52nd AIAA/ASME/ASCE/AHS/ASC Structures, Structural Dynamics and Materials Conference<BR> 19th AIAA 2011-2083

4 - 7 April 2011, Denver, Colorado

On Snap-Through Buckling

R. Wiebe∗ and L.N. Virgin †

Duke University, Durham, NC, 27708

I. Stanciulescu‡
Rice University, Houston TX, 77005

S.M. Spottswood§
Air Force Research Laboratory, Dayton OH, 45433

Snap-through buckling can reduce the life-span of structural systems such as aircraft
surface paneling. This is envisioned to be a specific problem in hypersonic vehicles in which
severe thermal loading and acoustic excitation conspire to create an especially hostile envi-
ronment for structural elements. A shallow arch, and two simplified link models are used to
investigate the mechanisms of snap-through buckling from a fundamental, or phenomeno-
logical, standpoint. The complexities introduced by modal interactions are introduced and
a method for identifying snap-through buckling is developed.

I. Introduction
In order to meet ever increasing performance demands modern aircraft are being designed lighter than
ever before. This weight efficiency usually comes at the expense of structural mass, leading to more slender
structural components. Slender structures are however susceptible to vibration and instabilities, particularly
buckling, which can occur well before strength limits are reached.
Snap-through buckling is a particular type of buckling where a structure snaps from one state to another
remote state. Such large deflections pose a hazard since they can cause fatigue.1, 2 The two primary types
of snap-through buckling, post buckled/bifurcated snap-through and limit point snap-through, are shown in
Figure 1.3
Postbuckled snap-through occurs in axially loaded structures, where a structure snaps between the two
(often) symmetric buckled states under some perturbation. Figure 2 shows the cross sections and center
displacement time series of a buckled plate before and after snap-through when exposed to a noisy lateral
sound pressure, F .4 This can also occur in aircraft surface panels that may buckle due to thermal loading,5–7
also shown in Figure 2.
Limit point buckling occurs in structures where the stiffness decreases (with increasing loading) to a
vanishing point, called the limit point. At the limit point the structure jumps to the remaining stable
equilibrium. The canonical example of this form of snap-through is the shallow arch. Figure 3 shows two
co-existing stable equilibria for an elastica shallow arch at a load slightly below the limit point. Similar
to axially buckled plates and columns, small external perturbations cause the arch to jump back and forth
between these two states.
∗ Doctoral Student, Department of Civil and Environmental Engineering, Box 90287 Hudson Hall, Durham, NC 27708,

Student AIAA Member.

† Professor, Department of Mechanical Engineering and Materials Science, Box 90300 Hudson Hall, Durham, NC 27708,

Senior AIAA Member.

‡ Assistant Professor, Department of Civil and Environmental Engineering, 6100 Main St., 208 Ryon Lab, MS 318, Houston,

TX 77005 , Senior AIAA member.

§ Senior Aerospace Engineer, Structural Sciences Center, Air Force Research Laboratory, Wright-Patterson Air Force Base,

Dayton, Ohio 45433, Senior AIAA member.

1 of 9

American
Copyright © 2011 by the American Institute of Aeronautics and Institute
Astronautics, Inc. All of Aeronautics
rights reserved. and Astronautics
 unstable
Load stable

Bifurcation point
Limit point

Deflection

Figure 1. Typical nonlinear force-displacement relationships with co-existing equilibria.

P P

δe
F
−δe δe
F F
t

−δ e

Time

Figure 2. A panel that exhibits snap-through in its postbuckled state.

A key aspect of both of these forms of buckling is the existence of an unstable equilibrium between
the stable states. Although not ordinarily realizable in an experimental (or even simulation) context, these
unstable equilibria have a profound influence on global behavior, and for example, define the basins of
attraction for the stable states, i.e., given co-existing states at the same load level means that the stability of
an equilibrium is confined to perturbations about it’s local neighborhood. The static equilibria, both stable
and unstable, are therefore of great importance in studying the dynamics of the systems. Continuous systems
such as aircraft panels however introduce difficulties in dimensionality as there are complex interactions
between mode shapes with the possibility of several co-existing stable and unstable equilibria. A shallow
arch and two link models are therefore used below as a starting point to investigate these complexities.

