Professional Documents
Culture Documents
of Cantilevers Subject
to Motion Constraints
Hamed Farokhi The nonlinear extremely large-amplitude oscillation of a cantilever subject to motion
Department of Mechanical and constraints is examined for the first time. In order to be able to model the large-
Construction Engineering, amplitude oscillations accurately, the equation governing the cantilever centerline rota-
1
Corresponding author. 2 Model Development
Contributed by the Applied Mechanics Division of ASME for publication in the
JOURNAL OF APPLIED MECHANICS. Manuscript received July 22, 2018; final manuscript
The system under consideration, as shown in Fig. 1, is a cantile-
received November 7, 2018; published online December 17, 2018. Assoc. Editor: ver under base excitation subjected to motion constraints. The
George Haller. motion of the cantilever is analyzed in a Cartesian coordinate
The relations in Eq. (1) allow for deriving the rotational equation ðL 2
1 @hð x; tÞ
of motion of the cantilever. The advantage of the rotational equa- KE ¼ qI dx
2 0 @t
tion of motion over the transverse equation of motion is that it is
ðL ðx 2
capable of predicting very-large-amplitude oscillations even when 1 @hð x; tÞ
þ qA sin hð x; tÞ dx dx
the tip angle is larger than p/2. In what follows, the kinetic and 2 0 0 @t
potential energies as well as the work of damping are obtained in ð L ð x 2
1 @hð x; tÞ
terms of the rotation angle. þ qA cos hð x; tÞ dx þ w0 x cosðx tÞ dx
The axial strain of the cantilever under inextensibility assump- 2 0 0 @t
tion can be formulated as [19] (7)
@hð x; tÞ in which q denotes the mass density and A stands for the cross-
e ¼ z (3) sectional area.
@x
The motion constraints are modeled via linear springs with very
Based on the Kelvin–Voigt energy dissipation mechanism, the large stiffness coefficients. Hence, the work of the motion con-
axial stress can be expressed as straints (both sides) can be mathematically represented as
where H denotes the Heaviside function, sgn stands for the sign function, dD represents the Dirac Delta function, and the vertical bars,
i.e., jj, indicate the absolute value.
Employing generalized Hamilton’s principle
ð t2
ðdKE dP þ dWv þ dWc Þdt ¼ 0 (9)
t1
one can derive the equation governing the rotational motion of the cantilever under transverse base-excitation and subject to motion
constraints as
(12)
Wiring the equation of motion in dimensionless form allows for a general parametric analysis [21]. In order to be able to examine the
nonlinear dynamical characteristics of the cantilever numerically, Eq. (12) is reduced to a discretized set nonlinear ordinary differential
equations utilizing the Galerkin technique. To this end, the cantilever centerline rotation is expanded as
XN
hðx; tÞ ¼ pm ðtÞwm ðxÞ;
m¼1 (13)
wm ðxÞ ¼ ½sinhðbm xÞ þ sinðbm xÞ vm ðcoshðbm xÞ cosðbm xÞÞ
where pm ðtÞ denotes the mth generalized coordinate for the rotational motion and vm ¼ ½coshðbm Þ þ cosðbm Þ=½sinhðbm Þ þsinðbm Þ; bm
is the mth root of the equation 1 þ coshðxÞ cosðxÞ ¼ 0.
Substitution of Eq. (13) into the equation of motion, i.e., Eq. (12), and application of the Galerkin method results in
ð1 ! ð1 ! ð1 !
1 XN X N XN
00 00
w w dx p€n wm wn dx pn gd wm wn dx p_ n
a n¼1 0 m n n¼1 0 n¼1 0
8 !ð ð 2 !!2 ! ! !3 9
ð1 < XN x x XN XN 2 X N XN =
4 @ @
wm sin pn ðtÞwn ð xÞ pn ðtÞwn ð xÞ cos pn ðtÞwn ð xÞ þ 2 pn ðtÞwn ð xÞ sin pn ðtÞwn ð xÞ 5dxdx dx
0
: 1 0 @t n¼1 @t n¼1 ;
n¼1 n¼1 n¼1
8 !ð 0 2 ! !2 !
