STRUCTURAL STABILITY

CE 5251
Dr. T. P. Tezeswi
Assistant Professor
Department of Civil Engineering
NIT-Warangal
email: [email protected]
Syllabus

Code provisions for structural stability

Stability

[1]
 Discrete Systems:
Simple Spring-Bar System

(a)

•A straight vertical rigid bar of length L is restrained at its bottom by an elastic spring
with a spring constant k. At the top of the bar there is applied a force P acting along its
longitudinal axis.
•The moment caused by the axial load acting through the displacement is resisted by the
spring reaction kθ.
•The symbol θ represents the angular rotation of the bar in radians.
[1]
 Discrete Systems:
Simple Spring-Bar System: Critical load
“Perfect System” (Small Deformation Theory)
•The critical load is the load that, when placed on the structure, causes it to pass
from a stable state to an unstable state.
•Consider the deformed shape and slightly perturb the system with a rotation θ
and impose equilibrium on the deformed state.
•Summing moments about point A we obtain:

•Solving for P at equilibrium, we obtain:

•Considering very small deformations, we can utilize small

displacement theory (a.k.a small strain theory) which allows us to simplify
using: sin θ≈ θ, tan θ≈ θ, cos θ≈ 1
•Substituting sin θ≈ θ, we determine the critical load Pcr of the spring-bar
model to be:
[1]
 Discrete Systems:
Simple Spring-Bar System: Critical Load
“Perfect System”
The equilibrium is in a neutral position: it can exist both in the un-deformed
and the deformed position of the bar. The small displacement response of
the system is shown below:

[1]
 Discrete Systems:
Simple Spring-Bar System: Critical Load
“Perfect System”

•The load ratio PL/k =1 is also known as the critical load, the buckling load, or
the load at the bifurcation of the equilibrium.

•The bifurcation point is a branch point; there are two equilibrium paths after
Pcr is reached, both of which are unstable.

•The upper path has an increase in P with no displacement. This equilibrium

path can only exist on a perfect system with no perturbation and is therefore not
a practical solution, only a theoretical one.

[1]
 Discrete Systems:
Simple Spring-Bar System: Critical Load
“Perfect System” Virtual Work Method
The total potential P of an elastic system is defined as Π= U+Vp
Where:
•U is the elastic strain energy of a conservative system. In a conservative
system the work performed by both the internal and the external forces is
independent of the path traveled by these forces, and it depends only on the
initial and the final positions. U is the internal work performed by the internal
forces; U = Wi.
•Vp is the potential of the external forces, using the original deflected position
as a reference. Vp is the external work; Vp = -We.

•The strain energy is the work done by the spring, U = Wi = ½ Kθ2

•The potential of the external forces is equal to Vp = - We = - PL(1-cos θ)

•The total potential in the system is then given by:

Π= ½ Kθ2 - PL(1-cos θ)

[1]
 Discrete Systems:
Simple Spring-Bar System: Critical Load
“Perfect System” Virtual Work Method

•According to the principle of virtual work the maxima and minima are
equilibrium positions, because if there is a small change in θ, there is no
change in the total potential. In the terminology of structural mechanics, the
total potential is stationary.
•It is defined by the derivative: dΠ/dθ = 0
•For the spring bar system, equilibrium is obtained when
•dΠ/dθ = 0 = Kθ- PL sinθ
•To find Pcr, we once again apply small displacement theory sin θ≈ θ and
obtain Pcr= k/L as before. [1]
 Discrete Systems:
Simple Spring-Bar System: Critical Load
“Perfect System” Virtual Work Method

Summary of Important Points

•Instability occurs when a small change in load causes a large change in
displacement. This can occur on a local, member or system level.
•The critical load, or buckling load, is the load at which the system passes from
a stable to an unstable state.
• The critical load is obtained by considering equilibrium or potential energy of
the system in a deformed configuration.
• Small displacement theory may be used to simplify the calculations if only the
critical load is of interest.

[1]
 Discrete Systems:
Simple Spring-Bar System: Stable and unstable behavior
“Post Buckling History”
•Reconsider the simple spring-bar model but introduce a disturbing moment,
Mo at the base of the structure.