II. Shallow Arch

Shallow arches typically snap-through in a symmetric mode shape.8–12 This makes it relatively easy to ob-
tain the force-displacement curve via a displacement control method. Figure 4 shows the force-displacement
(displacement measured at center) curve of the arch in Figure 3 under a point force at the center obtained
by sweeping through the center displacement. The arch is relatively shallow with end angles of 14 degrees
above horizontal and a total length of 305mm, which ensured that the symmetric snap-through mode was
preferred. By controlling displacement, however, one loses the ability to measure the stability of the state.
However, as expected, it was observed that the states on the down-sloping portion of the curve in the upper
part of Figure 4 (between the dashed red lines) were unstable, i.e., once disconnected from the displacement
control mechanism they jumped to a remote state.
Another indicator of stability is the frequency of oscillation about an equilibria. As an equilibrium path
begins to lose stability its natural frequency decays to zero.3, 13 The lower part of Figure 4 shows the natural
frequencies (of small amplitude oscillations) obtained using a laser vibrometer about the equlibria shown in

2 of 9

American Institute of Aeronautics and Astronautics

 Figure 3. Experimental shallow arch in pre- and post snap-through configurations.

4
Force
(N) 3

-5 5 10 15 20 25 30
-1 Displacement (mm)

50
Natural
Frequency 40
(rad/s)
30

20

10

0 0 5 10 15 20 25 30
Displacement (mm)

Figure 4. Experimental force-displacement data and corresponding natural frequencies for a shallow arch.

the upper part, clearly indicating the stability switch at the turning points on the force-displacement curve.
Of course the natural frequency cannot be determined in the unstable region, as the system response is
divergent not oscillatory.
The force-displacement curve provides some insight into the global dynamics and snap-through of the
system. The range over which snap-through can occur is readily visible from the plot as the range over which
multiple equlibria exist. Another, albeit more abstract, statement may be made about the likelihood of snap-
through from the distance between the stable and unstable equilibria. The global dynamics of the shallow
arch however prove difficult to capture due to the complexities of the continuous model, and detract from
the goal of investigating snap-through buckling. A one-dimensional link model arch is therefore introduced
in the following section as a simplified model which preserves the underlying snap-through susceptible force-
displacement relationship.
Another shortcoming of the above analysis is that the displacement control method used to create Figure 4
is limited in application since it essentially forces the system into a specified mode shape. Deep arches and
two-dimensional systems such as curved panels may buckle or snap-through in many different mode shapes.
Therefore this approach is not suitable for investigating the interaction between mode shapes. To overcome
this a two link model is used to investigate the modal interactions of higher dimensional systems.

3 of 9

American Institute of Aeronautics and Astronautics

 III. Single Degree-of-Freedom Link Model Arch
Snap-through buckling in an arch or buckled panel may be investigated, at least qualitatively, by a single
degree-of-freedom (SDOF) link model arch. Figure 5 shows such a link model arch where M is mass, L is
link length, k is spring stiffness, P is external force, y0 is the height at which the springs are unstretched,
and y is the measured position of the mass. The model is reduced to a SDOF system as the lateral motion
of the mass is held at zero, which must be true in the static case. The effect of gravity is not included in
the analysis, i.e., the implicit assumption is that the system is effectively horizontal: this was done to match
with future experiments in which the system will be placed in a horizontal test bed. Therefore the mass has
only an inertial effect. The potential energy of the system is given by
p p 2
V (y) = k L2 − y0 2 − L2 − y 2 + P y. (1)

The stationary points of the potential energy represent the equilibria and are thus given by
p !
L 2 − y0 2
Pequil. = 2k 1 − p y. (2)
L2 − y 2

L M L

k k
y0, y θ θ0

Figure 5. Schematic of SDOF link model arch.

