Negative and 10E-5 frequency in modal analysis.....why?

[sbernardini]

Hi to all,

i'm trying to learn how to use femap.
I was making a normal mode analysis of a structure composed by 2 octagonal plates and four long bolts.
To set the connection between bolts and plates, i'm using the following procedure:
MESH/CONNECTION/COINCIDENT LINK
Then i choose rigid connection and mark all the followings:
Tx, Ty, Tz, Rx, Ry, Rz
I choose coincident link because i have 2 coincident nodes which are one at the end of the beam representing the bolt, and the other node is the corresponding, on the octagonal plate. I created this 2 different nodes by creating nodes on poit function.
If that is the right procedure to simulate the bolt and plate, why am i getting 4 negative frequencies and why are these that close to zero , in fact they are about 10E-5
Is that a free body behaviour?
Is that correct that only from the fifth mode i get an expected behaviour?.
Why do i get these 4 neg. frequencies only when i constarin points on the plates?
In fact if i constrain any node on any of the four beams, i mean the bolts, i get all natural modes correct with positive and above zero frequencies (not smaller than 30Hz)
Thanks to all in advance.
I can post the problem in my web site to better illustarte the problem and the structure

sbernardini
Rome, Italy

http://digilander.libero.it/sbernardini

 

[pjhype]

Nodes on plate elements do not have stiffness in the rotational degree of freedom normal to the plate. Most finite element codes will automatically constrain these degrees of freedom. I suspect that when you connected beam elements to the plates through rigid elements, your solver did not automatically constrain the plate normal rotation causing the apparent rigid body modes. Alternately, you could connect the plates to the beams in all degrees of freedom except for the rotational degree of freedom normal to the plate. Your solver should then recognize that there is not rotational stiffness and automatically constrain this degree of freedom. If not, you will need to add a permanent constraint to this degree of freedom.

pj

 

[sbernardini]

Hi pjhype,
you have no idea of how thankfuul i am that you replied me....thanks.
Well, i will try not setting the Rx, Ry and Rz and see what happens.
I'd had many question to ask you about femap...if you are not too busy, please tell me if i could ever ask them.
I'll try some more tests and then reply again....i'm in big trouble because i need to set the model within few days and this is my first experience.
Ciao sbernardini
Rome, Italy

http://digilander.libero.it/sbernardini

 

[pjhype]

sberardini,

You should un-constrain the rotation normal to the plate only (not all of the rotations). If your bolts are in-line, you will define a hinge and get a rigid body rotation mode.

In order to determine whether you need to model the bolts, you should first constrain the plates to each other with rigid elements at the bolt locations. This will give you an upper bound on your analysis. Typically, with (thin) plates, the local bending stiffness of the plates around the bolts is several times lower than the axial or shear stiffness of the bolts and you might be able to ignore the bolts in future analyses.

pj

 

[sbernardini]

Hi again,

well i really miss that we are not neighboors...he he
I guess you are very far from Rome....and that is really too bad...he he
some fun just to have a better time than what i'm really having now with this model.....
I'll keep trying...using your suggestions

Some more info's:
the 4 bolts are long about 25cm and the 2 (top and bottom) plates are octagonal shapes (each side 15 cm)and they are 1cm thick. It means these 2 plates are separated by 23cm and they are kept in this distance by 4 spacers coaxials with the bolts.
I'll try to add a pic of the structure and give you the link

sbernardini
Rome, Italy

http://digilander.libero.it/sbernardini

 

[corus]

When a zero frequency is calculated then this is due to rigid body motion as has been explained. The fifth mode calculated will in fact be the first natural frequency of the system if you had the correct restraints. Your results are still valid however although the mode number is incorrect. Modifying the bolt connections may remove the zero frequencies but rigid body motion of the whole system may still occur if the whole system is not restrained in some way.

 

[sbernardini]

Thanks to you too Corus,

well i have been discovering many new things in a few days...very interesting.
I have been given a task probably bigger than what i was supposed to be doing.
I thought that to simulate the 4 bolts it would have been enough to get the end node of the bolt (using a beam element) and make a rigid connection to the equivalent node on the base (Using a plate element)
I have seen how it cannot be done in that way.
Following pjhype and your suggestions i am getting closer to understand it and make it work....
I have seen other people asking for bolt connections and i'll read the other topics too....too bad i have very few time left
sbernardini
Rome, Italy

http://digilander.libero.it/sbernardini