  w 
 xx     (1)
concentrated mass is a function of the deflection itself. Large x 2  x 
amplitude vibrations of beams with tip mass have also been where u and w are the displacements along X and Z axes,

Downloaded from https://asmedigitalcollection.asme.org/IDETC-CIE/proceedings-pdf/IDETC-CIE2012/45035/363/4257462/363_1.pdf by University of Michigan user on 29 August 2019

investigated in the literature. Hijmissen and Horssen analyzed respectively.
the weakly damped transverse vibrations of a vertical beam Using equation (1), the axial strain of an element at distance
with a tip mass [10]. Zavodney and Nayfeh studied the non- z from the neutral axis, along the Z direction, may be
linear response of a slender beam carrying a lumped mass to a expressed as follows [13].
principal parametric excitation [11]. But the axial dynamics of u 1  w 
2
 2 w 
the beam, which can become important at large deflections, was  xx      z 2 
not considered in these formulations. x 2  x   x  (2)
This paper presents an analytical investigation of the non- Using equation (2), the strain energy of the beam for a
linear in-plane oscillations of a flexure beam with a tip mass, linear elastic material would be
while including axial stretching. The Homotopy Perturbation
E E 
l  u
2 2
1  w  
Method (HPM) is employed because it does not depend upon 
     xx dA.dx        
2
the assumption of small parameters in the non-linear equations 2 V 2 0  x 2  x  
and takes full advantage of the traditional perturbation methods 
 2 w  u 1  w   
2
as well as homotopy techniques. HPM has been used to  2w  2
2

investigate non-linear vibrations of beams in the recent   2  z  2z 2       dA.dx (3)

 x  x  x 2  x   
literature. For example Moeenfard et al. used a combination of
HPM and the modified Lindstedt-Poincare technique to analyze 2 2
EA l  u 1  w   2
EI l   2 w 
2 0  x 2  x  2 0  x 2 
non-linear free vibrations of Timoshenko micro-beams [12]. In       dx    dx
this paper, HPM is utilized in conjunction with the method of 
multiple time scale perturbation method to solve the non-linear where l is the un-deformed length of the beam, A is the area of
dynamics of the beam. the cross section, and I is the second moment of the area of the
cross section about the neutral axis.
2. PROBLEM FORMULATION In long slender beams where u(x,t) = O(w(x,t)2), the axial
The beam with end-mass considered in this analysis is inertia of the beam can be ignored compared with the
shown in Figure 1. The dashed line represents the un-deformed concentrated inertial loads applied at the tip of the beam [4].
state, while the solid line represents a general deformed state. Assuming that axial damping is also negligible, the axial strain
The gravitational field, if any, is assumed normal to the plane εxx=∂u/∂x+(∂w/∂x)2/2 would remain constant along the length
and is therefore does not affect the planar analysis considered of the beam. In such a condition, the potential energy given in
here. equation (3) can be simplified as
2
EA  l  u 1  w   
2

      dx 
2l  0  x 2  x   
2
EI L  2w  EA 
   2  dx   u l, t   (3)
2 0
 x  2l 
2 2
1 l  w   EI L   2 w 
2

2 0  x   2 0  x 2 
  dx     dx

Figure 1 Schematic view of a beam with a tip mass
As the first step, the equations of motion and boundary
conditions corresponding to the transverse and axial vibrations
of a slender beam will be derived using the generalized
Hamilton’s principle. In the Euler–Bernoulli beam theory, plane
cross-sections remain plane and perpendicular to the neutral
axis after deformation, which implies that distortions due to
shear are neglected. These assumptions are applicable for long
and slender beams, with length much greater than the thickness
[5]. Since the beam undergoes large deflections, the non-linear
It can be shown that for an infinitesimal element of the
beam, the ratio of the rotational kinetic energy to the
translational kinetic energy is approximately of the order of
(h/l)2, where h is the thickness of the beam. Since for a long and
slender beam h  l , the rotational kinetic energy may be
ignored [4, 5]. Additionally, since in a planar beam flexure
u(x,t) is approximately two orders of magnitude smaller than
w(x,t), i.e. u(x,t) = O(w(x,t)2), the axial kinetic energy of a
beam element is at least four orders of magnitude smaller than
its transverse kinematic energy, and therefore may also be