ð1 < X ðx
N x
@ 4 @ X N XN
þ wm cos pn ðtÞwn ð xÞ w0 xe sinðxe tÞ þ pn ðtÞwn ð xÞ sin pn ðtÞwn ð xÞ
0
: 1 0 @t n¼1
n¼1 n¼1
! !#! )
@2 X N X N
2 pn ðtÞwn ð xÞ cos pn ðtÞwn ð xÞ dxdx dx
@t n¼1 n¼1
8 !ð " ! !
ð1 < X ð x X
N 1 N
þ wm K1 cos pn ðtÞwn ð xÞ dD ð xxc Þ H sin pn ðtÞwn ð xÞ dx g0
0 : n¼1 x 0 n¼1
ðx ! ! 0 ð ! !# )
X N x X N
sgn sin ð Þw
pn t n ð xÞ dx @ ð Þw
pn t n ð xÞ dx g0 dx dx ¼ 0; m ¼ 1;2;…;N
sin
0 n¼1 0 n¼1
(14)
In order to ensure accurate results even at very large-amplitude cantilever subject to motion constraint is examined and discussed
oscillations, the trigonometric functions of h are kept intact as in Sec. 3.2. The numerical results are obtained for a cantilever of
they are integrated from 0 to 1, which results in substantially large dimensions h ¼ 1 mm, b ¼ 5 h, and L ¼ 200 h. It should be men-
equations of motion. The presence of such large equations and tioned here once again that all the reported values in this section
nonlinear inertial terms makes the numerical simulation even are dimensionless as defined in Eq. (11) with the asterisk notation
more challenging. In this study, five generalized coordinates are being dropped for convenience.
retained in the rotational motion series expansion, resulting in a
5DOF system which ensures converged results; in other words,
3.1 Nonlinear Response of the Cantilever Without Motion
the contribution of higher modes (more than 5) to the cantilever
Constraint. In the absence of a motion constraint, the cantilever
response is small enough that can be safely neglected. After
is free to undergo large-amplitude oscillations. The transverse
obtaining the amplitude of the rotational motion, the longitudinal
amplitude of oscillation depends of course on the base-excitation
and transverse displacements are calculated using Eq. (1). The
amplitude. Figure 2 shows the nonlinear frequency-amplitude
5DOF discretized system of equations is solved numerically via
plots of the cantilever undergoing large-amplitude oscillations due
the use of a pseudo-arclength continuation technique; addition-
to transverse base-excitation. For this case, w0 ¼ 0.018 and
ally, a stability analysis was conducted using the Floquet theory.
gd ¼ 0.0057. Subfigures (a, b) show the cantilever tip transverse
The employed method used is capable of capturing all types of
and longitudinal displacements, while subfigure (c) shows the can-
periodic motions as well as detecting different bifurcations [22].
tilever tip angle. It is worth mentioning that solving the model
developed in Sec. 2 results in the value of centerline angle. Hav-
ing obtained the centerline angle, the transverse and longitudinal
3 Numerical Results displacements are calculated via using Eq. (1). As seen in Fig. 2,
The nonlinear large-amplitude oscillations of the cantilever the system shows a weak nonlinear hardening behavior. Further-
with and without motion constraints are studied in this section. In more, subfigure (c) shows that the tip angle maximum value is
order to better illustrate the capability of the developed model in more than p/2, indicating that the tip of the cantilever bends more
capturing very large-amplitude oscillations of the cantilever, the than 90 deg at large-amplitude oscillations. This is the main
motion constraint is initially removed and the nonlinear resonant advantage of the model developed in this study, as it overcomes
response of the system under base-excitation is examined; this is the limitation of the classical nonlinear transverse model of the
discussed in Sec. 3.1. Then, the detailed nonlinear response of the cantilever (i.e., the tip angle being between p/2 and p/2). To
Fig. 4 Comparison between the frequency-amplitude plots of the constrained cantilever to those of a
cantilever with no constraint: (a) maximum tip transverse displacement, (b) maximum tip longitudinal dis-
placement, and (c) maximum tip angle. The solid line showing stable solution, while the dotted line show-
ing unstable one. g0 5 0.03, K1 5 2.0 3 104, and w0 5 0.018.
