Seismic analysis

[Anonymous]

Which is the correct way to perform a seismic analysis with Nastran in Cad?

I ask this because I read some different methods and it seems there is not a lot of documentation about this topic, at least not in the several websites that I visited.

By the way, I see 3 ways. I briefly resume them here with the essential steps:

• Choose a Response spectrum analysis; set the spectrum; set the number of modes that involve at least 85% of the participating mass; define damping; apply the constraint “spectrum response”; connect a heavy concentrated mass to the part in order to simulate the surrounding mass. The results are stress and displacement that the earthquake causes.

• Choose a Random Response analysis; set the number of modes that involve at least 85% of the participating mass; define damping; apply the constraint; apply an enforced motion (acceleration) along the axis where the earthquake takes place; define a PSD. The result RMS tells the Root Mean Square of the excitation in terms of stress and displacement; the NPX tells which frequencies cause the result shown by RMS.

• Choose a Linear Static Analysis; take note of the highest acceleration given by the spectrum of your seismic spectrum (plateau) and insert it as static acceleration load (setting it as a component of the gravity load).

I’ve also seen a brief Tutorial about Response Spectrum with InCad and it tells what I summed up at the point 1, but I’m also interested if the other two approaches could be correct.

Thanks in advance

Luca

[KubliJ]

Hi @Anonymous,

I am familiar with the second two analysis types as used for a earthquake analysis. 

I would like to note though that the setup (large mass) for the first method can be used in the random response.  The advantage of this method is that you can use simple modal damping (which isn't allowed in enforced motion based random analysis).  The different damping methods have different effects and we can definitely discuss that further

As to which one is correct, based off of my knowledge, most if not all earthquake data is PSD based, so I would say the Random Response analysis is the most accurate method.  Something to note about the results is that these will be a statistical result.  So the stress reported will be within 1 sigma of confidence.  If you multiple your values by 3, then you will have more confidence in the max stress the system may experience in an actual random vibration event.

The linear static method is a gross over approximation of the dynamics of the system.  I've never validated an analysis like this, but in theory, if your model passes this method, it should be more than strong enough for the actual vibration.

Thanks,

James