The equation of motion for the system, assuming a small amount of linear viscous damping (β), under
time varying external force P (t) is given by
p !
L 2 − y0 2
M ÿ + β ẏ − 2k 1 − p y = −P (t). (3)
L2 − y 2
Linearizing the stiffness term in Eq. (3) yields the linear undamped natural frequency in the neighborhood
of the equilibria as given by
p !
2k L 2−y 2
0
ω0 2 = −1 . (4)
M (L2 − y 2 )3/2
Figure 6 shows plots of the force-displacement
√ curve and the square of the natural frequency for M =
1, L = 1, k = 1, β = 0.1, and y0 = 1/ 2 which corresponds to a 45 degree initial angle. Again, as expected
the natural frequency vanishes as the equilibria become unstable. Furthermore, a negative squared natural
frequency indicates a divergent local response in the unstable region.
Large oscillations about and between the coexisting stable equilibria are more complex. In fact, for an
oscillating force the underlying force-displacement is constantly changing. But, it is possible to distinguish
between trajectories that snap-through from those that do not using energy.

A. Identifying Snap-Through
Consider a harmonic external forcing P (t) = P0 cos (Ωt + φ0 ), where Ω is the forcing frequency, and φ0 is
the initial forcing phase. For the arch to snap-through it is required that this external forcing add sufficient
energy to the system. One anticipates that the average total energy of the system, i.e. kinetic energy of the
mass plus strain energy in the springs will be much larger for trajectories that traverse across the unstable
equilibrium than for those that oscillate about a single equilibrium. Figure 7 shows the average total energy

4 of 9

American Institute of Aeronautics and Astronautics

 P (a)
0.4 0.5
P (c)
0.2
y 0.0

-1.0 -0.5 0.5 1.0

-0.2
-0.5
0.5
-0.4
V
0.0
1.0 Ω 20 (b)

0.5 -0.5
-1.0
y -0.5
-1.0 -0.5 0.5 1.0 0.0
0.5
-0.5
y 1.0

-1.0

Figure 6. Force-displacement, linearized natural frequency, and potential energy.

over a span of 100 forcing cycles against the forcing frequency for a load level of P0 = 0.169. This forcing
corresponds to the static case of P = 0.9Psnap from Figure 6. To alleviate the problem of initial condition
sensitivity the simulations were allowed to run for a period of time prior to initiation of the energy averaging
algorithm. For the periodically excited system there is also the possibility of coexisting oscillations, this was
not the focus of the current study.3
To highlight the different types of response, Figure 8 shows several selected time series whose frequencies
are indicated by the dashed red lines in Figure 7. It is clear that trajectories that dwell in either the pre-
snap or post-snap configuration (i.e., those that do not consistently snap-through) have lower average total
energy than those that traverse between them. The chaotic trajectory in Figure 8(f) has an average total
energy in between that of non snap-through and periodic snap-through trajectories. This is expected as it
intermittently exhibits both small and large amplitude motion.
Total Energy
0.30

0.25

0.20

0.15

0.10

0.05

0.00
0.0 0.5 1.0 1.5 2.0
Ω

Figure 7. Average total energy of the response as a function of forcing frequency.

5 of 9

American Institute of Aeronautics and Astronautics

 1.0 1.0
y y
0.5 0.5

0.0 0.0
2020 2040 2060 2080 t 2020 2040 2060 2080 t
-0.5 -0.5

-1.0 -1.0

(a) Ω = 0.30 (b) Ω = 0.51

1.0 1.0
y y
0.5 0.5

0.0 0.0
2020 2040 2060 2080 t 2020 2040 2060 2080 t
-0.5 -0.5

-1.0 -1.0

(c) Ω = 0.60 (d) Ω = 0.86

1.0 1.0
y y
0.5 0.5

0.0 0.0
2020 2040 2060 2080
t 2050 2100 2150 t
-0.5 -0.5

-1.0 -1.0

(e) Ω = 0.94 (f) Ω = 1.28

1.0 1.0
y y
0.5 0.5

0.0 0.0
2050 2100 2150 t 2020 2040 2060 2080
t
-0.5 -0.5

-1.0 -1.0

(g) Ω = 1.37 (h) Ω = 2.14

Figure 8. Time series for the sample frequencies shown.