2 Copyright © 2012 by ASME

ignored [4]. Thus, the total kinetic energy is simply given by: w
2 wˆ  (10)
1  w 
l l
2 0  t 
T    . A.dx u
(4) uˆ  (11)
l
1   u  l , t    w  l , t   
2 2

 M      E .I
2   t   t   tˆ  t (12)
   . A.l 4

Downloaded from https://asmedigitalcollection.asme.org/IDETC-CIE/proceedings-pdf/IDETC-CIE2012/45035/363/4257462/363_1.pdf by University of Michigan user on 29 August 2019

where  is the material density and M is the tip mass. By substituting these dimensionless quantities into
Assuming that the beam is vibrating in viscously damped equations (6) and (7), dropping the hats, and assuming that E.I
media and assuming that the axial damping is negligible with is constant with respect to coordinate x, and ρ.A and M are
respect to transverse damping, the virtual external work done constant with respect to time, the following equations may be
on the beam by distributed damping loads would be derived:
l w  x, t 
 We    ct  w  x, t  dx   2 w
2
(5) 4 w 1 1  w 
 
2 0  x 
0 x    u t    dx  2
x 4
1

where ct is the damping coefficient per unit length in the   x (13)
transverse direction. 2 w w  2 w  x, t  
 2  Ct 2 f 1  0
Now using the generalized Hamilton’s principle, the t t t 2
equations governing the non-linear dynamics of a beam
d 2u  t   1 1  w 
2

undergoing large in-plane motions and the related geometric 1  

u  t      dx   0

(14)
boundary conditions are obtained as follows. dt 2
 2 0  x  
 2 w  E. A   2 w
2
2  1 l  w  where u(t) is the normalized axial displacement of the tip mass
 
2 0  x 
 E . I    u l , t    dx  and
x 2  x 2  l   2
 x (6)
Al 2
 w    w   1  (15)
   . A.     M  f l  I
t  t  t  t  M
2  (16)
d 2u  l , t  E. A  1 l  w 
2
  . A.l
M   u  l , t   0   dx   0 (7)
dt 2 l  2  x   M I
1  (17)
w  x, t   . A.l Al 2
w  0, t   0 (8) ct l 2
x x 0 Ct  (18)
 . A.E.I
In equation (6), fˆ  x  is the Dirac delta function which is
The first mode of a typical system is generally the most
used to model the concentrated inertial load at x = l. Figure 2 important one. When the system is excited, most of the input
shows a schematic view of the Dirac delta function. excitation energy is injected to its first mode. Therefore, in
many practical situations, the only dominant mode in the
dynamics of dynamical systems is its first mode. Assuming this
to be the case for the system being considered, one may employ
the Galerkin projection method [4]. Accordingly, the response
of the system to an initial disturbance can be assumed to be as
follows:
  x
w  x, t   w t  (19)
 1
Here, w(t) is the transverse displacement of the beam tip.
Furthermore, φ(x) is the first linear un-damped transversal
vibrational mode of the system. φ(x) can be used as the basis
function for describing the non-linear behavior of the system.
Figure 2 Dirac delta function fˆ  x  where   0  For a beam with a tip mass, φ(x) is given by [5]:
  x   cos   x   cosh   x  
For convenience, the following dimensionless variables are
cos     cosh    (20)
introduced.
sin     sinh   
sin   x   sinh   x 
x
xˆ  (9)
l

3 Copyright © 2012 by ASME

 In this equation, β is the absolute value of the smallest c3 d1
positive root of equation (21). n  (34)
c1
1 M
1   tan     tanh     0 (21)
cos    cosh    m C1 
c2
d1 (35)
Substituting equation (19) into equation (13), multiplying it c1
by φ(x) and then integrating the resulting equation over the c4 d1
dimensionless domain, the following non-linear ordinary C2  (36)

