COSMOS - Negative Natural Frequencies 1

[phili22]

I have been conducting a modal analaysis of a long tube hanging vertically. The tube is constrained at one end and free at the other.

For lengths less than 2m, the natural frequencies obtained from COSMOS seem reasonable. However, at lengths greater than 2m, I am getting a negative natural frequency for the 1st mode shape. Can anybody comment on this?

I suspect that the L/D ratio of the geometry is getting too large for the FEA code to solve it properly. The outer diameter of the tube is only 17.3mm with a 2.1mm wall thickness.

See the attached powerpoint for an example of the model. Any help would be appreciated.

 

[GregLocock]

What is the constraint at the 'fixed' end?

Cheers

Greg Locock


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

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

 

[phili22]

The constraint at the end is "fixed" No movement is allowed in any direction with this constraint.

 

[GregLocock]

OK. That's a standard case, you should be able to check the frequencies for your successful runs against a handcalc.

What you are describing sounds unbelievable to me, can you do a diff between the input files for a good run and a bad run to check that nothing odd has changed?



Cheers

Greg Locock


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

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

 

[phili22]

Thanks Greg,

Vibrations was never one of my stronger subjects. Could you direct me to the equation that I can verify the frequencies with by hand cals?

 

[GregLocock]

http://www.amazon.com/Formulas-Natural-Frequency-Mode-Shape/dp/0894648306

Cheers

Greg Locock


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

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

 

[MikeyP]

Are you using beam elements or shell/solid elements? What is the element size? What is the element order?

M

--
Dr Michael F Platten

 

[btrueblood]

Mike,

He's using the embedded Cosmos solver bundled with Solidworks, i.e. only allows auto-meshing in solid elements. Would guess that the mesh is becoming quite distorted if the automesher is not reset/re-run from case to case as the modelled length is varied.

 

[phili22]

I have re-done the mesh at each length and the quality looks good. Its an extremely simple geometry so the auto mesher should have no trouble with it.

I tried to figure out what type of elements I am using and it stated that they are solids. I varied the number of points up to 16 and get the exact same answer.

So, I'm at a loss at this point as to what might be going on. Something is clearly wrong because the hand calcs using a cantilevered beam give me way different answers.

 

[GregLocock]

So do your handcalcs agree with FEA for a static deflection test?

Cheers

Greg Locock


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

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

 

[hacksaw]

the first four criticals for a 2m beam with an ideally rigid support (shear and rotation inertia included) are:

3.7 Hz
23.4 Hz
65.6 Hz
128.3 Hz

 

[hacksaw]

negative frequencies are sometimes associated with buckling, good luck

 

[FeX32]

Negative natural frequencies are not permissible in the real world. But, your mode shape may tell you why you get this.
What are the nodes of the mode shape?


Fe

 

[StrykerTECH]

model it as an extrude thin, instead of a solid extrude. Then it works out well for me. I ran it through simulation and I don't get any negative frequencies.
I have the full package so maybe COSMOS express isn't as robust.

Ryan Marks
Mechanical Engineer

http://www.stryker-tech.com/