Fig. 5 Details of the motion of the system of Fig. 4 at xe/x1 5 0.9775 (i.e., in the resonance region before hitting
the motion constraint), showing, respectively, the time trace and phase-plane portrait of: ((a), (b)) tip transverse
displacement, ((c), (d)) tip longitudinal displacement, and ((e), (f)) tip angle. tn denotes normalized time with
respect to the period of the oscillation.
reveals a value of for the 0.0790 unconstrained system and a value before and after hitting the motion constraint. As seen in Fig. 5,
of 0.1292 for the constrained, i.e., an increase of 63.5%. the cantilever displays smooth periodic response in the region
The detailed dynamical characteristics of the constrained canti- where there is not contact between the cantilever and the con-
lever at two different excitation frequencies, xe/x1 ¼ 0.9775 and straint. However, as seen in Fig. 6, the dynamical characteristics
1.1030, are illustrated in Figs. 5 and 6, respectively. xe/ of the cantilever in the region where it hits the constraint are
x1 ¼ 0.9775 is in the resonance region before hitting the motion totally different, showing nonsmooth vibrational behavior.
constraint, while xe/x1 ¼ 1.1030 is in the region where the canti- It would be interesting to compare the response of the cantile-
lever is hitting the motion constraint. The goal here is to compare ver subject to motion constraint at both sides to that subject to
the time traces and phase-plane portraits of the cantilever motion motion constraint only at one side; the comparison between the
Fig. 6 Details of the motion of the system of Fig. 4 at xe/x1 5 1.1030 (i.e., in the resonance region after hitting
the motion constraint), showing, respectively, the time trace and phase-plane portrait of ((a), (b)) tip transverse
displacement, ((c), (d)) tip longitudinal displacement, and ((e), (f)) tip angle. tn denotes normalized time with
respect to the period of the oscillation.
two cases is shown in Fig. 7. As seen, when the cantilever motion important to note that asymmetric motions can arise even in oscil-
is constrained from both sides, the change in the slope of the reso- lators with perfectly symmetric properties.
nance response after hitting the constraint is larger compared to As mentioned in Sec. 2, in this study, the motion constraints are
the case with one-side constraint. Furthermore, the cantilever with modeled as linear springs located at distance xc from the clamped
constraints at both sides shows wider resonance frequency band base and at both sides of the cantilever with a gap (clearance) g0.
compared to the cantilever with a constraint only at one side. It To this end, the spring stiffness coefficient should be large enough
should be noted that for the rest of the figures in this paper, it is to ensure that the spring is a realistic representative of the motion
assumed that the cantilever is constrained at both sides. It is constraint. The effect of the spring stiffness coefficient on the
Fig. 8 Effect of the motion constraint stiffness on frequency-amplitude plots of the constrained cantile-
ver: (a) maximum tip transverse displacement and (b) maximum tip longitudinal displacement. The solid
line showing stable solution, while the dotted line showing unstable one. g0 5 0.03 and w0 5 0.018.
Fig. 9 Effect of the motion constraint gap-width on the frequency-amplitude plots of the constrained
cantilever: (a) maximum tip transverse displacement and (b) maximum tip longitudinal displacement.
The solid line showing stable solution, while the dotted line showing unstable one. K1 5 2.0 3 104 and
w0 5 0.018.
Fig. 11 Effect of the position of the motion constraint on the frequency-amplitude plots of the cantilever:
(a) maximum tip transverse displacement and (b) maximum tip longitudinal displacement. The solid line
showing stable solution, while the dotted line showing unstable one. g0 5 0.03, K1 5 2.0 3 104, and
w0 5 0.018.
frequency-amplitude diagrams of the constrained cantilever is occurs at smaller excitation frequencies. Furthermore, as the gap
shown in Fig. 8; the frequency-amplitude plot of the uncon- width is decreased, the resonance frequency band increases and
strained cantilever is plotted as well for better comparison. As the peak amplitude occurs at larger excitation frequencies.
