Stability Chapter 01
Stability Chapter 01
1. Basic Concepts
1.1 Introduction
The present chapter introduces the basic concepts about stability of structures. Making reference
to simple discrete systems, it deals with the basic methods and approaches to be applied and pursuit
when the study of structures stability is of concern.
Definition of equilibrium conditions is the first subject of the discussion; it is based on
“quantitative” relationships involving forces acting on the system and its relevant geometric and
mechanical properties, such as dimensions, stiffness and mass. Discussion about stability of structures
begins just after such a definition of equilibrium conditions; it basically examines the aspect of the
“quality” of the equilibrium configuration possibly achieved by the system. As intuitive, “quality” of
equilibrium – for both rigid bodies and elastic or even non-linear systems - can be classified on the
basis of its possible evolution after a possible (even small) perturbation on the system while resting in
its equilibrium condition. Under this standpoint, basic mechanical knowledge and physical every-day
intuition draw to the following three possible definitions for an equilibrium configuration of the
system:
- stable equilibrium, if the system after perturbation come back to the initial equilibrium
configuration, possible through small oscillations in its neighbours;
- unstable equilibrium, if a perturbation however small in intensity results in a change of the
initial equilibrium configuration;
- neutral equilibrium, if the system achieves in a new equilibrium configuration, passing
through other equilibrium configurations, depending on the intensity of the perturbation.
The three rather different situations described above are usually emphasized through a very
popular illustration dealing with the static equilibrium of rigid body reported in Figure 1.1.
structures can be based on an energy approach for which a conceptual correspondence can be stated with
Figure 1.1, as will be explained within the next paragraphs.
The basic definitions given for the “quality” of equilibrium configuration with reference to the
simple free body depicted in Figure 1.1 need to be generalized to structural systems usually made of
flexible member connected one another for facing the loads and action applied on them. Nevertheless,
a hopefully meaningful application of the three mentioned approaches will be carried out in following
sections of the present chapter on simpler systems made of rigid bodies connected one another by a
finite number of flexible elements (basically linear springs) with the aim of providing the readers with
the basic insights on stability of structural systems. These structures are usually named “discrete
system” and so will be done in the following of this text.
Euler definition of stability will be firstly referred to discrete structures pointing out the condition
of possible alternative equilibrium configuration available for the system (affected by no imperfections
and in equilibrium in its reference position) when (axial) loads achieve a certain “critical” value: this
possible availability of more than one configuration is currently named as “bifurcation” of equilibrium.
Moreover, imperfections do play a relevant role in the so called “equilibrium path” (namely the
relationship between the current value of loads and the corresponding equilibrium configuration of the
system) and their influence will be even analyzed in this chapter with reference to the mentioned
discrete systems.
Finally, bifurcation, instability onset (namely “buckling”) and the way in witch initial
imperfections of the system affect them will be analyzed in the light of the Koiter’s theorems which are
a very powerful tool for understanding the “quality” of a given equilibrium position.
length is L, connected to the soil by a hinge and a rotational spring k and loaded of the opposite end by
an axial load P (Figure 1.2).
L k
P P
θ
L
Figure 1.3: Discrete system #1: deformed configuration for applying the static approach
A simple equilibrium condition can be stated in terms of moments around the base hinge; only
two contributions participate play a role in the equilibrium of the above system: the axial load applied
on the deformed configuration and the rotational spring, was moment is proportional to the actual
rotation θ:
PL sin θ − kθ = 0 . (1.1)
If the first of the two Euler hypotheses applies (namely, small displacements), then the value of
the sine can be approximated by its argument and equation (1.1) can be written as follows:
[ PL − k ] ⋅ θ = 0 . (1.2)
The above equation points out that two possible solutions (and two corresponding equilibrium
configurations) can be found for the system:
- the trivial configuration, described by the solution θ = 0 which firstly satisfy the above
equilibrium equation;
- a bifurcated solution obtained as the first factor of equation (1.2) vanishes:
k
PL − k = 0 ⇒ P = = PE ; (1.3)
L
such a condition occurs as the external load attains the value PE, which is a bifurcation load in
the sense that under PE the system can switch out from its (trivial) configuration toward a
bifurcated one.
Static approach carried out under the mentioned Euler hypotheses, leads to describe the
equilibrium paths as a bifurcation phenomenon which can have place for a precise value of the external
load P equal to PE. Figure 1.4 show the equilibrium path, namely the couples (θ, P) obtained by solving
equation (1.2).
