Anda di halaman 1dari 46

Lecture 5

Buckling Analysis

ANSYS Mechanical
Introduction to Structural Nonlinearities
1 2013 ANSYS, Inc. July 16, 2014
Chapter Overview
A. Background on Structural Stability

B. Linear (Eigenvalue) Buckling Procedure

C. Background on Nonlinear Buckling Techniques

D. Nonlinear Stabilization

It is assumed that the user has already covered Lecture 2


The capabilities described in this Chapter are generally applicable to ANSYS
Structural licenses and above.

2 2013 ANSYS, Inc. July 16, 2014


A. Background on Structural Stability
Many structures require an evaluation of their structural stability. Thin
columns, compression members, and vacuum tanks are all examples of
structures where stability considerations are important.
The instability could be global (such as a snap-through) or local (such as yielding
or buckling at a concentrated load or support).

Instability problems usually pose convergence difficulties and therefore require


the application of special nonlinear techniques.
3 2013 ANSYS, Inc. July 16, 2014
... Background on Structural Stability
At the onset of instability (buckling) a structure will have a very large change in
displacement {u} under essentially no change in the load (beyond a small load
perturbation).

F F
u

Stable Unstable

4 2013 ANSYS, Inc. July 16, 2014


... Background on Structural Stability
An idealized fixed end column will exhibit the following behavior under
increasing axial loads (F).

F F
F
Bifurcation Point Unstable Equilibrium

Neutral Equilibrium
Fcr
u
Stable Equilibrium

5 2013 ANSYS, Inc. July 16, 2014


... Background on Structural Stability
Bifurcation Point

A bifurcation point is a point in load history where two branches of


the solution are possible.

In the case of the idealized fixed end column, F F


at the critical load (Fcr), the column can buckle
to the left or to the right. Thus two load paths
are possible. In the case of real structures the P
existence of geometric imperfections or force
perturbations (P 0) will determine the direction
of the load path.
u

6 2013 ANSYS, Inc. July 16, 2014


... Background on Unstable Structures
Stable, Unstable, and Neutral Equilibrium

Consider the equilibrium of the ball shown below. If the surface is


concave upward the equilibrium is stable, the ball will return to its
original position if perturbed. If the surface is concave downward
the equilibrium is unstable, if perturbed the ball will roll away. If
the surface is flat the ball is in neutral equilibrium, if perturbed the
ball will remain in its new position.

Stable Unstable Neutral

7 2013 ANSYS, Inc. July 16, 2014


... Background on Structural Stability

Critical Load

At a value of F < Fcr the column is in stable equilibrium. If a small


perturbing force (P 0) is introduced and then removed, the
column will return to its original position. At values F > Fcr the
column is in unstable equilibrium, any perturbing force will cause
collapse. At F = Fcr the column is in neutral equilibrium, and this is
defined as the critical load.

8 2013 ANSYS, Inc. July 16, 2014


... Background on Structural Stability
Limit Load
In real structures the critical load can rarely be achieved. A
structure generally will become unstable at a load lower than the
critical load because of imperfections and nonlinear behavior.

F
Bifurcation Point

Actual structural response,


instability occurs below
Fcr critical load.

9 2013 ANSYS, Inc. July 16, 2014


... Background on Structural Stability
Analysis techniques for pre-buckling and collapse load analysis include:
Linear Eigenvalue Buckling
Nonlinear Buckling Analysis (using Stabilization)

F
Nonlinear
Linear Buckling Idealized Load Path
Eigenvalue
Imperfect Structures
Buckling
Load Path

Pre-buckling u

10 2013 ANSYS, Inc. July 16, 2014


B. Linear (Eigenvalue) Buckling Procedure
A linear buckling analysis is based on a classic eigenvalue problem.
To develop the eigenvalue problem, first solve the load-displacement
relationship for a linear elastic pre-buckling load state {P0}; i.e. given {P0}
solve for
{P0} = [Ke]{u0}
to obtain
{u0} = the displacements resulting from the applied load {P0}
{s} = the stresses resulting from {u0}

