Resonance Frequency-modal Analysis

[ataras]

Using MSC Nastran 103 to calculate BIW (car's sheet metal platform only)resonance frequency by FE method, do we need to apply any constrains?
I think, a free modal analysis means no constrains, however, final outcome (Hz)will depends on where forces are applied to the model ? Cross car torsional stiffness frequency will be diferent from front to back. Please, advise.

 

[FredGarvin]

I would think, at the bare minimum, you would constrain the panel as it would be in the installation. A panel confined around its edges is going to act differently than a free floating one. Just thinking of the solving of the differential equations, you need to specify the BCs to get a particular solution.

 

[Mech_Engineer]

To get results that are worth something, you need to constrain the model as it will be constrained in real-life. While it might be possible that the solver will solve without constraints (depends on the software) the results are meaningless. You'll end up with an infinite number of vibrational modes that CANNOT happen in real life because there should be a constraint there.

Think of the possible differences just on a simple structure like a standard rectangular beam. Simply supported, single cantilever, or double cantilever will all give you far different modes of vibration, yet the structure looks exactly the same in each case, only the boundary conditions have changed.

 

[ataras]

Thanks, location of excitations points is crucial. Every time you change them you do get different results. Point I was trying to make is that you induce vibration through excitation points on particular modes without having to constrain the object.

 

[Mech_Engineer]

Thanks, location of excitations points is crucial. Every time you change them you do get different results. Point I was trying to make is that you induce vibration through excitation points on particular modes without having to constrain the object.

If you're doing a modal analysis, you shouldn't need to have any excitation points defined. The program should solve for the natural modes of vibration without them; but boundary conditions are crucial. This solution should show you the nature of the mode, and it's modal ferquency.

If you're doing a frequency-response analysis, you will need to define an excitation force, a range of frequencies, AND constrain the model fully. This solution can show you the model's reaction to an input vibration. As before, if the model is unconstrained you will not get any results that are useful (rigid-body motion, modes that don't exist in the constrained model, etc).

Basically, no matter what, if you want to believe the results you're getting, you need to constrain the model as it will be constrained in real-life. There is a VERY fine line between getting numbers, and getting numbers that make sense in FEA.