Reply To: Reliability of LS-DYNA output graphs

[peteroznewman]

Hi,

There are many questions to answer before this data can be cleaned up.

1. What units are in the chart?  Is the X axis in seconds?
2. What is the measured mass of the physical structure?
3. What is the mass of the Finite Element Model?
4. Run a Modal Analysis to learn the undamped natural frequencies of the structure.
5. Show the input load time history.
6. Plot the frequency content of the input load such as a Shock Response Spectrum. What is the primary frequency?
7. Show the boundary conditions. How is the structure supported?
8. What time step was used when LS-DYNA was solving?
9. What time step was used between writing output data?
10. What damping was applied to the model?

If the time units are in seconds, you might consider running the simulation in an implicit transient structural solver instead of an explicit dynamics solver such as LS-DYNA.  The reason is that explicit dynamics results contain a lot of noise and special care is needed to filter that noise.  The implicit solver results do not have a lot of noise and may not need any filtering.

Here are some resources talking about why explicit dynamics solutions have noise and a link to a NASA paper on how to filter that noise.

https://feassistant.com/forums/ls-dyna/generic-question-about-when-to-use-a-filter-for-xy-data-in-ls-prepost/

https://feassistant.com/forums/ls-dyna/remove-oscillations-in-nodfor-output-data/