11 2013 ANSYS, Inc. July 16, 2014


... Linear (Eigenvalue) Buckling Procedure
Assuming the pre-buckling displacements are small, the incremental
equilibrium equations at an arbitrary state ({P}, {u}, {s}) are given by

{P} = [[Ke] + [Ks(s)]]{u}

where
[Ke] = elastic stiffness matrix
[Ks(s)] = initial stress matrix evaluated at the stress state {s}

12 2013 ANSYS, Inc. July 16, 2014


... Linear (Eigenvalue) Buckling Procedure

At the onset of instability (the buckling load {Pcr}), the structure can
exhibit a change in deformation {u} in the case of

{P} 0

By substituting the above expression into the previous incremental


equilibrium equations for the pre-buckling range we have

[[Ke] + l[Ks(s0)]]{u} = {0}


The above relation represents a classic eigenvalue problem.

13 2013 ANSYS, Inc. July 16, 2014


... Linear (Eigenvalue) Buckling Procedure
Hence, for a linear buckling analysis, the eigenvalue problem below is solved
to get the buckling load multiplier li and buckling modes yi:

K li S y i 0
Where:
[K] = stiffness matrix (constant)
[s] = stress stiffness matrix (constant)
i = ith buckling load factor
i = ith buckling mode shape
Linear elastic material behavior is assumed.
Small deflection theory is assumed
Nonlinear properties, if defined, are ignored.
It is important to remember these assumptions related to performing linear
buckling analyses in Mechanical.
14 2013 ANSYS, Inc. July 16, 2014
Linear (Eigenvalue) Buckling Procedure
A Static Structural analysis will need to be performed prior to (or in conjunction
with) a buckling analysis.

15 2013 ANSYS, Inc. July 16, 2014


Linear (Eigenvalue) Buckling Procedure
Any type of geometry supported by Mechanical may be used in
buckling analyses:
Solid bodies
Surface bodies (with appropriate thickness defined)
Line bodies (with appropriate cross-sections defined)
Only buckling modes and displacement results are available for line bodies.

Although Point Masses may be included in the model, only inertial loads
affect point masses, so the applicability of this feature may be limited in
buckling analyses

For material properties, Youngs Modulus and Poissons Ratio are required as
a minimum

16 2013 ANSYS, Inc. July 16, 2014


Linear (Eigenvalue) Buckling Procedure

Nonlinear contact, if defined, is ignored or reduced to a linear behavior


depending on type and initial status.

Linear Buckling Analysis


Contact Type
Initially Touching Inside Pinball Region Outside Pinball Region
Bonded Bonded Bonded Free
No Separation No Separation No Separation Free
Rough Bonded Free Free
Frictionless No Separation Free Free

17 2013 ANSYS, Inc. July 16, 2014


Linear (Eigenvalue) Buckling Procedure
At least one structural load, which causes buckling, should be applied to the
model:
All structural loads will be multiplied by the load multiplier (l) to determine the buckling
load (see below).
Compression-only supports are not recommended.
The structure should be fully constrained to prevent rigid-body motion.

F x l = Buckling Load

In a buckling analysis all applied loads (F) are scaled


by a multiplication factor (l) until the critical
(buckling) load is reached

18 2013 ANSYS, Inc. July 16, 2014


Linear (Eigenvalue) Buckling Procedure
Special considerations must be given if constant and proportional loads are
present.
The user may iterate on the buckling solution, adjusting the variable loads until the load
multiplier becomes 1.0 or nearly 1.0.
Consider the example of a column with self weight WO and an externally applied force A.
A solution can be reached by iterating while adjusting the value of A until l = 1.0. This
insures the self weight = actual weight or WO * l WO .