•A straight, vertical rigid bar of length L is restrained at its bottom by an

infinitely elastic spring with a spring constant k. At the top of the bar there is
applied a force P acting along its longitudinal axis.
•The right fig. shows the deformation of the bar if a disturbing moment Mo is
acting at its base. This moment is resisted by the spring reaction kθ, and it is
augmented by the moment caused by the product of the axial force times the
horizontal displacement of the top of the bar. [1]
 Discrete Systems:
Simple Spring-Bar System: Stable and unstable behavior
“Post Buckling History”
•Taking moments about the base of the bar (point A) we obtain the following
equilibrium equation for the displaced system:
ΣMA =0 = PL sinθ+Mo-kθ
•Letting θo = Mo/k and rearranging, we can write the following equation:
PL/k = (θ- θo) /sin θ
This expression is plotted for three values of the disturbing action: θo =0; θo
=0.01; θo =0.05

[1]
 Discrete Systems:
Simple Spring-Bar System: Stable and unstable behavior
“Post Buckling History”
•When θo =0, (i.e., PL/k = θ/sin θ), there is no possible value of PL/k less than
unity since θ is always larger than sin θ. Thus no deflection is possible if
PL/k<1.
•At PL/k>1 deflection is possible in either the positive or the negative direction
of the bar rotation.
•As θ increases or decreases the force ratio required to maintain equilibrium
becomes larger than unity. However, at relatively small values of θ, (<0.1
radians, or ~5o), the load deformation curve is flat for all practical purposes.
•Approximately, it can be said that equilibrium is possible at θ = 0 and at a
small adjacent deformed location, say θ <0.1 rad or so.
•The load PL/k =1 is thus a special type of load, when the system can
experience two adjacent equilibrium positions: one straight and one deformed.
•The equilibrium is thus in a neutral position: It can exist both in the un-
deformed and the deformed position of the bar.

[1]
 Discrete Systems:
Simple Spring-Bar System: Stable and unstable behavior
“Post Buckling History”
•The other two sets of solid curves in previous Fig are for specific small values
of the disturbing action θo of 0. 01 and 0.05 radians.
•These curves each have two regions: When θ is positive, that is, in the right
half of the domain, the curves start at θ = θo when PL/k = 0 and then gradually
exhibit an increasing rotation that becomes larger and larger as PL/k =1.0 is
approached, finally becoming affine to the curve for θo = 0 as θ becomes very
large.
•While this in not shown, the curve for smaller and smaller values of θo will
approach the curve of the bifurcated equilibrium.
•The other branches of the two curves are for negative values of θ. They are in
the left half of the deformation domain and they lie above the curve for
θo = 0.
•They are in the unstable region for smaller values of θ, that is,
they are above the dashed line defining the region between stable and unstable
behavior, and they are in the stable region for larger values of θ.
[1]
 Discrete Systems:
Simple Spring-Bar System: Stable and unstable behavior
“Imperfect System”
•The nature of the equilibrium, that is, its stability, is examined by disturbing
the already deformed system by an additional small rotation θ*.

The equilibrium equation of the disturbed geometry is:

ΣMA =0 = PL sin(θ+ θ*) + Mo- k(θ+ θ*)
After rearranging we get, noting that θo = Mo
PL/k = (θ+ θ*- θo)/sin (θ+ θ*)
[1]
 Discrete Systems:
Simple Spring-Bar System: Stable and unstable behavior
“Imperfect System”
•The nature of the equilibrium, that is, its stability, is examined by disturbing
the already deformed system by an additional small rotation θ*.
•Since sin (θ+ θ*) = sinθcos θ*+ cos θsin θ*. For small values of u we can use
cos θ* ≈1.0; sin θ* ≈ θ*, and therefore:
PL/k = (θ+ θ*- θo)/sin (θ+ θ*cos θ)
•This equation can be rearranged to the following form:
(PL/k sinθ)- θ+ θo+ θ ((PL/k) cosθ-1)= 0.
•However, (PL/k)sinθ-θ+θo=0 and θ ⱡ0, and thus (PL/k)cosθ-1 = 0

[1]
 Discrete Systems:
Simple Spring-Bar System: Stable and unstable behavior

In the discussion so far we have derived three hierarchies of results, each

requiring more effort than the previous one:
1. Buckling load of a perfect system.
2. The post-buckling history of the perfect system.
3. The deformation history of the „„imperfect‟‟ system.
[1]
 Discrete Systems:
Simple Spring-Bar System: Virtual Work Method
•In the previous derivations the equilibrium condition was established by
utilizing the statical approach. Equilibrium can also be determined by using the
theorem of virtual work.