B. Snap-Through Boundaries
The forcing magnitude and frequency obviously play a large role in dictating whether the system will snap-
through. If the forcing frequency is such that the forcing and system response velocity are often in phase then
the system will gain sufficient energy to snap-through (however as opposed to linear systems this behavior
is bounded). Figure 9 shows contours of average total energy in parameter space (P0 , Ω) in which dark
red corresponds to low energy and white corresponds to high energy. The horizontal line at P0 = 0.168
corresponds to the forcing magnitude used to generate the data in Figures 7 and 8. Plots such as this could
prove to be a useful design tool showing all the regions in parameter space of low energy response.

IV. Two Degree-of-Freedom Link Model Arch

Aircraft typically have surface panels of varying curvature over their exterior, which can lead to complex
modal interactions.14 Improper buckling mode analysis of curved panels can also lead to errors when the
incorrect instability mode is assumed.15
Unfortunately it is difficult to analyze modal interactions on continuous models of arches and shells.
For continuous models there exist an infinite set of displacement modes and buckling modes, which do not
necessarily coincide. The mode shapes themselves are also continuous functions, which makes the force-
displacement relationship difficult to characterize, especially graphically. This complexity diverts attention

6 of 9

American Institute of Aeronautics and Astronautics

 Po
0.15

0.10

0.05

0.00
0.0 0.5 1.0 1.5 2.0
Ω
Figure 9. Total energy contours of a SDOF link model arch as a function of the forcing parameters. the color
scheme goes from low (dark red) to high (white) energy levels.

from the goal of analyzing the interactions between modes. To simplify the analysis the two DOF link
model in Figure 10 is introduced instead. This system also includes imperfections in the initial symmetry of
the arch, in that the initial angles θ10 and θ20 at which the springs are unstretched and untorqued are not
necessarily the same. The torque springs are necessary to ensure that the system is not a mechanism.

P
M M

αL
L L

k1 k2 k2 k1
θ 1 θ10 θ 20 θ
2

Figure 10. Schematic of two DOF link model arch. In this case M = 0.

The potential energy of the system, which in this case is better described in angular coordinates, is given
by

 2
1
q q
2 2 2 2 2
V (θ1 , θ2 ) = k1 L cos θ1 + cos θ2 + α − (sin θ1 − sin θ2 ) − cos θ10 − cos θ20 − α − (sin θ10 − sin θ20 )
4
1 h 2 2
i 1
+ k2 (θ1 − θ10 ) + (θ2 − θ20 ) + P L (sin θ1 + sin θ2 ) . (5)
2 2
The force-displacement equations are omitted for brevity, however, as always in a potential system they
are given located at the stationary points of the potential energy function, i.e., ∂V /∂θ1 = ∂V /∂θ2 = 0.
The stability of the equilibria can be obtained by investigating the determinant of the Hessian matrix
(matrix of second derivatives) of the potential energy function. A positive definite Hessian indicates a local
minimum and hence stable equilibria, while a negative definite or non-definite matrix indicates an unstable

7 of 9

American Institute of Aeronautics and Astronautics

equilibria.16, 17 This is equivalent to analyzing the local linear natural frequency.
Figure 11 shows the force-displacement relationship for a perfect system (L = 1, α = 2, k1 = 1, k2 =
0.1, θ10 = θ20 = π/3) and imperfect system (θ10 = π/3, θ20 = 0.9π/3). Dashed lines are used to indicate
unstable equilibria. This figure shows that the system prefers the unsymmetric equilibria. The results are
also similar to related work by other researchers on bistable structures.18
1 1
θ2 θ2 0
0 (a) (b)
1 1
1.0 1.0
P P
0.5 0.5

0.0 0.0

-0.5 -0.5
1 1
θ1 0 1 θ1 0
1

Figure 11. Force-displacement relationship of two DOF link arch.

The force-displacement relationship of the two-link system illustrates the difficulties introduced when
a system has multiple equilibrium paths, since there is no unique path along which the system can snap-
through. This makes it more difficult to assign an energy threshold required for snap-through. This is the
focus of current research by the authors.