19 2013 ANSYS, Inc. July 16, 2014


Linear (Eigenvalue) Buckling Procedure
Buckling analyses are always coupled to a structural analysis within the project
schematic.
The Pre-Stress object in the tree contains the results from a structural analysis.
The Details view of the Analysis Settings under the Linear Buckling branch allows the
user to specify the number of buckling modes to find.

20 2013 ANSYS, Inc. July 16, 2014


Linear (Eigenvalue) Buckling Procedure
After setting up the model the buckling analysis can be solved along with the static
structural analysis.
A linear buckling analysis is more computationally expensive than a static analysis on the
same model.
The Solution Information branch provides detailed solution output.

21 2013 ANSYS, Inc. July 16, 2014


Linear (Eigenvalue) Buckling Procedure
After the solution is complete, the buckling modes can be reviewed:
The Load Multiplier for each buckling mode is shown in the Details view as well as the
graph and chart areas. The load multiplier times the applied loads represent the
predicted buckling load.

Fbuckle = (Fapplied x l)

22 2013 ANSYS, Inc. July 16, 2014


Linear (Eigenvalue) Buckling Procedure
Interpreting the Load Multiplier (l):
The tower model below has been solved twice. In the first case a unit load is applied. In
the second an expected load applied (see next page)

23 2013 ANSYS, Inc. July 16, 2014


Linear (Eigenvalue) Buckling Procedure
Interpreting the Load Multiplier (l):

BucklingLoad l *Unit _ Load


BucklingLoad l

BucklingLoad l * Actual _ Load


BucklingLoad
l Safety _ Factor
Actual _ Load

24 2013 ANSYS, Inc. July 16, 2014


Linear (Eigenvalue) Buckling Procedure
The buckling load multipliers can be reviewed in the Timeline section of the
results under the Linear Buckling analysis branch
It is good practice to request more than one buckling mode to see if the structure may
be able to buckle in more than one way under a given applied load.

25 2013 ANSYS, Inc. July 16, 2014


... Workshop
Please refer to your Workshop Supplement for instructions on:
W5A- Linear (Eigenvalue) Buckling

26 2013 ANSYS, Inc. July 16, 2014


C. Background on Nonlinear Buckling Techniques

Shown below is a generalized nonlinear load deflection curve. This figure


illustrates the idealized load path, an imperfect structures load path, and
the actual dynamic response of the structure.

Limit Point Idealized Load Path


Bifurcation
Point Imperfect Structures
Load Path
Idealized
Static Actual Dynamic
Behavior Response

Pre-buckling Post-buckling u

27 2013 ANSYS, Inc. July 16, 2014


... Background on Nonlinear Buckling Techniques
There are various analysis techniques available in WB-Mechanical for solving
the nonlinear static force deflection response of a structure. These techniques
include:

Load Control
Displacement Control
Nonlinear Stabilization

28 2013 ANSYS, Inc. July 16, 2014


... Background on Nonlinear Buckling Techniques
Load Control:
Consider the snap through analysis of the shallow arch shown below. When
the solution to a problem is performed with incrementally applied forces (F)
the solution is performed using load control.

F
Can Fapp be
achieved with
Fapp
F load control?

F u

29 2013 ANSYS, Inc. July 16, 2014


... Background on Nonlinear Buckling Techniques

Load Control:

The difficulty of using load control with the Newton-Raphson is that the
solution can not progress past a point of instability. At the point of instability
(Fcr) the tangent stiffness matrix KT is singular. Using load control, the
Newton-Raphson method will not converge. However, this type of analysis
can be useful to characterize the pre-buckling behavior of a structure.

Fapp Only Fcr can be


achieved using
Fcr KT = 0 load control.

KT < 0 u

30 2013 ANSYS, Inc. July 16, 2014


... Background on Nonlinear Buckling Techniques
Displacement Control:
When the arch is loaded with an incrementally applied displacement, as
opposed to a force, the solution is performed using displacement control. The
advantage of displacement control is that it produces a stable solution beyond
Fcr. (The imposed displacement provides an additional constraint at the point
of instability.)

