Continue to Site

Eng-Tips is the largest engineering community on the Internet

Intelligent Work Forums for Engineering Professionals

  • Congratulations IDS on being selected by the Eng-Tips community for having the most helpful posts in the forums last week. Way to Go!

Comparing ANSYS Modal Analysis to Experimental Sine Sweep

Status
Not open for further replies.

Transient1

Mechanical
Jan 31, 2007
267
My modal analysis indicates participation factors of less that .01 for X,Y,X-rotation and Z directions. It has a participation factor for the first mode of .37 in the Z-rotation and -.41 for the 3rd mode in the Y-rotation.
The first three modes are 919 Hz,1042 Hz,1562 Hz.
The Sine sweep does not show these resonances, but shows resonances at about 400-600 Hz depending on which axis is being monitored. Of course, some of my modeling assumptions could be wrong or my mesh not refined enough (which I don't think is the issue). My question is, if in the range of interest there are only dominant rotational modes, how do they present themselves in a sine sweep X,Y,Z axis response graph. Is there some constitutive relationship between the resonance peak in three axes that would add up to the modal analysis results?

What would be a good reference for vibration FEA? Is there any book that clearly spells out the do's and do nots of such an anlysis?
 
Replies continue below

Recommended for you

Hi Transient1,
Im not an ANSYS user, Im starting to learn it this week, however, Im a Pro Mechanica user and I have corelated between the software and data. Just from a FEA point of view, it looks like your ANSYS model is stiff and your UUT is a little looser than what you have in your model. Are there any fasting points in your structure? Some times in FEA models if you have three parts put together, the software will fuse it all together as if it was acting like one part. But, in reality, where there are fasting points, this will cause the structure (UUT) to be more loose in vibration mode. It is either some how make your model "looser" or make the UUT "tighter". Now, if your fasteners are at it's proper torques, then it means you have to "loosen" you model some how till you get the frequency that was tested.

Remember the freqeuncy from test is your ansewer and the frequency from your model was your best guess. Some people make the mistake that the FEA study is the final ansewer and at the end(at qual)get bit in the a$$. The FEA study is half the ansewer, the other half is on the vib table.

Good luck

Tobalcane
"If you avoid failure, you also avoid success."
 
Thanks Tobalcane,

That answer makes sense to me. What are you abreviating with UUT? So what you are saying is that for a panel held by fasteners around the perimeter the answer can vary greatly by considering the local bonding at the fastner location rather than fixing the mating surfaces.
 
Transient1,

Oh sorry..UUT is Unit Under Test.

"...considering the local bonding at the fastner location rather than fixing the mating surfaces. "

Yes. So if you had a box fasten to a wall, in a FEA program, don't have the whole surface of the back of the box constrained, just where the fasting points are located. This will "loosen" your model a little and may bring the frequncy down to the emperical data.



Tobalcane
"If you avoid failure, you also avoid success."
 
to add, it also depends where you place your response accelerometers. From your FEA medel, determin where you predict the displacemnts will occur. Place your accel at thoughs locations. The frequencies should match up.

Tobalcane
"If you avoid failure, you also avoid success."
 
Status
Not open for further replies.

Part and Inventory Search

Sponsor