seen, due to increased stiffness coefficient, the change in the slope Figure 10 shows the effect of the base-excitation amplitude on
of the frequency-response becomes sharper when the cantilever the nonlinear frequency-amplitude response of the constrained canti-
hits the motion constraint. Additionally, the frequency band of lever. It is seen that at a small base-excitation amplitude of 0.005, the
resonance increases with increasing stiffness coefficient. It is cantilever does not hit the motion constraint, hence displaying a
interesting to note that the resonance response of the constrained smooth response. At larger base-excitation amplitudes, on the other
cantilever changes dramatically when K1 is increased from hand, the cantilever does hit the motion constraint causing a sharp
2.0 102 to 2.0 103; however, a smaller change in the resonance change in the slope of the resonance response. As seen, by increasing
response is observed as K1 is increased from 2.0 103 to the base-excitation amplitude, the resonance frequency band increases
2.0 104. Increasing the stiffness coefficient even further, to and the peak amplitude occurs at larger excitation frequencies.
6.0 104, causes only very slight change in the resonance The frequency-amplitude plots of the constrained cantilever for
response, indicating convergence. Hence, K1 ¼ 2.0 104 can be two locations of the motion constraint are illustrated in Fig. 11.
selected as the stiffness of the constraint since it gives a reasona- As seen, as the motion constraint is moved slightly to the right
ble approximation of the motion constraint while allowing for (i.e., further away from the base), the cantilever hits it at smaller
smooth numerical simulations. excitation frequencies. Furthermore, the peak amplitude occurs at
The effect of the gap width (clearance) of the motion constraint larger excitation frequencies as the constraint is moved to the right
on the nonlinear resonance response of the constrained cantilever on the x-axis.
is depicted in Fig. 9. As seen, as the gap width is decreased, the The effect of the material damping coefficient gd on the reso-
cantilever hits the motion constraint at smaller oscillation ampli- nance response of the constrained cantilever is shown in Fig. 12.
tudes; as a result, the change in slope of the resonance response As seen, as gd is increased, the multivalued resonance region
becomes wider and the peak amplitude occurs at smaller excita- results in widened resonance frequency band. Furthermore, it is
tion frequencies. It is interesting to note that increasing gd from shown that as the clearance between the motion constraint and the
0.006 to 0.007 causes a further reduction in the amplitude com- cantilever is increased, the change in slope of the frequency-
pared to the case when gd is increased from 0.005 to 0.006. response, due to hitting the motion constraint, occurs at smaller exci-
tation frequencies. Examining the effect of the base-excitation ampli-
4 Conclusions tude on the nonlinear resonance response of the system shows that at
small base-excitation amplitudes the cantilever does not hit the
The nonlinear large-amplitude oscillation of a cantilever under motion constraint; as the base-excitation amplitude is increased to
base excitation has been examined in the presence of motion- larger amplitudes, the cantilever hits the motion constraint causing a
limiting constraints on both sides of the cantilever. Hamilton’s change in the slope of the frequency-amplitude. Additionally, due to
principle is employed to derive the equation governing the rota- increased base-excitation amplitude, the resonance frequency band
tional motion of the cantilever centerline, based on the increases and the peak amplitude occurs at larger excitation frequen-
Euler–Bernoulli beam theory and centerline inextensibility assump- cies. If the motion constraint is moved further away from the
tion. The equation of motion is discretized through use of the Galer- clamped end, the cantilever hits it at smaller excitation frequencies
kin technique; the trigonometric functions in the equation of and the resonance peak occurs at larger excitation frequencies.
motion are kept intact to ensure accurate results. The discretized
equations are solved making use of a continuation technique, while
examining the effect of various system parameters.