P/PE P
Bifurcation point
P (P=PE)
Figure 1.4: Discrete system #1: bifurcation phenomenon derived under the Euler assumptions
Although the static approach lead to a rather simple way for evaluating all the possible
equilibrium configurations of the system, nothing can be deduced for qualifying the single branches of
the equilibrium path in terms of stability. Physical sense could suggest that the vertical branch would be
of stable form the origin to P<PE and unstable above that value, but no quantitative evaluation can be
done on that issue. Moreover, nothing can be understood for the linear branch, where the key
hypothesis of small displacement is not always verified.
that can be written as follows in terms of moments and angular frequency (rather than forces F and
linear acceleration z because of the circular nature of motion):
M − Iθ = 0 . (1.5)
In the last equation M is the sum of moments of the external forces with respect to the hinged
end of the bar:
M = PL sin θ − kθ , (1.6)
while the second moment I of the bar with respect to the hinged end can be evaluated as follows:
L
μ L3
I = ∫ μ x 2 dx = , (1.7)
0
3
being μ the linear mass density of the system (Figure 1.5).
L sinθ
xθ μdx
θ
L
and the constant terms A and B could be found imposing the suitable initial conditions in terms
of displacement θ ( 0 ) and velocity θ ( 0 ) . For whatever values of such initial condition the
motion consists in oscillations of the body around the initial configuration due to the oscillatory
nature of trigonometric functions: consequently, if P<PE the equilibrium in stable;
3 k
3[
- PL − k ] > 0 , namely P > = PE : in this case the equation of motion takes the following
μL L
form:
θ − ω 2θ = 0 , (1.13)
whose general solution is:
θ ( t ) = Ae ωt + Be −ωt , (1.14)
and the constant terms A and B could be found imposing the suitable initial conditions in terms
of displacement θ ( 0 ) and velocity θ ( 0 ) . Nevertheless, the nature of motion is always diverging
since the positive exponential of equation (1.14) is a strictly increasing function which describe
the motion of the body which abandon the initial equilibrium position toward another
configuration: consequently, if P>PE the equilibrium in unstable.
Finally, besides its procedural difficulties, dynamical approach is the closest one with respect to
the intuitive definition of equilibrium stability represented in Figure 1.1 for a rigid ball within the
gravitational field. Consequently, further information about the quality of equilibrium has been drawn
for the two vertical branches of the equilibrium path represented in Figure 1.4., showing that the lower
part (P<PE) is stable in nature, while the upper one (P>PE) is unstable. On the contrary, nothing can be
said about the nature of the horizontal post-bifurcation branch: large displacements occur after
bifurcation and Euler can only provide the bifurcation point, but is not fit for describing post-buckling
behaviour.
- equilibrium is stable if the function Π has a local minimum point, namely, in the cases of
single degree of freedom systems, the second derivative is positive;
- equilibrium is unstable if the function Π has a local maximum and, consequently, the second
derivative is negative.
The second derivative of Π can be easily determined as follows:
d 2Π
= k − PL cosθ ≈ k − PL , (1.19)
dz 2
and its sign can be easily derived:
d 2Π
- > 0 as k > PL ⇒ P<PE , meaning that equilibrium is stable in the lower part of the
dz 2
vertical branch;
d 2Π
- < 0 as k < PL ⇒ P>PE , meaning that equilibrium is unstable in the upper part of the
dz 2
vertical branch.
Finally, the same conclusions derived through the dynamic approach can be achieved by the
energy-based approach for determining the equilibrium positions of the systems and its stability. No
further insights have been derived about the post-bifurcation branch as a result of the limitation
imposed by the small displacement hypothesis that will be finally removed in the next section for
understanding the post-buckling behaviour in terms of equilibrium path and its stability conditions.
PL/k P
Bifurcation point
P (P=PE)
θ
Figure 1.6: Discrete system #1: Post-Buckling behaviour
The quality of equilibrium throughout this curved path reported within the (θ, P/PE) plane can
be discussed by looking after the second derivative of the Total Potential Energy Π already evaluated
and reported in equation (1.19) where the expressions without small displacement simplification is of
interest for the post-buckling behaviour:
d 2Π
= k − PL cosθ . (1.22)
dz 2
The sign of the second derivative of Π throughout the post-buckling path can be deduced by
introducing equation (1.20) in (1.22) and obtaining the following condition:
d 2Π θ
> 0 ⇔ 1− >0 , (1.23)
dz 2
tan θ
which is true for every value of θ belonging to the range [0, π/2). Consequently the post-buckling
behaviour is stable.
