Determining stiffness of a model from natural modes

[GenericUser]

Hello,

I am trying to determine the stiffness of a part I simulated with enforced displacement. My first method was to use Hooke's law F=kx but applying an external non-pressure force was problematic in Abaqus. So I think I may have found a way using modal frequencies, but I have doubts if my method is sound.

Basically any continuous part has infinite number of natural modes. So w=sqrt(k/m) is only one of many frequencies. What I did is animate each mode until I found a mode shape that closely resembles the deformation pattern I am looking for and then calculating w using w=(2*pi*f) and then calculating the stiffness k by saying k = mw^2. Is this method correct or is it wrong to apply discrete vibration theory to a continuous part in this manner?

 

[ESPcomposites]

That doesn't sound too reliable to me.

In Patran there is something called a freebody tool. You can get the summation of forces for a section cut (and other free bodies). Then it is just F=kx.

I am not familiar with the ABAQUS equivalent to the Patran freebody tool, but I would look for something that does the same thing.

Brian

http://www.espcomposites.com

 

[GregLocock]

That will produce a number that may be meaningful in some sense, but I struggle to think what.

Applying a static external force is FEA101.

Cheers

Greg Locock


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

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

 

[GenericUser]

I tested it with a cantilever beam, used my hand calcs and found the numbers to be way off.

A static external force would be the first option. While I am new to Abaqus, I have not found any option that would allow a user to assign non-pressure force on anything other than a point or edge (no surfaces). Using those cause very large localized displacements that do not propagate to the rest of the model.

I'll find another way to address the problem, but thanks for the responses.

 

[GregLocock]

Good, I'm glad that applying infinite pressure to a part of a model causes large local deformation.

So what you can do is apply a pressure load to an area, or apply your load to several nodes, or, and this should be regarded with total disdain, use a spider of rigid elements to distribute your point load.

Incidentally there is a technique to use modal results to establish static stiffness, as the summed response of /all/ the modes referred back to 0 hz is the static stiffness, it does work, but I've never done it in FEA.

Cheers

Greg Locock


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

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

 

[GenericUser]

Oh I wasn't surprised about the effects of the point load, I was working with the options available.

The problem with pressure loads is that by definition they are normal to the surface they are applied. Works fine for flat surfaces, but for non-flat surfaces where you don't want the force to be always normal, its a pain.

I understand why the rigid-spiderweb-links would not be the best idea. Does Abaqus have weighted links? I could not find any reference to it, but I am guessing the option might exist under a different name.

 

[GregLocock]

Sorry, i'm not an abaqus person, I don't know.

Cheers

Greg Locock


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

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