Appendix: Verification and Convergence
It is shown that unlike other nonlinear models of the cantilever In order to verify the model and numerical results of the present
(based on transverse motion), the present model can predict large- study, a nonlinear static analysis is conducted on a cantilever with-
amplitude oscillations even when the tip angle becomes larger out a stopper and the obtained results are compared to three-
than p/2. It is shown that the presence of a motion constraint dimensional (3D) finite element analysis (FEA) results obtained
reduces the amplitude of the oscillation and widens the band of using ABAQUS/CAE. In order to examine very large-amplitude dis-
resonance frequency vibration. Examining the effect of the stiff- placements, the cantilever is assumed to be under a concentrate
ness of the motion constraint revealed that, as the stiffness is tip load, always perpendicular to the beam. For such a system,
increased, the change in the slope of the frequency-response the nonlinear dimensionless static equation of motion can be
becomes sharper when the cantilever hits the motion constraint; this derived as
Fig. 13 Comparison between the cantilever tip deflections under perpendicular tip load: (a) transverse
deflection and (b) longitudinal deflection. Solid line: present study; symbols: 3D FEA.
References
[1] Liu, H., Lee, C., Kobayashi, T., Tay, C. J., and Quan, C., 2012, “Piezoelectric
Fig. 15 Strain distribution as a function of x at z 5 2h/2 for MEMS-Based Wideband Energy Harvesting Systems Using a Frequency-
cantilever under static load fs 5 7.68 (the case with the largest Up-Conversion Cantilever Stopper,” Sens. Actuators A: Phys., 186, pp.
242–248.
force in Fig. 14)
[2] Huicong, L., Chengkuo, L., Takeshi, K., Cho Jui, T., and Chenggen, Q., 2012,
“Investigation of a MEMS Piezoelectric Energy Harvester System With a
Frequency-Widened-Bandwidth Mechanism Introduced by Mechanical
" ð1 ð1 # Stoppers,” Smart Mater. Struct., 21(3), p. 035005.
@2h [3] Moon, F. C., and Shaw, S. W., 1983, “Chaotic Vibrations of a Beam With Non-
þ fs cos h ðdD ð x 1Þcos hÞdx þ sin h ðdD ð x 1Þsin hÞdx Linear Boundary Conditions,” Int. J. Non-Linear Mech., 18(6), pp. 465–477.
@x2 x x [4] Shaw, S. W., and Holmes, P. J., 1983, “A Periodically Forced Piecewise Linear
¼0 (15) Oscillator,” J. Sound Vib., 90(1), pp. 129–155.
[5] Choi, Y. S., and Noah, S. T., 1988, “Forced Periodic Vibration of Unsymmetric
in which fs ¼ fL2 =ðEIÞ with f being the forcing amplitude Piecewise-Linear Systems,” J. Sound Vib., 121(1), pp. 117–126.
[6] Heiman, M. S., Bajaj, A. K., and Sherman, P. J., 1988, “Periodic Motions and
(dimensional). Bifurcations in Dynamics of an Inclined Impact Pair,” J. Sound Vib., 124(1),
Using the same discretization procedure and numerical tech- pp. 55–78.
nique introduced in Sec. 2, the nonlinear static response of the [7] Shaw, S. W., and Rand, R. H., 1989, “The Transition to Chaos in a Simple
cantilever is obtained numerically. The cantilever nonlinear static Mechanical System,” Int. J. Non-Linear Mech., 24(1), pp. 41–56.
[8] Li, G. X., Rand, R. H., and Moon, F. C., 1990, “Bifurcations and Chaos in a
deflection is also obtained using ABAQUS/CAE, using continuum Forced Zero-Stiffness Impact Oscillator,” Int. J. Non-Linear Mech., 25(4), pp.
shell elements. The results obtained by the present study are com- 417–432.
pared to those obtained through 3D FEA and plotted in Figs. 13 [9] Natsiavas, S., 1990, “On the Dynamics of Oscillators With Bi-Linear Damping
and 14. The transverse and longitudinal displacements of the can- and Stiffness,” Int. J. Non-Linear Mech., 25(5), pp. 535–554.
[10] Natsiavas, S., 1993, “Dynamics of Multiple-Degree-of-Freedom Oscillators
tilever tip are shown in Fig. 13 for the two models. As seen, the With Colliding Components,” J. Sound Vib., 165(3), pp. 439–453.
model developed in this study predicts almost the same displace- [11] Nigm, M. M., and Shabana, A. A., 1983, “Effect of an Impact Damper on a
ments as the nonlinear 3D FEA. Additionally, the deformed Multi-Degree of Freedom System,” J. Sound Vib., 89(4), pp. 541–557.