03-06-2015 03:00 AM
Hello.
I want to use a boltfor preload in a sol 103/111.
To test the effects I've created a test model with only one beam mesh with a fixed constrain on both ends and solve this model in a sol 103. With an additional preload I expected that the modes are the same and the frequencies increase.
But the results looks strange. The modes has changed and aren't symmetric anymore. Some modes get lost so instead of 10 only 8 modes are calculated and the modes and frequencies are dependent from where the bolt preload is defined.
Using an enforced displacement that enlarge a shorter beam mesh to the length of the pre-loaded one (the resulting axial force is comparable) brings the expected results.
In my opinion the model with the bolt preload should create the same results like the model with the enforced displacement.
What can I do to succesfully use bolt preloads in dynamic Solutions?
Solved! Go to Solution.
03-08-2015 07:48 AM
Dear Andreas,
In my opinion NX NASTRAN runs correctly as expected. Here you are the first mode shape f1=915 Hz, being Mode#1 and Mode#2 orthogonal bending modes of identical value.
After applying the bolt-preload effect the shape of mode#1 is an axial mode caused by the bolt preload, but if you plot Mode#2 and Mode#3 you will see orthogonal bending modes of value 26490 Hz. Bolt-preload modifies not only the value of regular eigenvalues (from 915 Hz to 26490 Hz, impressing difference!!), but also the mode shape.
For linear static analysis (SOL101), when BOLT PRELOAD is prescribed with the NX Nastran solver the entire run is automated and allows for direct entry of the bolt preload forces in the FE model. During the run, the model is solved twice.
• The first solution calculates the deformed shape of the bolted medium resulting from bolt preload forces. The software then performs an intermediate calculation. The appropriate compressive forces to apply to the bolts during the second solution are calculated correcting for differences in the length of the bolt model and the loaded length of the actual bolt.
• Solving the model a second time gives the stress state resulting from bolt preload forces and, optionally, service loads.
Then a bolt preload condition can be included in a normal mode solution (SOL 103), and in an optional dynamic response calculation (SOLs 111 and 112).
Best regards,
Blas.
03-11-2015 05:13 AM - edited 03-11-2015 05:13 AM
Blas,
thanks alot for your detailed answer.
Your results seem realistic (even with the very high eigenfrequencies). What magnitude force did you use?
I defined my model in NX. But what I am wondering about is that my modes have a different shape. The amplitude of mode1 and 2 (the orthogonal bending modes) are not in the middle of the beam mesh (as could be seen in the following picture).
Is it possible to get your dat file? I want to check if I can find any differences in definition. Which Nastran Version did you use?
Best regards
Andreas
03-11-2015 11:36 AM
This looks like a bug. If you examine the displacements in the static subcase, they look correct - the preloaded element gets shorter and the other 9 elements get longer. However, if you examine the XX strains, they are constant which does not make any sense. The preloaded element should have a compressive strain equal to 9 times the tensile strain in the other elements.
If I manually apply the preload force to the bolt, the displacements and stresses are the same as the bolt preload model, but the strains now look correct. The mode shapes in the modal subcase also look correct.
Something is messed up in the strain calculation and the differential stiffness calculation.
03-12-2015 06:09 AM
Jim,
thank you for your response.
I tried to replace the bolt preload by the manual defined forces like in your description for the modal subcase and the results are the same like with the bolt preload. I get 8 modes with strange shapes.
It seems that in your input the bolt preload in the eigenvalue subcase prevent the solution to use the results from the static load case.
03-12-2015 06:45 AM
Yes, I found the same late yesterday and did not get a chance to reply. With the preload in the static subcase only these two cases yield the same results. These results do look correct. 9 of the beams have tensile stress so they are becoming stiffer and their frequency is increasing. The preloaded beam has compressive stress so it is becoming softer and the frequency decreases. Since you have an even number of beams, the stiffness is no longer symmetric with respect to the midpoint, so neither are the mode shapes.
03-12-2015 07:20 AM
That is right. Defining the preload on the mid-element of a mesh with an uneven count of elements leads to a symmetric shape of the calculated modes and the frequencies are nearly the same like using an enforced displacement. But there are still 2 modes missing. That are the modes3 and 4 from the example with enforced displacement which perform a full 2pi sinus (as you can see on the picture below).
04-29-2015 01:50 AM
Hello again,
are there any ideas why these 2 modes are missing?
It would be great if I could understand this behavior.
Best regards
Andreas
04-30-2015 08:48 AM
This looks like a bug. You model has 54 unconstrained DOF and your EIGRL is asking for the first 10 modes, but you are only getting 8.
Please report this to GTAC.