UY Fapp can be achieved


with displacement
Fapp
control. (Fapp is now
UY the reaction force at
the imposed
u displacement UY.)
UY

31 2013 ANSYS, Inc. July 16, 2014


... Background on Nonlinear Buckling Techniques
Displacement Control:
The disadvantage of displacement control is that it only works when you
know what displacements to impose! If the arch is loaded with a pressure
load as opposed to a concentrated force, displacement control is not
possible.

P
With a more complicated
loading it is generally not
clear which displacements
to impose.

32 2013 ANSYS, Inc. July 16, 2014


... Background on Nonlinear Buckling Techniques
Dynamics:
A static stability problem using load control, can be solved by using a
nonlinear transient dynamic analysis.
With a dynamic analysis the softening response will not be calculated
when the structure snaps-through (it will be a dynamic snap).
The main disadvantage of dynamics is that it is not always easy to damp
out unwanted dynamic effects (ringing).

Fapp
F Dynamic
Response
F u

33 2013 ANSYS, Inc. July 16, 2014


... Background on Nonlinear Buckling Techniques
Force-controlled with Arc-length method
Typical force-controlled method uses Newton-Raphson (NR) method by default. With
this approach (NR) the applied loads are always increased, causing problems at the
point of buckling.
The arc-length method, however, adjusts the applied load based on a relationship
between the calculated incremental displacement and the arc-length radius, thus
enabling the applied load to decrease, if needed As a result, this can help capture
post-buckling behavior beyond the point of instability.
F
ri
ri Arc Length Radii
Converged Substeps
ri ri
Equilibrium Path
ri
Click to edit Master text styles
u
Arc-length method is not directly exposed in WB-Mechanical, but can be
implemented with command object using ARCLEN, MAXARC, MINARC commands
34 2013 ANSYS, Inc. July 16, 2014
... Background on Nonlinear Buckling Techniques
Nonlinear stabilization
Nonlinear stabilization can be thought of as adding artificial dampers to all of the
nodes in the system.
Before the critical load, the system may have low displacements over a given
timestep this can be thought of as a low pseudo velocity thus less resistive force
from the artificial dampers.
When buckling occurs, larger displacements occur over a small timestep, thus large
pseudo velocity and hence large resistive force. Thus stiffness matrix at the limit
point is not singular.
D
Path OACD is traceable.
State O
A C
State A KT=
0
O Click to editBMaster text styles
State D

Nonlinear Stabilization is available in WB-Mechanical (discussed in next Section)

35 2013 ANSYS, Inc. July 16, 2014


D Nonlinear Stabilization
Nonlinear stabilization can be understood as adding an artificial damper or
dashpot element at each structural element node
The solver calculates a damping force (fd) proportional to the relative pseudo
velocity of the two nodes of the artificial dashpot element.
The pseudo velocity is calculated as a displacement increment divided by the time
increment of the substep

f
d

36 2013 ANSYS, Inc. July 16, 2014


Nonlinear Stabilization
With the dashpot model from previous slide in view:
Any DOF that tends to be unstable has a large displacement increment causing
a large damping (stabilization) force; this force, in turn, reduces the
displacements at the DOF so that stabilization is achieved.
For the DOFs that are stable, the dashpot elements have little effect on the
results because the displacements and the stabilization forces are small relative
to the physical forces.
The coefficient used to calculate the stabilization force is also referred to as the
damping factor.
Although it has the same physical meaning and unit as physical damping, it is
purely numerical in nonlinear stabilization.

37 2013 ANSYS, Inc. July 16, 2014


Nonlinear Stabilization
There are two methods available in Mechanical for controlling the stabilization
force
Energy
Damping factor calculated automatically
Factor can vary from element to element.
Best suited for local instability (i.e. plasticity)

Damping
User defines a damping factor directly.
Same factor used for all applicable elements.
Recommended when user has a specific damping
factor in mind.