Downloaded from https://asmedigitalcollection.asme.org/IDETC-CIE/proceedings-pdf/IDETC-CIE2012/45035/363/4257462/363_1.pdf by University of Michigan user on 29 August 2019

c1
differential equation is obtained.
c5 d1
d 2wt  dw  t  C3  (37)
2
  c2 c1    c3 c1  w  t  c1
dt dt (22)
D1  d2 (38)
  c4 c1  w  t    c5 c1  w  t  u  t   0
3

The natural frequency ωn in equation (32) is not the actual

Furthermore, by substituting equation (19) into equation frequency but instead a normalized one.
(14), the following equation is obtained for the axial
displacement of the beam tip.
3. SOLUTION PROCEDURE
d 2u  t 
 1 d1  u  t    d 2 d1  w  t   0
2
(23) A beam with a tip mass with characteristics given in Table 1
dt 2 is considered.
In equations (22) and (23), ci’s 1≤ i ≤ 5 and dj’s 1≤ j ≤ 4 are
defined as follows. Table 1: Characteristics of the simulated beam and its tip mass

 
1
   x      x  f 1 dx
2 2
c 
1 2
(24) symbol definition value
0
1 Young’s Modulus
c2  Ct    x  dx
2
(25) E of elasticity of the 69 GPa
0
beam material
1 d 4  x 
c3     x  dx (26)  Density of the
0 dx 4 7800 kg m3
beam material
 d  x  
2
1 1
l 0.15 m
2 0 
c4    dx Beam’s length
2 1  dx 
(27) b Beam’s width 0.015 m
1 d 2  x 
    x dx h Beam’s thickness 0.001 m
dx 2
0
M Tip mass 0.050 kg
1 d 2  x 
c5  1    x  dx (28)
0 dx 2
In order to give the reader some insight into the order of
d1  1 (29)
magnitude of all intermediate parameters defined in the paper,
 d  x  
2
1 1 their values are compiled Table 2.
2 0 
d2   dx (30)
2 1  dx  Table 2: Values of the intermediate parameters
defined in the analysis. Parameters listed without any
In order for the coefficients of equations (22) and (23) to
units represent normalized quantities.
appear at the same order, the following dimensionless variable
is introduced. A  1.5  105 m 2 c3  13.729
  t d1 (31) I  1.25  10 12
m 4
c4  8.845  105
Substituting equation (31) into equations (22) and (23), the
m  1.755  102 kg c5  1.477  106
following equations are obtained.
d 2 w   dw    1  2.7  105 d1  1.055 105
 n2 .w    C1  C2 w  
3

d 2
d (32)  2  2.849 d 2  0.598
 C3 w   u    0 1  1.055 10 5
n  3.334  103
d 2 u     0.993 C2  0.716
 u    D1 w    0
2
(33)
d 2
c1  13.033 C3  1.196
where ωn, Ci’s (1≤ i ≤ 3) and Dj’s (1≤ j ≤ 2) are defined as D1  0.598

4 Copyright © 2012 by ASME

 In addition, the damping coefficient Ct is selected such that d 2 w  
the final damping coefficients become C1 = 0.001. For a 1  P, w   , u      n2 .w  
d 2 (39)
practical choice of dimensions, as listed in Table 1, even though
dw  
the non-normalized natural frequency is finite (equal to 37.6  P  C2 w    C3 w   u     0
3
 P.C1 2

rad/s), the normalized natural frequency  n in equation (32) is d

very small. This is simply a consequence of the fact that time is d 2u  
2  P, w   , u      u    PD1 w    0 (40)
2
normalized via equation (31) using the natural frequency of the d 2

