piping linear modal analysis and nonlinear support

[r77]

Hello forum,

I'm approaching piping modal analysis and subsequent spectral recombination, and I need some help to undestand this point : modal analysis is a linear tool, so how can it be applied in the very ordinary situations where Y+ vertical supports are present (sleepers, saddles..etc). While a full vertical restraint keeps the problem in a linear domain, a unilateral support will introduce nonlinear boundary conditions, and as a consequence a loss of linearity.
How should I manage this kind of support in a modal analysis???
Is the only solution to carry out a dynamic nonlinear analysis? This would be very unpleasant....
I hope someone can help me

regards

 

[4Pipes]

If you are using something like Caesar then you tell it which load case to use for the modal analysis. If a support is in contact in this load case, it is converted to a bi-directional, no gap linear support. If it is not in contact, the support takes no further part in the analysis.

 

[r77]

Thank you 4Pipes for your explanation,
I understand that if thare is no contact you can neglect the support; I'm less sure you can obtain correct results replacing a simple support with a full vertical restraint.

What about ignoring the support in any case?

best regards

 

[BigInch]

When you have a full vertical restraint, there is a resulting net reaction that is in only one direction of the two possibilities, up or down. The net result is that, even though the restraint is in both up and down directions, the net reaction doesn't care about which restraint direction actually supplied the holding action giving you that reaction and one half of the restraint can essentially be removed. Ie, you wind up with either a "virtual" support, or a "virtual" hanger, never both. If the net reaction was in the +Y direction the full restraint acted just like a suport, if the net reaction was in the -Y direction, it acted like a hanger. Specifying a full vertical restraint just makes it possible for the restraint to act in either direction as needed to supply whatever final net force that is required to make the net vertical movement = zero.

We are more connected to everyone in the world than we've ever been before, except to the person sitting next to us. Lisa Gansky

 

[r77]

I quote BigHinch :
<<there is a resulting net reaction that is in only one direction of the two possibilities, up or down>>,

As far as static analysis is concerned I agree with you. As far as vibration and modal analysis are concerned there's something I don't understand:
in vibrations analysis you neeed to consider both direction at the same time and it's easy to understand that a fully vertical restrained pipe has a completely different vibratory behaviour than a simply supported pipe, hasn't it?
How can you replace a simple support by a full restrain?

 

[BigInch]

OK, I read 4Pipe's response and drifted off into where I thought he was going.

Now I'm not sure. Considering vibrations, it would seem that there must be one distinct mode when in contact and another distinct mode when not in contact, possibly suggesting that there may be a critical case in which one or the other could be ignored, or analyzed separately, or perhaps it is possibly creating some kind of third combination of mode 1 and 2, when a support is in intermitant contact. You could treat the support as a pulsating load first in combination with mode 1, then 2.

A support in intermitant contact might be seen to be a pulsating load, creating either a damping, or a reinforcing action, on other more prominent modes, damping if away from one or the other of those frequencies, or reinforcing if close to one or the other of those frequencies. You might try seeing if you can make any progress from that perspective.



We are more connected to everyone in the world than we've ever been before, except to the person sitting next to us. Lisa Gansky

 

[r77]

Hi BigInch,

As you say the best would be considering separately the two modes. But this becomes soon cumbersome if you have many such supports,as you should calculate all possible combinations.
My idea is, to avoid a nonlinear transient analysis, to ignore the support at all. this would be conservative in terms of safaty because it would lower the eigenvalues of the system respect to the original, so if the excitation remains tha same the dynamic amplification should be higher than in the original case. This is true for a one-degree-of-freedom system. I hope it can be extrapolated to a more complex one.

best regards

 

[BigInch]

Yes, that was my first thought too; just going for the most conservative approach and, if that works, leave it at that. Hopefully it will work and you won't have to go any deeper.

Only put off until tomorrow what you are willing to die having left undone. - Pablo Picasso

 

[4Pipes]

I would guess that the bottom line is that if a dynamic response is important then the system should not use non-linear supports - if only to be able to produce valid calculations. If the calculation implies that pipe will bouncing on the supports, then you need to fit hold-downs or snubbers and try again.

 

[r77]

Ok, but can you assess a priori whether a dynamic analysis is important? can you decide to replace all sleepers with hold-downs just because your client asked for a seismic assessment?
I think such a decision should be backed by calculations, so we are again back to the old question: there is a conservative way to check the safety of a system with nonlinear supports by a linear modal anlysis? Or is a dynamic nonlinear analysis the only way?
And if dynamic analysis is the only solution, does it mean that in the case of piping with saddles or sleepers (i.e. the 99% of the cases) modal analysis is not applicable??

 

[BigInch]

I think if it lifts off, there's eitehr probably a better place to put a support, OR just use a full hold-down clamp is what I'd do.

Only put off until tomorrow what you are willing to die having left undone. - Pablo Picasso