Vibration Specification - Correct Way to Complete Analysis

[rhmeng]

We receive PSD curves for our vibration specification requirements. The curves are given for each axis (X,Y,Z). I have always set up a linear dynamic study for each axis (one study for X, one for Y and one for Z) and ran a study for each axis. I then look at the stress for each axis to make sure none of the structural members are failing.

My question, is should I be setting up the analysis so that each axis is ran in the same study? I have always kept each axis separate, this is how i was taught. But now I am thinking I should have a base excitation with a PSD input curve in all three axes in the same study. In real life the axes are not separate so why should they be in the FEA. I would imagine I would get much higher stresses if I do it this way also.

Any input is appreciated.

 

[JoshPlumSE]

So, it depends on what type of structure and analysis you're doing. Also, what seismic design category you're in.

ASCE-7 for example says that for Design Category C (and above) that you should combine 100% of the seismic force in the primary lateral direction with 30% of the force for the other horizontal direction. And, it sort of ignores the vertical effect (not completely, but it assumes that about about 20% of the lateral design force in the vertical direction).

But, if you're doing nuclear work, there is usually either a SRSS (square root sum of squares) technique when apply all three directions at once or a 100%+30%+30% to cover all directions.

I should note that the reason why the codes combine the multiple directions is because of the idea that the earthquake can occur in a direction that is not based on the orthogonal axes of your structure. Kind of similar to the wind design flaw that caused all the concern for the Citicorp Center in the late 70s.

 

[driftLimiter]

If the structure were free from coupled vibration modes, and elements resisting vibrations from each direction. A 1 D analysis in each direction captures the peak.

However if the structure has coupled modes or elements that resist vibrations from each directions. Some combination of the vibrations SRSS or 1X + 0.3Y for example aught to be used.

 

[rhmeng]

So the structure is a product that we design that will essentially go on a rocket. The PSD curves are for a flight environment. We are not given requirements on how to run the FEA (ie 1X + 0.3Y, etc).

I do not see how our structure could be free from coupled vibration modes. When analyzing the stresses in a screw for example, the screw will see the stresses from each axis all at the same time, regardless of if the modes are coupled or not.

I think I really need to understand how the PSD curves were derived to understand how to analyze the system correctly.

 

[driftLimiter]

I've mostly seen SRSS or CQC methods used to combine. However I've delt mostly with earthquake excitation which there is a low probability of the peak excitation frequency occurring in each direction simultaneously.

 

[SWComposites]

rhmeng - this forum is mostly for civil structures engineers. suggest you delete this thread and repost in the Spacecraft Engineering forum.