•Examine the large displacement behavior of the system using the

energy approach described previously.

•The strain energy is the work done by the spring, U = Wi = ½ Kθ2

•The potential of the external forces is equal to Vp = - We = - PL(1-cos θ)

•With θo= Mo/k, the total potential in the system is then given by:
Π/k= ½ θ2 – PL/k(1-cos θ)- θoθ

[1]
 Discrete Systems:
Simple Spring-Bar System: Virtual Work Method
•The total potential is plotted against the bar rotation in Figure 1.9 for the case of uo ¼
0:01 and PL=k ¼ 1:10.
•In the range 1:5 u 1:5 the total potential has two minima (at approximately u ¼ 0:8
and 0:7) and one maximum (at approximately u ¼ 0:1).
•According to the Principle of Virtual Work, the maxima and minima are equilibrium
positions, because if there is a small change in u, there is no change in the total
potential.
•In the terminology of structural mechanics, the total potential is stationary.
•It is defined by the derivative dP/du =0.
•From equation 1.6, dP/du =0 =2u2 PL k sin u uo, or

[1]
 Review
• Equilibrium
– Consider three cases of equilibrium of a ball

(a) (b) (c)

(a) Ball on concave spherical surface: if displaced slightly, will return to its original
position upon the removal of the disturbance. Stable equilibrium

(b) Ball on convex spherical surface: if displaced slightly, will continue to move
farther away from the original position. Unstable equilibrium

(c) Ball on horizontal plane: it remains at the position to which the small
disturbance has taken it. Neutral equilibrium
 Discrete Systems:
Simple Spring-Bar System: Softening Structure
The previous spring-bar system exhibited a particular post-buckling
characteristic: The post-buckling deflections increased as the load was
raised above the bifurcation point.
Such hardening behavior is obviously desirable from the standpoint of safety.
However, there are structural systems where the post-buckling exhibits a softening
character.

[1]
 Discrete Systems:
Simple Spring-Bar System: Softening Structure

[1]
 Discrete Systems:
Simple Spring-Bar System: Softening Structure

[1]
 Discrete Systems:
Simple Spring-Bar System: Softening Structure

[1]
 Discrete Systems:
Simple Spring-Bar System: Softening Structure

[1]
 Discrete Systems:
Simple Spring-Bar System: Softening Structure

[1]
 Discrete Systems:
Post buckling behavior

[1]
 Discrete Systems:
Post buckling behavior

•Plates are insensitive to initial imperfections, exhibiting

reliable additional strength beyond the buckling load.

•Shells and columns that buckle after some parts of their

cross section have yielded are imperfection sensitive.

•Elastic buckling of columns, beams, and frames have little

post-buckling strength, but they are not softening, nor are
they hardening after buckling.

[1]
 Discrete Systems:
Snap Through buckling
Consider a two-bar structure where the two rigid bars are at an angle to each other.
One end of the right bar is on rollers that are restrained by an elastic spring.

loading and geometry

deformed shape after the load is applied.

[1]
 Discrete Systems:
Snap Through buckling
Equilibrium is determined by taking moments of the right half of the deformed
structure about point A.

From the deformed geometry it can be shown that

The equilibrium equation thus is determined to be

The state of the equilibrium is established by disturbing the deflected structure by an

infinitesimally small virtual rotation θ. After performing trigonometric and algebraic
manipulations it can be shown that the curve separating stable and unstable equilibrium is

If we substitute PL=k into previous equation , we get, after some elementary operations, the
following equation that defines the angle θ at the limit of stable equilibrium

[1]
 Discrete Systems:
Snap Through buckling

[1]
 Discrete Systems:
Snap Through buckling

[1]
 Discrete Systems:
Snap Through buckling

[1]
 Discrete Systems:
MDOF Systems
A structure is made up of three rigid bars placed between a roller
at one end and a pin at the other end. The center bar is connected
to the two edge bars with pins. Each interior
pinned joint is restrained laterally by an elastic spring with a spring
constant k.

Buckling analysis is performed by assuming small deflections and an initially perfect geometry.
Thus, the only information to be gained is the critical load at which a straight and a buckled
configuration are possible under the same force. [1]
Discrete Systems:
MDOF Systems

[1]
Discrete Systems:
MDOF Systems

[1]
 Discrete Systems:
MDOF Systems
The smaller of the two critical loads is then the buckling load
of interest to the structural engineer.
Substitution each of the critical loads into equation 1.26
results in the mode shapes of the buckled configurations, as
illustrated.
Finally then, Pcr= kL/3 is the governing buckling load, based
on the small deflection approach.

