PSD Random Vibration

[Anonymous]

1)  When inputting a PSD vibration spectrum the graph is plotted linear/linear. PSD is more commonly input as log/log.  Can it be shown as a log/log?

2) The resultant stress levels seem to be 1 sigma or 1 standard deviation.  Is that correct?  Can I plot 3 sigma stress

[John_Holtz]

Hi klinde,

The answer to both questions is no.

I think those ideas would be good suggestions on the Ideas page.

[Anonymous]

I assume by your response that you agree with my statement that the output is 1 sigma stress.

Thanks,

  Karl

[John_Holtz · Accepted solution]

Hi Karl,

Yes, I agree that it is one sigma (one standard deviation). Sorry I missed that part of the answer before

[pier-olivier.duval · Accepted solution]

To consider 3 sigma stress, you just need to multiply your stress by a factor of 3...  it is not exact for the non vectored value (as VonMises) because it's a resultant scalar computed from the vectored values.  Nerveless, it is considered conservative in the industry to consider the 3*VonMises as the 3 sigma stress criteria.

Good thing to know...  You need to use the stress in conjunction with the positive crossing number (NPX set in Nastran).  This is giving to you the frequency at which the stress is occurring.  Using both this results, you can compute your cumulative damage, element per element.  If you use Femap (like us) you can rapidly create an API to compute and generate a graph of the cumulative damage.

Feel free to ask if you have question relatively to vibration and dynamics analysis

Pier-Olivier Duval

FEA/CFD Manager

Ametek - Creaform

[Anonymous]

Pier-Oliver Duval,

  Thank you for your help. That was a very informative reply.

Karl Linde