Downloaded from https://asmedigitalcollection.asme.org/IDETC-CIE/proceedings-pdf/IDETC-CIE2012/45035/363/4257462/363_1.pdf by University of Michigan user on 29 August 2019

axial direction dynamics, given by equation (23). At this step, the independent variable τ is expanded in terms
From equations (6) and (7), it can be mathematically proven of multiple time scales T0= τ and T1= Pτ so that the first and the
that second time derivatives become
D1  C2 C3 (38) d  
This relation is corroborated by the numerical values of D1,      P   (41)
d T0 T1
C2, and C3 listed in Table 2. By studying the beam constraint
model, Awtar and Sen [13] showed that the above-mentioned d2 2 2 2
2   2      P 2 2   (42)
  2P
equivalence is an invariant characteristic and independent of d T0 T0 T1 T1
the beam’s dimensions given in Table 1. Therefore, using The solution of equations (39) and (40) are sought in the
equation (38) and (33), equation (32) may be rewritten as form
d 2 w   dw   w T0 , T1   w0 T0 , T1   P.w1 T0 , T1  (43)
 n2 .w    C1
d 2
d (38) u T0 , T1   u0 T0 , T1   Pu1 T0 , T1  (44)
d 2 u  
 C3 w   0 By substituting equations (41), (42), (43) and (44) into the
d 2 homotopy forms and equating the coefficients of like powers of
In Figure 3, the numerical solution of equation (38) and that P, the following equations are obtained.
of its corresponding linear form are compared for the initial
conditions w(0) = 0.1 and u(0) = ‒ 0.006. It is observed that P0 :
these two solutions, which represent the transverse dynamics,  2 w0 T0 , T1  (45)
do not differ appreciably. However, the non-linear terms in  n2 w0 T0 , T1   0
T 0
2
equation (33) considerably affect the axial dynamics.
 u0 T0 , T1 
2

 u0 T0 , T1   0
0.1
(46)
0.08 Solution of the inear equation
Solution of the non-linear equation
T02
0.06
P1 :
 2 w1 T0 , T1   2 w0 T0 , T1 
0.04

0.02  n2 w1 T0 , T1   2 (47)

w()

0
T02 T1T0
-0.02 w0 T0 , T1 
 C1
-0.04 T0
-0.06
 u1 T0 , T1 
2
 2 u0 T0 , T1 
 u1 T0 , T1   2
T T0 T1
-0.08 2
0 500 1000 1500 2000 2500

3000 3500 4000 4500 5000
0 (48)
 D1 w0 T0 , T1 
2

Figure 3 Comparison of the solutions of equation (38) and

its corresponding linear form Equations (45) and (46) constitute a system of linear
ordinary differential equations with constant coefficients and
In the next section, HPM is used in parallel with the their solution can be written as:
multiple time scale perturbation method to solve the non-linear
w0 T0 , T1   A1 T1  exp  I T0   A1 T1 
system of ODE’s given in equations (32) and (33). (49)
exp   I T0 
3-2. SOLUTION PROCEDURE
u0 T0 , T1   B1 T1  exp  IT0   B1 T1 
Now, HPM is utilized in parallel with the multiple time (50)
scale perturbation method to derive analytical closed form exp   IT0 
solutions for equations (32) and (33). To do so, the homotopy where A1(T1) and B1(T1) are complex functions and A1 T1  and
forms i  P, w   , u    for i = 1 and 2 are constructed as B1 T1  are the complex conjugate of A1(T1) and B1(T1)
respectively.

5 Copyright © 2012 by ASME

 Substituting equations (49) and (50) into equations (47) and solving the resulting equation, u1(T0,T1) is obtained as equation
(48), these equations can be solved easily using the theory of (65).
linear ODEs. But any particular solution of equations (47) and  D1a12 exp  C1T1 
(48), contains a secular term, T0exp(±Iω0T0) and T0exp(±IT0) u1  T0 , T1  
2
unless the coefficients of T0exp(IωT0) and T0exp(IT0) in the (65)
right hand side of equations (47) and (48) are zero respectively.  cos  2nT0  21  
 1  
Therefore the following equations have to be satisfied in order  1  4 2 

