Lecture 1

Overview
14. 0 Release

ANSYS Mechanical
Basic Structural Nonlinearities
1 © 2011 ANSYS, Inc. May 23, 2012
 Chapter Overview
In this chapter, an overview of the basics of nonlinear finite-
element analysis (FEA) is presented:

A. What is “Nonlinear” Behavior?

B. Types of Nonlinearities
C. Nonlinear solution using linear solvers
D. Nonlinear FEA issues

The purpose is to give you an understanding of the fundamental nature of

nonlinear FEA.
The capabilities described in this section are generally applicable to Structural
licenses and above.

2 © 2011 ANSYS, Inc. May 23, 2012

 A. What is nonlinear behavior?
Recall, in the 1600s, Robert Hooke discovered a simple linear relationship
between force (F) and displacement (u), known as Hooke’s Law:

F = Ku
• The constant K represents structural stiffness.
• A linear structure obeys this linear relationship.
• A common example is a simple spring:
F
F
K
K u
u
• Linear structures are well-suited to finite-element analysis, which is
based on linear matrix algebra.

3 © 2011 ANSYS, Inc. May 23, 2012

 ... What is nonlinear behavior?
Significant classes of structures do not have a linear relationship between force
and displacement.

Because a plot of F versus u for such structures is not a straight line, such
structures are said to be nonlinear.
• The stiffness is no longer a constant, but varies as you progress through the load path
• KT (tangent stiffness) represents the tangent to the force deflection curve at a particular
point in the load path.

F KT

4 © 2011 ANSYS, Inc. May 23, 2012 u

 ... What is nonlinear behavior?
A structure is nonlinear if the loading causes significant changes in stiffness.

Typical reasons for stiffness change are:

• Strains beyond the elastic limit (plasticity)
• Large deflections, such as a loaded fishing rod
• Changing Status (Contact between two bodies, Element birth/death)

5 © 2011 ANSYS, Inc. May 23, 2012

 B. Types of Nonlinearities
There are three main sources of nonlinearities:
• Geometric nonlinearities: If a structure
experiences large deformations, its
changing geometric configuration can
cause nonlinear behavior.

• Material nonlinearities: A nonlinear stress-strain

relationship, such as metal plasticity shown on
the right, is another source of nonlinearities.

• Contact: A “changing status” nonlinearity, where

an abrupt change in stiffness may occur when
bodies come into or out of contact with each
other.

6 © 2011 ANSYS, Inc. May 23, 2012

 … Types of Nonlinearities
Of course, all three types of nonlinearities can be encountered in combination.
Mechanical can readily handle combined nonlinear effects.

Rubber Boot Seal

An example of nonlinear
geometry (large strain and
large deformation),
nonlinear material
(rubber), and changing
status nonlinearities
(contact).

7 © 2011 ANSYS, Inc. May 23, 2012

 B. Nonlinear solution using linear solvers
How does Mechanical solve for a changing stiffness?
• In a nonlinear analysis, the response cannot be predicted directly with a set of linear
equations.
• However, a nonlinear structure can be analyzed using an iterative series of linear
approximations, with corrections.
• Mechanical uses an iterative process called the Newton-Raphson Method. Each iteration
is known as an equilibrium iteration.

Load

F A full Newton-Raphson iterative

4
3 analysis for one increment of load.
2 (Four iterations are shown.)

u Displacement

8 © 2011 ANSYS, Inc. May 23, 2012

 … Nonlinear solution using linear solvers
The actual relationship between load and displacement (shown
with a blue dotted line) is not known beforehand.

Consequently, a series of linear approximations with corrections is performed.

This is a simplified explanation of the Newton-Raphson method (shown as
solid red lines)

• In the Newton-Raphson Method, the total Newton-Raphson Method

load Fa is applied in iteration 1. The result Fa

is x1. From the displacements, the internal 4
3
forces F1 can be calculated. If Fa  F1, then
the system is not in equilibrium. Hence, F1
2

a new stiffness matrix (slope of dotted line) is

calculated based on the current conditions. 1
The difference of Fa - F1 is the out-of-balance
or residual forces. The residual forces must
be ‘small’ enough for the solution to converge. x1 x
• This process is repeated until Fa = Fi. In this example, after iteration 4, the system achieves
equilibrium and the solution is said to be converged.
9 © 2011 ANSYS, Inc. May 23, 2012
 … Nonlinear solution using linear solvers
• The difference between external and internal loads, {Fa} - {Fnr}, is called the
residual. It is a measure of the force imbalance in the structure.
• The goal is to iterate until the residual becomes acceptably small; that is,
until the solution is converged.
• When convergence is achieved, the solution is in equilibrium, within an
acceptable tolerance.

