3.3 Flowchart For Troubleshooting Nonlinear Static Analysis
3.3 Flowchart For Troubleshooting Nonlinear Static Analysis
1. Run your model in a linear static solution and make sure it completes and that the results appear
correct.
2. If it runs okay in linear static, run the model as a nonlinear analysis and use the VIS solver. The
VIS solver is chosen in the Autodesk Nastran Editor options (on the left side) under Program
Control Directives – DECOMPMETHOD (double click to change). Also change
MAXSPARSEITER to 500 iterations to shorten the analysis time. The goal is to force a solution
to help diagnose the problem. Unlike all other solvers the VIS will not produce a fatal error for an
ill-conditioned stiffness matrix.
3. On the NLPARM card, change the number of iterations to 1 (field 3) and change the maximum
number of iterations to 1 (field 7).
4. Run this analysis and look at the results. Please keep in mind this is only a diagnostic run and
should not be used as actual results. If you see an area or node with a large amount of
displacement/stress, this is the probable cause of the fatal error. This may be caused by badly
distorted elements.
WARNING: E5074: No
Convergence in Increase the maximum number of iterations
Maximum Number of
Iterations Permitted
WARNIN: E5079:
Increase the number of increments and/or increase the
Excessive Element
Stress Fraction Limit (FSTRESS on NLPARM)
Internal Loads Detected
NO
S1110 Use 64-bit Autodesk Nastran to overcome
Insufficient memory issues.
Surface Contact in
Memory error
Model?
Occurs
Analysis is
Surface Contact in Try varying the solvers as
running NO
Model? described in Section 9.
slow
Autodesk Nastran 11
Autodesk Nastran 2022 Nonlinear Analysis Handbook
If the effects of inertia, damping, and transient loading are significant, then a nonlinear transient analysis
should be used. Additionally, “quasi-static” models that undergo buckling or other instable loading
conditions will often converge better in a nonlinear transient analysis due to the inertia effects keeping
the model stable.
An important element to having a stable NLT (nonlinear transient) solution is to provide damping in the
model. There are two types of damping that can be applied in NLT solutions. The first type is a global
damping value specified using a PARAM,G followed by a PARAM,W3 which defines the frequency at
which to apply the damping (see the Autodesk Nastran User’s Manual, Section 5 for more detailed
information on damping). The second type is material based damping and is defined on each material
card directly. PARAM,W4 is needed to define the frequency at which to apply the material based
damping. Note that the units of W3 and W4 are radians per unit time. The increased flexibility of
material based damping (i.e., different damping values can be applied to different areas/materials of the
model) makes it the logical choice for NLT analysis.
A note of caution when using damping in a NLT solution is that for models where the velocity/inertia is
the main driver of the analysis such as in an impact solution, damping can have a significant effect on the
acceleration/velocity/displacement of the model. This is because the solver cannot make a distinction
between rigid body motion/velocity, and flexible motion/velocity, so the damping is applied to any part of
the structure that has a velocity. For impact analysis it is recommended to use no damping or a small
“stability” damping value (i.e., 1.E-6).
There are a few guidelines to follow when performing an impact analysis that will have a large impact on
solution time and quality of the results.
This is a very important and often overlooked stage. We need to know the linear response
characteristics of the structure to get some idea of what the actual nonlinear frequencies and mode
shapes are going to be. It can never be an exact representation, but it gets us in the right ball park for
several key input parameters:
Constrain (fully fixed) the area of the model that you expect to make contact with the ground (or other
impactor) and run a normal modes solution with ~20 modes. Look at the mode shapes and find the
mode you would consider to be the “dominate” response of the structure during/after impact. A look at
the modal effective mass table in the *.OUT file may also help determine the critical mode. The
frequency of the mode can be used to calculate the key input parameters above:
• Frequency range of interest – This would be the frequency of the dominant mode.
• Size of time step – This can be calculated using 1/f, and then assuming 100 data points per cycle
would net: dt = 1/ (100*f).
• Duration of analysis – This largely depends upon the velocity of the impact, the size of the model,
and the flexibility of the model. A good estimate is to run the analysis for 2-5 cycles.
Autodesk Nastran 12