Downloaded from https://asmedigitalcollection.asme.org/IDETC-CIE/proceedings-pdf/IDETC-CIE2012/45035/363/4257462/363_1.pdf by University of Michigan user on 29 August 2019

to avoid any secular term in the response. By substituting T0 = τ into equation (63) and substituting
dA T  equations (64), (65), P = 1, T0 = τ and T1 = τ into equation (44),
2 I n 1 1  I n C1 A1 T1   0 (51)
dT1 the zero order and first order approximate solutions for w(τ) and
dB1 T1  u(τ) respectively are obtained as follows.
2I 0 (52)  1 
dT1 w    a1 exp   C1  cos n  1  (66)
For solving equations (51) and (52), it is convenient to write  2 
A1(T1) and B1(T1) in the form  D1a12 exp  C1 
u    b1 cos   1  
1 2
A1 T1   a1 T1  exp 1 T1  I  (53) (67)
2  cos  2n  21  
1  1  
B1 T1   b1 T1  exp  1 T1  I  (54)  1  4 2 
2 Figure 4 and Figure 5 compares the result of the presented
By substituting equations (53) and (54) into equations (51) analytical model with the numerical simulation for an un-
and (52), making the necessary simplifications and equating damped and a damped system respectively. It is observed that
both the real and imaginary parts of these equations with zero, when there is no internal resonance in the system, the analytical
the following equations are obtained. results well follow the numerical ones and as a result, the
da1 T1  1 presented analysis can be used to investigate the dynamics of
 C1a1 T1   0 (55)
dT1 2 flexure beams in flexure mechanisms.
0.1
d 1 T1  (a) Numerical results
0 (56) 0.08
Zero order HPM results
dT1 0.06

db1 T1  0.04

0 (57) 0.02
dT1
w()

0
d 1 T1  -0.02
b1 0 (58)
dT1 -0.04

Equations (55) to (58) can be solved consequently and the -0.06

results would be as follows. -0.08

 1  -0.1
a1 T1   a1 exp   C1T1 
0 500 1000 1500 2000 2500 3000 3500 4000
(59) 
 2  -3
x 10
1 T1   1
1
(60) (b) Numerical results
First order HPM results

b1 T1   b1 (61) 0

1 T1   1 (62) -1

Using equations (49), (50), (53), (54), (59), (60), (61) and -2
u()

(62), w0(T0,T1) and u0(T0,T1) are obtained as follows. -3

 1 
w0 T0 , T1   a1 exp   C1T1  cos nT0  1  (63) -4
 2 
u0 T0 , T1   b1 cos T0  1 
-5
(64)
As it will be seen later, a zero order approximation is -6
0 500 1000 1500 2000 2500 3000 3500 4000

sufficient for predicting the time domain behavior of w(τ), but 

for accurate prediction of the behavior of u(τ) at least a first Figure 4 Comparison of analytical results with numerical
order perturbation approximation is required. So, by simulations for an un-damped system with initial conditions
substituting equations (63) and (64) into equation (48) and w(0) =0.1 and u(0)= ‒ 0.006. (a) Normalized transverse
displacement w(τ), and (b) Normalized axial displacement u(τ)

6 Copyright © 2012 by ASME

 0.1
(a) Numerical results
Nevertheless, a closed-form solution is presented here for the
0.08
Zero order HPM results sake of completeness.
0.06
-4
0.04 2 x 10
Numerical results
(a) C1 = 0 First order HPM results
0.02
w()

0
0

-0.02 -2

Downloaded from https://asmedigitalcollection.asme.org/IDETC-CIE/proceedings-pdf/IDETC-CIE2012/45035/363/4257462/363_1.pdf by University of Michigan user on 29 August 2019

-0.04

u()
-4
-0.06

-0.08 -6
0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000

-3

x 10 -8
1
(b)
0 -10
350 400 450 500 550 600

-1
Numerical results
First order HPM results

-4
x 10
1
-2
(b) C1 = 0.001 Numerical results
u()

