Top Four Mistakes Made Interpreting Analysis Results

The most important part of an analysis is the interpretation of results, also called post-processing. Too much focus is often put on problem setup and solution – load magnitudes and directions, mesh particulars, solver choices, material properties – but all that effort is totally worthless if you misinterpret the results! Here are the most common mistakes, still made by the best of us.

The Singularity

SingularityThe most common source of numerical error is still the most common source of interpretation errors. This is when the solver divides by zero, resulting in erroneously high stresses as it tries to solve for infinity. The most common singularity locations are at zero-radius corners (sharp corners), and zero-area loads/restraints/contacts (edges and vertices), and zero-displacement fixtures.

The best defense is to always run your study a second time, with a different mesh size, and ensure your results are similar. This technique is called “mesh-independent results”, and is the single best way to get confidence in your answers.

Displacement Scale

You might think your parts interfere when they don’t. Or the other way around.

Displacement

Wacky exaggerated expansions and shape changes will usually grab your attention, but linear scaling of linear results may produce a plot that isn’t so obviously wrong and leads you to jump to the wrong conclusions.

Color Me Bad

ColorAt SOLIDWORKS World 2005, the star of American Chopper, Paul Teutel Sr., famously quipped, “Red is bad, blue is good.” True for SimulationXpress, but outside that app it isn’t always so straightforward.

Yes, the default Von Mises Stress will reliably give you red colors at highest stress values, since it always represents a scalar value greater than zero. However other stress types, most notably Normal Stresses and Principal Stresses may have their most significant value in a compressive direction, a negative number that will be blue by default. Similarly, the regions of zero stress might end up in the yellow-green portion of the spectrum.

You can learn more about adjusting color and deflection scales here.

Imaginary High Stress Values

I already talked about singularities. This one is about numerical values that accurate, but are wrong because they are too large for real life. I’m referring to stresses above yield. Designers in a hurry may glance at where the stress concentrations appear and forget to double-check the stress values. If you have any values that exceed the yield stress of your material, then your study violates your assumptions of linear static equilibrium. You need to perform a nonlinear plasticity study, or at least view your results with a very high degree of skepticism. All you really know is that your part probably yields; you don’t actually know where or how, because the FEA is simply finding force over area that balances equilibrium. It doesn’t matter to the solver if that equals 1,000,000 psi and your aluminum can only take 50,000 psi.

Stress

A nice tip is to change your settings so values above yield are given a different color entirely, like gray. Go to Simulation – Options – Default Options – Plot – Color Chart to turn on that setting which applies to vonMises plots.

Stress

CAPINC
CAPINC was founded on one core principle: Provide the best solutions and services to assist our customers in designing and developing better products. CAPINC provides premier solutions and services in New England to assist our customers in accelerating their design and development process for better mechanical products. Our award winning technical support team is comprised of industry experts with hundreds of years of combined practical experience in mechanical design, design validation and analysis, product data management, and technical communication. We are the award winning 3D solutions partner offering SolidWorks software and training, and the entire Stratasys line of 3D printers and production systems. For more tech tips and blog tutorials check out CAPINC's blog and videos.
CAPINC