[1]
Discrete Systems:
MDOF Systems

Buckled Shapes

[1]
 Discrete Systems:
MDOF Systems: Energy Approach
The energy method can also be used for arriving at a solution
to this problem.

[1]
 Discrete Systems:
MDOF Systems: Energy Approach

[1]
 Discrete Systems:
MDOF Systems: Energy Approach

[1]
 Discrete Systems:
Summary
• The critical, or buckling load, of geometrically perfect systems
• The behavior of structures with initial geometric or statical
imperfections
• The amount of information obtained by small deflection and
large deflection analyses
•The equivalence of the geometrical and energy approach to
stability analysis
•The meaning of the results obtained by a bifurcation analysis, a
computation of the post-buckling behavior, and by a snap-
through investigation.
•The hardening and the softening post-buckling deformations
•The stability analysis of multi-degree-of-freedom systems

[1]
Eigenvalue Problem: Pinned-Pinned Column
Equilibrium is again formulated on the deformed deflection
configuration.

[1]
Eigenvalue Problem

[1]
Eigenvalue Problem

[1]
Eigenvalue Problem

[1]
Eigenvalue Problem

[1]
 Eigenvalue Problem: Discrete Systems
In an eigenvalue buckling problem we look for the loads for which the model stiffness
matrix becomes singular, so that the problem has nontrivial solutions.
For a system with n degrees of freedom (DOF), the eigen-system has the following form

Ax = λBx
Where: A and B are square matrices of order n,
the n eigenvalues λi are the critical load factors
the eigenvectors xi are the associated buckling shapes

•Generally only the smallest eigenvalue, which will be denoted by λ1, will be of interest for
engineering design.
•Matrices A and B may be obtained by either equilibrium analysis, or by energy
methods
•If equilibrium approach based on doing Free Body Diagrams (FBD) on a slightly
perturbed configuration is used, the matrices A and B are not necessarily symmetric, but
may be symmetrized if necessary.
•If the energy method is used, the symmetry of both matrices is guaranteed from the
start.
[1]
 Eigenvalue Problem
•How to find P? Solve the eigenvalue problem.
[A]{y}+[B]{y}={0}
•Standard Eigenvalue Problem
[A]{y}={y}
Where,  = eigenvalue and {y} = eigenvector
•Can be simplified to [A-I]{y}={0}
•Nontrivial solution for {y} exists if and only if
| A-I|=0
•One way to solve the problem is to obtain the characteristic
polynomial from expanding | A-I|=0
•Solving the polynomial will give the value of 
•Substitute the value of  to get the eigenvector {y}

Note: This is not the best way to solve the problem, and will not
work for more than 4or 5th order polynomial
 Eigenvalue Problem: Numerical Solution
Calculation of Buckling Loads by the Finite Element Method
•The calculation of the linear bifurcation point is based on the eigenvalue problem
constituted from the linear stiffness matrix [ K] and the geometrical stiffness matrix [Kσ ]
as:
([Ko] +λ [Kσ]) δ {D} =0
=

[KT] = [K]+ [Kσ] = Tangent stiffness matrix

[Ko] contains the stiffness of the structure
[Kσ] contains the membrane forces
where the smallest root λ defines the smallest level of load where bifurcation occurs
 Eigenvalue Problem: Numerical Solution
•Inverse Power Method: Numerical Technique to Find Least
Dominant Eigenvalue and its Eigenvector
•Based on an initial guess for eigenvector and iterations Algorithm
1) Compute [E]=-[A]-1[B]
2) Assume initial eigenvector guess {y}0
3) Set iteration counter i=0
4) Solve for new eigenvector {y}i+1=[E]{y}i
5) Normalize new eigenvector {y}i+1={y}i+1/max(yji+1)
6) Calculate eigenvalue = 1/max(yji+1)
7) Evaluate convergence: i+1-i < tol
8) If no convergence, then go to step 4
9) If yes convergence, then = i+1 and {y}= {y}i+1
 References
[ 1] Theodore V. Galambos Andrea E. Surovek , Structural Stability of Steel: Concepts
and Applications for Structural Engineers, 2008 John Wiley & Sons, In