Fa
{ Fnr
{Fa} {Fnr}

u
10 © 2011 ANSYS, Inc. May 23, 2012
 ... Nonlinear solution using linear solvers
The Newton-Raphson method:
• Is not guaranteed to converge in all cases!
• Will converge only if the starting configuration is inside the radius of
convergence.

Load Load
Diverging!
Converged

F F

ustart u Displacement ustartu Displacement

Starting outside the Starting inside the radius

radius of convergence of convergence
11 © 2011 ANSYS, Inc. May 23, 2012
 ... Nonlinear solution using linear solvers
Two techniques can help you obtain a converged solution:

F
F

F1

ustart u ustart u
Apply load incrementally to move Use convergence-enhancement
the target closer to the start tools to enlarge the radius of
convergence

Mechanical combines both strategies to obtain convergence.

12 © 2011 ANSYS, Inc. May 23, 2012

 ... Nonlinear solution using linear solvers
As a general rule, sudden changes to any aspect of a system will cause
convergence difficulties.
With this in mind, it is useful to understand how loads are managed
• Load steps differentiate changes in general loading.
– In the Figure at the bottom right, Fa and Fb are loadsteps.
• Substeps apply the loads in an incremental fashion
Fb
– Because of the complex response, it
Fb2
may be necessary to apply the load
Fb1
incrementally. For example, Fa1 may be
Fa
near 50% of the Fa load. After the load
for Fa1 is converged, then the full Fa load Fa1

is applied. Fa has 2 substeps while Fb

has 3 substeps in this example xa xb
• Equilibrium iterations are the corrective
solutions to obtain a converged substep
– In this example, the iterations between the dotted lines indicate equilibrium iterations.
13 © 2011 ANSYS, Inc. May 23, 2012
 C. Nonlinear FEA Issues
Three main issues arise whenever you do a nonlinear finite element analysis:
• Obtaining convergence
• Balancing expense versus accuracy
• Verification

It takes care and skill to

juggle these three
issues successfully!

14 © 2011 ANSYS, Inc. May 23, 2012

 … Nonlinear FEA Issues
Obtaining convergence…
Usually your biggest challenge.
Solution must start within the radius of convergence.
• The radius of convergence is unknown!
– If solution converges, the start was within the radius.
– If solution fails to converge, the start was outside the radius.
• Trial-and-error is sometimes required.
• Experience and training reduce your trial-and-error effort.
Difficult problems might require many load increments, and many iterations at
each load increment, to reach convergence.
• When many iterations are required, the overall solution time increases.

15 © 2011 ANSYS, Inc. May 23, 2012

 … Nonlinear FEA Issues
Balancing expense versus accuracy…
• All FEA involves a trade-off between expense (elapsed time, disk and memory
requirements) and accuracy.
• More detail and a finer mesh generally lead to a more accurate solution, but
require more time and system resources.
• Nonlinear analyses add an extra factor, the number of load increments, which
affects both accuracy and expense. More increments =improve the accuracy,
with increase the expense.
• Other nonlinear parameters, such as contact stiffness (discussed later), can also
affect both accuracy and expense.
• Use your own engineering judgment to determine how much accuracy you
need, how much expense you can afford.

16 © 2011 ANSYS, Inc. May 23, 2012

 … Nonlinear FEA Issues
Verification…
• In a nonlinear analysis, as in any finite-element analysis, you must verify your
results.
• Due to the increased complexity of nonlinear behavior, nonlinear results are
generally more difficult to verify.
• Sensitivity studies (increasing mesh density, decreasing load increment, varying
other model parameters) become more expensive.

Stress

Mesh Density
Typical Sensitivity Study
Later chapters will provide modeling tips for different nonlinear situations.

17 © 2011 ANSYS, Inc. May 23, 2012