V. Conclusions
The fundamental mechanisms of snap-through buckling is investigated using a shallow arch, and a pair of
link models. A promising method of distinguishing snap-through from non snap-through response based on
average total energy is introduced. The complexities of higher dimensional systems are investigated through
a two DOF link model arch. This research is being extended to higher dimensional (continuous) systems to
provide a useful tool in the practical design of snap-through susceptible systems such as aircraft panels.

Acknowledgments
The authors gratefully acknowledge the support of the US Air Force, grant FA9550-09-1-0204. RW for
the award of a Canadian NSERC scholarship.

References
1 Clarkson, B., “Review of sonic fatigue technology,” Tech. rep., NASA Contract Report 4587, 1994.
2 Suresh, S., Fatigue of Materials, 2nd edition, Cambridge University Press, 1998.
3 Virgin, L., Vibrations of Axially Loaded Structures, Cambridge University Press, 2007.
4 Murphy, K., Virgin, L., and Rizzi, S., “Experimental snap-through boundaries for acoustically excited, thermally buckled

plates,” Experimental Mechanics, Vol. 36, 1996, pp. 312–317.

5 Timoshenko, S. and Woinowsky-Krieger, S., Theory of Plates and Shells, 2nd Edition, McGraw Hill, 1959.
6 Chen, L. and Chen, L., “Thermal Buckling of Laminated Composite Plates,” Journal of Thermal Stresses, Vol. 10, No. 4,

1987, pp. 345–356.

7 Murphy, K., Virgin, L., and Rizzi, S., “The effect of thermal prestress on the free vibration characteristics of clamped

rectangular plates: Theory and Experiment,” Journal of Vibration and Acoustics, Vol. 119, 1997, pp. 243–249.
8 Hsu, C., “Stability of shallow arches against snap-through under timewise step loads,” Journal of Applied Mechanics,

Vol. 35, 1968, pp. 31–39.

9 Rehfield, L., “Nonlinear flexural oscillations of shallow arches,” AIAA Journal , Vol. 12, 1974, pp. 91–93.
10 Johnson, E. and McIvor, I., “The effect of spatial distribution on dynamic snap-through,” Journal of Applied Mechanics,

Vol. 45, 1978, pp. 612–618.

11 Gregory, W. and Plaut, R., “Dynamic stability boundaries for shallow arches,” Journal of Engineering Mechanics,

Vol. 108, 1982, pp. 1036–1050.

8 of 9

American Institute of Aeronautics and Astronautics

 12 Plaut, R. and Hsieh, J.-C., “Oscillations and instability of a shallow arch under two-frequency excitation,” Journal of

Sound and Vibration, Vol. 102, 1985, pp. 189–201.

13 Virgin, L., “Parametric studies of the dynamic evolution through a fold,” Journal of Sound and Vibration, Vol. 110,

1986, pp. 99–109.

14 Gordon, R. and Hollkamp, J., “Reduced-Order Modeling of the Random Response of Curved Beams using Implicit

Condensation,” Proceedings of the Fourty-Seventh AIAA Structures, Structural Dynamics, and Materials Conference, AIAA,
Newport, Rhode Island, 2006.
15 Wardle, B., “The Incorrect Benchmark Shell Buckling Solution,” Proceedings of the Fourty-Seventh AIAA Structures,

Structural Dynamics, and Materials Conference, AIAA, Newport, Rhode Island, 2006.
16 Crisfield, M., Nonlinear Finite Element Analysis of Solids and Structures, Vol.1: Essentials, Wiley, 1997.
17 Crisfield, M., Nonlinear Finite Element Analysis of Solids and Structures, Vol.2: Advanced Topics, Wiley, 1997.
18 Maurini, C., Pouget, J., and Vidoli, S., “Distributed piezoelectric actuation of a bistable buckled beam,” European

Journal of Mechanics A/Solids, Vol. 26, 2007, pp. 837–853.

9 of 9

American Institute of Aeronautics and Astronautics