modal test phase shifts

[hacksaw]

traditional modal testing involves the sinusoidal steady state with the mode shapes more or less easily identified, mode enumeration is another discussion altogether, but when the modal testing is performed by means of impulse testing (tap tests) the response is no longer a matter of sinusoidal steady state, so the equestion is this

how is the phase shift associated with sensor placement interpreted?

 

[GregLocock]

Sorry, I don't understand your question. The results, on a linear system, are identical, in theory.



Cheers

Greg Locock


New here? Try reading these, they might help FAQ731-376

http://eng-tips.com/market.cfm?

 

[hacksaw]

modal analysis describes steady-state condition as I understand it, i.e. time is completely absent,

if I do a tap test with null damping, i should get the same result,

the problem is per an earlier posting in this forum, is what happens when the damping is finite but small, and the wave length of the excitation comparable to the dimensions of the structure being tested.

my quandry is that time is now a factor in the measurements.

so how does this affect the accuracy of the measurements and sensor placement

 

[GregLocock]

Well practically, it doesn't. Take a Body in White. First bending is at around 25-40 Hz, wavelength is obviously about one car length. You can hit it with a 300g hammer, or shake it with a shaker. You'll get the same mode shape, and it'll be lightly damped.

Propagation time as such isn't an issue, but I still don't see what you are driving at. Perhaps you are puzzled by a possible time delay between the driving point acceleration and the point at the other end of the car? That is, how does the other end of the car 'know' that it has to move in phase with the driving point?



Cheers

Greg Locock


New here? Try reading these, they might help FAQ731-376

http://eng-tips.com/market.cfm?

 

[hacksaw]

that's basically the question

in dealing with transient excitation and using fourier analysis to reduce the time series data, it seems that the propagation delays between the source and sensor are going to give rise to phase shifts

thats what puzzles me

 

[GregLocock]

It is a neat question, now I've had a think.

How about we set up a 3 mass 2 spring model, masses are 1,2,1 kg, and springs are 10^5 N/m.

The mode shapes are obvious.

Now set up a time based sim of that and see how the third mass responds when the first is struck.

I haven't done this but when i get a free cpu I'll try it.



Cheers

Greg Locock


New here? Try reading these, they might help FAQ731-376

http://eng-tips.com/market.cfm?

 

[hacksaw]

followup:


for old school types, the modes with phase shift are classified as "complex modes", basically the dynamic mode shapes with damping present. They are in the form of traveliing waves.

"Normal modes" are the stationary modes with identifiable nodes with fix phase relationships, either in-phase (0deg) or out-of-phase(180Deg). They are for the damping free case only.