0
First order HPM results
-3 -1

-2
-4
-3

-4

u()
-5
-5
-6 -6
0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000
 -7

Figure 5: Comparison of the analytical results with numerical -8

simulations for a damped system with C1 = 0.001 and initial -9

conditions w(0) =0.1 and u(0)= ‒ 0.006. (a) Normalized -10

400 450 500 550 600 650 700

transverse displacement w(τ), and (b) Normalized axial 

displacement u(τ) Figure 6: Zoomed view of the normalized axial displacement
(a) Un-damped system, and (b) Damped system, C1 = 0.001
In Figure 4 (b) and Figure 5 (b), the normalized axial
displacement is composed of a large-amplitude, low-frequency In the case ωn ≈ 1/2, the nearness of ωn to 1/2 can be
component and a small-amplitude, high-frequency component. expressed as follows.
The former is due to the effect of the transverse vibration of the 1
beam on its axial vibration while the latter is the direct  n   P (68)
2
consequence of the large axial stiffness of the beam. Zoomed which leads to
views of the latter component are shown in Figure 6. At higher
 IT 
values of normalized time, the difference between the results is exp   I nT0   exp   0  exp   I  T1  (69)
due to a slight difference between the frequency of the  2 
numerical and analytical solutions. exp   IT0   exp  2 I nT0  exp  2 I  T1  (70)
By using equations (49) and (50) and substituting equations
The case n  1 2 (69) and (70) into the right hand side of equations (47) and (48)
Next, one may mathematically analyze the case when ωn is respectively, the terms capable of producing secular terms are
close to 1/2, which represents a condition of internal resonance obtained as
in the system of non-linear ordinary differential equations given dA T 
by (32) and (33). It is important to note that this normalized 2 I n 1 1  IC1 A1 T1  n  0 (71)
dT1
value of n actually corresponds to a natural frequency of 5874
rad/sec. At such large frequencies, the approximations made in dB1 T1 
 D1 A1 T1  exp  2 I  T1   0
2
2I (72)
deriving equations (32) and (33) break down. To accurately dT1
analyze the dynamics of the system in this frequency range,
Substituting equations (53) and (54) into equations (71) and
several transverse and axial modes will need to be considered
(72) and equating the real and imaginary parts of both
and the axial kinetic energy cannot be ignored. Therefore,
equations with zero gives
solving the above equations for the case when ωn ≈ 1/2 is a
da T  1
strictly mathematical exercise and of little physical relevance. n 1 1  n C1a1 T1   0 (73)
dT1 2

7 Copyright © 2012 by ASME

 d 1 T1  coupled non-linear ordinary differential equations, using the
0 (74) first transverse mode of the corresponding linearized system.
dT1
To solve these equations, the homotopy perturbation method
db1 T1  1 was utilized in conjunction with the method of multiple time
 D1a1 T1  sin  21 T1 
2

dT1 4 (75) scales. These equations reveal an internal resonance state in the
system arising from the coupling between the transverse and
 1 T1   2 T1   0 axial directions. The time domain response of the system
d 1 T1 

Downloaded from https://asmedigitalcollection.asme.org/IDETC-CIE/proceedings-pdf/IDETC-CIE2012/45035/363/4257462/363_1.pdf by University of Michigan user on 29 August 2019