For challenging nonlinear problems, it is sometimes helpful to employ both methods,


starting first with the Energy method and switching to the Damping method in a
subsequent restart
38 2013 ANSYS, Inc. July 16, 2014
Nonlinear Stabilization
Energy Method
Energy Dissipation Ratio
Ratio of the work done by stabilization forces to the
element potential energy.
Should be large enough to circumvent the
divergence, but small enough to avoid excessive
stiffness. (Typically between 0 and 1.0)

It is a good practice to examine the energies after the solution has completed
because the energy dissipation ratio of the solution could be greater than the
ratio initially specified.

39 2013 ANSYS, Inc. July 16, 2014


Nonlinear Stabilization
Regardless of which method is employed,
user has general options for how to apply
the stabilization force
Constant Option: Maintains the damping
factor as an unchanged constant during each
substep of a load step.
Reduced Option: Reduces the damping factor
linearly to zero at the end of a load step.

Although the constant option works well in most cases, convergence


difficulties may occur if stabilization is suddenly deactivated in a
subsequent load step. This can have the effect of a step change in loading.
In such a case, use the reduced option.

40 2013 ANSYS, Inc. July 16, 2014


Nonlinear Stabilization
There are also three activation options for
controlling stabilization for the first substep
- NO: Stabilization is not activated (Default)
- ON NONCONVERGENCE: Activate Stabilization only
when min time increment is reached and the analysis
has still not converged
- YES: Activate Stabilization regardless of time
increment for first substep.

In most well posed problems, no stabilization is necessary at first substep


because the structure is initially stable, assuming the time increment is
reasonable.
Stabilization is designed to overcome physical instabilities (i.e. properly
constrained models that fail due to buckling and/or localized material
yielding). It will not resolve numerical instabilities associated with ill
conditioned matrices (poorly constrained structures).

41 2013 ANSYS, Inc. July 16, 2014


Nonlinear Stabilization
In addition, user can receive feedback on
magnitude of Stabilization Forces relative to the
overall imbalance forces
- When the L2-norm of the stabilization force exceeds
the L2-norm of the internal force multiplied by the
stabilization force coefficient, the Solver issues a
message displaying both the stabilization force norm
and the internal force norm. Refer to Section
15.13.2 of Theory Manual for detailed description of
L2-norm calculations.
- Default is 0.2 (=20%)
- The message is intended to alert user to the
presence of an excessive stabilization force in the
run.
- User can choose to tighten or loosen this tolerance
depending on the application

42 2013 ANSYS, Inc. July 16, 2014


Nonlinear Stabilization
When Stabilization has been activated, it is always a good practice to determine
the extent of its influence on results accuracy
One way to do this is to post process the Strain Energy of the system and compare it
to the Stabilization Energy created by the artificial damping introduced to bring
about convergence

As a general rule, the stabilization energy should be small (<<10% of strain energy) in
comparison
43 2013 ANSYS, Inc. July 16, 2014
Nonlinear Stabilization
It is also helpful to examine the reaction forces and moments to gain an
understanding of how much Stabilization forces are introducing fictitious loads
into model

44 2013 ANSYS, Inc. July 16, 2014


Summary
Many structures require an evaluation of their structural stability.
Instability problems usually pose convergence difficulties and therefore
require the application of special nonlinear techniques.
Nonlinear stabilization can be understood as adding an artificial damper or
dashpot element at each structural element node
There are two methods available in WB-Mechanical for controlling the
stabilization force, Energy and Damping.
When Stabilization has been activated, it is always a good practice to
determine the extent of its influence on results accuracy by postprocessing
the stabilization energy and comparing with strain energy as well as
examining the force and moment reactions.

45 2013 ANSYS, Inc. July 16, 2014


... Workshop
Please refer to your Workshop Supplement for instructions on:
W5A- PostBuckling with Stabilization

46 2013 ANSYS, Inc. July 16, 2014