Nevertheless, a variety of possible behaviours can occur after bifurcation. To have an example of
a significantly different one, the system represented in Figure 1.7 can be examined.
k1
L θ
dΠ
= k1L2 ⋅ sin θ cosθ − PL ⋅ sin θ = 0 . (1.25)
dθ
If both the basic assumptions of the Euler approach apply, the first-order approximation can be
adopted for the above equation whose final expression is reported below
[k1L − P ] ⋅ θ = 0 , (1.26)
which defines the value of the Euler load PE=k1L resulting in a bifurcation in equilibrium path of the
system. Figure 1.8 shows that the second system analyzed under the Euler assumption behaves like the
first one, since the equilibrium path consists of a vertical branch corresponding to the trivial
configuration (θ=0) and an horizontal one (P/PE=1) representing the bifurcation condition attained for
the Euler critical value PE.
Post-buckling behaviour is even of interest for this system, since significant differences with
respect to the first one could be pointed out, even if a formal equivalence between the two systems has
been deduced under the Euler assumptions. Equation (1.25) describes the post-buckling behaviour in
large displacements and can be easily solved with respect to the external load P and introducing the
above definition of the Euler load PE:
P P
= = cosθ , (1.27)
k1L PE
Equation (1.27) describing the equilibrium path is even represented in Figure 1.8 which points out
as post-buckling is a decreasing branch stemming out from the bifurcation point.
P/PE P
Bifurcation point
P (P=PE)
the sign of the second derivative can be studied after dividing by k1L2 and introducing the equation
(1.27) of the post-buckling branch:
d 2Π
> 0 ⇔ cos 2 θ − 1 > 0 ; (1.29)
dθ 2
which is not true for any value of θ within the range [0, π/2). Consequently, the second system is
subjected to an unstable equilibrium bifurcation.
Finally, post-buckling analysis points out the key aspects related to the stability or instability of
the equilibrium path after bifurcation that can be even foreseen by the simple Euler theory was
assumptions are not restrictive since perfect systems have small displacements.
e P P
θ
L
1.60
P/PE 1.40
Euler
1.20 bifurcation
1.00
0.80 e/L
0.60
0.00
0.05
0.40 0.10
0.15
0.20
0.20
0.00
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60
θ
Figure 1.10: Discrete system #1: the role of imperfection on equilibrium paths
Although no further analytical developments will e proposed herein for the sake of brevity, it would be
easy to demonstrate that the equilibrium for the system is stable in nature since its second derivative is
strictly positive throughout the entire equilibrium path.
Finally, the system #1 (and, by extension all those systems with stable post-buckling behaviour)
are not imperfection sensitive in terms of ultimate load, but have only larger lateral displacement as a
result of those imperfection.
The second system can be now studied considering that it is affected by an imperfection whose
effect can be simulated by an initial out-of-verticality angle θ0 (Figure 1.11).
k1
θ0
L θ
and the corresponding equilibrium configuration of the system can be sought by solving the following
equation:
dΠ
= 0 ⇒ k1L2 ⋅ [sin θ − sin θ 0 ] cosθ − PL ⋅ sin θ = 0 . (1.35)
dθ
Simplifying the above equation and introducing the definition of Euler load PE stated by equation
(1.26), the following relationship can be obtained for the equilibrium path:
P ( sin θ − sin θ 0 )
= ⋅ cosθ , (1.36)
PE sin θ
and in the case of “ideal” system ( θ0 = 0 ) equation (1.36) reduces to (1.27). Figure 1.12 plots that
relationship as a dashed line; the continuous lines reproducing the behaviour of the system with
imperfection are hugely affected by imperfections in terms of both lateral displacements in the
ascending branch and ultimate load which is usually as smaller than PE as the imperfection amplitude
θ0 .
1.20
P/PE
Euler
bifurcation
1.00
*
P /PE
0.80
e/L
0.60 0.00
0.01
0.40 0.05
0.10
0.20
0.15
0.00
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40
θ
Figure 1.12: Discrete system #2: the role of imperfection on equilibrium paths
Consequently, two achievements have been drawn by the above two systems that can be assumed
as paradigms of two classes of problems:
- in the first one, stable (and symmetric) behaviour occurs after bifurcation imperfection only
affects the lateral flexibility of the system;
- for the second one unstable bifurcation occurs and, when imperfections are taken into
account, both lateral stiffness and ultimate load are hugely affected.
Finally, the intimate relationship between post-buckling behaviour and imperfection sensitiveness
will be emphasized in the next section as one of the key aspects of a more general theory.
1.7 Applications