1 around and far from this internal resonance state was
b1 T1   D1a1 T1  cos  21 T1 
2
investigated and analytical zero-order and first order
dT1 4 (76) expressions for the beam tip displacement have been presented.
 1 T1   2 T1   0 Comparison of the derived analytical results with the numerical
Equations (73) to (76) can be transformed into an simulations was used to validate the accuracy of the presented
autonomous system by letting closed-form solution approach. While the approximations and
assumptions made in this solution approach are justified by
 1 T1   21 T1   1 T1   2 T1 (77)
physical and mathematical arguments, the final results are yet
The results are to be validated via experimental measurements. In addition to
da1 T1  1 experimental validation, our on-ongoing research effort
  C1a1 T1  (78)
dT1 2 includes extending the above analysis approach to more
complex flexure modules and mechanisms.
d  1 T1  d  T1 
 2  (79)
dT1 dT1 REFERENCES
db1 T1  1 [1] M. Mojahedi, H. Moeenfard, and M. T. Ahmadian,
  D1a1 T1  sin   T1  
2
(80)
dT1 4 "Nonlinear free vibration of simply supported beams
considering the effects of shear deformation and rotary
d 1 T1 
1
b1 T1   D1a1 T1  cos   T1  
2
(81) inertia, a homotopy perturbation approach," International
dT1 4 Journal of Modern Physics B, vol. 25, pp. 441-455, 2011.
By solving equations (78) to (81), one can find a1(T1), [2] S. T. Smith, Flexures: elements of elastic mechanisms.
γ1(T1), b1(T1) and β1(T1). Gordon & Breach, 2000.
After eliminating secular terms, the solution of equations [3] W. Su and C. E. S. Cesnik, "Strain-based geometrically
(47) and (48) becomes nonlinear beam formulation for modeling very flexible
w1 T0 , T1   0 (82) aircraft," International Journal of Solids and Structures,
vol. 48, pp. 2349-2360, 2011.
1
u1 T0 , T1    D1 a1 T1 
2
(83) [4] A. H. Nayfeh and D. T. Mook, Nonlinear oscillations.
2 Wiley, 1979.
Using equations (63), (64), (82) and (83), the final solution [5] S. S. Rao , Vibration of continuous systems, Wiley, 2007.
for w(τ) and u(τ) would be [6] M. R. M. Crespo Da Silva, "Non-linear flexural-flexural-
 1  torsional-extensional dynamics of beams—I.
w    a1 exp   C1 1  cos nT0  1  (84)
 2  Formulation," International Journal of Solids and
1 Structures, vol. 24, pp. 1225-1234, 1988.
u     D2 a12 exp  C1 1   b1   cos   1    (85) [7] M. R. M. Crespo Da Silva, "A reduced-order analytical
2
model for the nonlinear dynamics of a class of flexible
multi-beam structures," International Journal of Solids
4. CONCLUSION
and Structures, vol. 35, pp. 3299-3315, 1998.
The importance of analytically studying the dynamics of [8] A. H. Nayfeh, "Nonlinear transverse vibrations of beams
flexure mechanisms to better inform their design and with properties that vary along length," Journal of the
optimization is well-recognized. However, such an Acoustical Society of America, vol. 53, pp. 766-770, 1973.
investigation is complicated by the fact that in many [9] C. L. Zaretzky and M. R. M. C. da Silva, "Experimental
applications, geometric non-linearities in flexure mechanics Investigation of Non-Linear Modal Coupling in the
play an important role in the dynamics of the system. As a Response of Cantilever Beams," Journal of Sound and
starting point in a broader investigation, we considered a simple Vibration, vol. 174, pp. 145-167, 1994.
beam flexure with a tip mass in this paper and analyzed its large [10] J. W. Hijmissen and W. T. van Horssen, "On the weakly
amplitude in-plane oscillations. Euler beam approximation was damped vibrations of a vertical beam with a tip-mass,"
made while retaining the geometric non-linearity associated Journal of Sound and Vibration, vol. 310, pp. 740-754,
with beam arc-length conservation, resulting in a set of coupled 2008.
non-linear partial differential equations of dynamics. The
Galerkin's projection approach was employed to reduce these to

8 Copyright © 2012 by ASME

[11] L. D. Zavodney and A. H. Nayfeh, "The non-linear
response of a slender beam carrying a lumped mass to a
principal parametric excitation: Theory and experiment,"
International Journal of Non-Linear Mechanics, vol. 24,
pp. 105-125, 1989.
[12] H. Moeenfard, M. Mojahedi, and M. T. Ahmadian, "A
homotopy perturbation analysis of nonlinear free

Downloaded from https://asmedigitalcollection.asme.org/IDETC-CIE/proceedings-pdf/IDETC-CIE2012/45035/363/4257462/363_1.pdf by University of Michigan user on 29 August 2019

vibration of Timoshenko microbeams," Journal of
Mechanical Science and Technology, vol. 25, pp. 557-
565, 2011.
[13] S. Awtar and S. Sen, "A Generalized Constraint Model for
Two-Dimensional Beam Flexures: Nonlinear Strain
Energy Formulation," Journal of Mechanical Design, vol.
132, pp. 081009-11, 2010.

9 Copyright © 2012 by ASME