Problem in Dynamic Analysis

[Sayan]

Hiii,
Thank you,,Though I don't know much but will try to gather some knowledge about the recompilation procedure in my windows 10 operating system.I will be thankful if anyone take my problem as test case....One more thing,I have found in my problem,,when i was running my structure with less number of mesh it's not getting any segmentation fault error but because of using less elements results are not matching.....Last time you checked my meshing comparing with your own meshing..Can u please tell me how many elements you used? if it's lesser than me,as it was also giving good result in terms of eigen value I would like to try that one as I am not getting any other way to skip that stated error message..

[raback]

Hi

If you are not some sort of wizard don't bother trying to compile the Windows stuff from scratch. I hope these nightly builds will be updated automatically,

https://www.nic.funet.fi/pub/sci/physic ... n/windows/

Most serious users tend to use Elmer under Linux and hence there is little focus on Windows problems. Unfortunately. Best user experience may be obtained using a ready made virtual machine that comes with all necessary bells and whistles,

https://www.nic.funet.fi/pub/sci/physic ... lMachines/

It may run in VirtualBox under Windows. In fact that is exactly my platform when using my laptop.

-Peter

[Sayan]

Hiii,
Thanks for your references... I installed modified ElmerGUI window AMD64.exe to run the problem...Even run that program directly in the command line by typing "ElmerSolver"..but every time at different point segmentation fault error is coming and "backtrace for this error" number varies between 27 and 28 after which it stops...I am getting no clue to move forward,,,Only option left is to try VM for which I am installing .ova file...I believe the same problem will persist there also..Is that a bug or I am having a problem in my own program?

[kevinarden]

My mesh had more elements than yours. However I think you could reduce your problem size by using parabolic brick elements instead of parabolic tetrahedrons. The accuracy of the eigenvalues depends on the mesh size. i was able to converge on the eigenvalues with 261 nodes and 32 elements using this mesh.

beam.grd

(594 Bytes) Downloaded 170 times

[kevinarden]

With the above mesh and this sif it solved out to 26 seconds, no errors, and reasonable results. It took 11 minutes.

nltcase.sif

(2.28 KiB) Downloaded 217 times

26sec.png

(81.12 KiB) Not downloaded yet

[Sayan]

Hiiii,
Thank you so much...I was using your .grd file...It's really helpful..Is there any option in elmer to extract steady-state response from total transient time-step analysis?

[Sayan]

Hiii,
I am wondering whether the harmonic analysis provides "displacement amplitude" as result under forced vibration combining all modal responses or the displacement result is something else..If it is so, how many modes it considers in case the number of modes is not specified in the input file....

[raback]

Hi

Harmonic analysis gives reaction to the specific harmonic wake. Eigen analysis gives the different modes.

There is a module "FilterTimeSeries" that can make Fourier transform on-the-fly for transient systems.

-Peter

[Sayan]

Hiii,
If possible can you refer to some references on how can I implement it....Actually I am trying to get only steady-state response for my geometrically non-linear dynamic problem between 24s to 28s...Till now with the immense support from you all experts, as an amateur like me in this platform have successfully performed transient time-step analysis, which produces total displacement response with time i.e(transient+Steady State)..But for my future work, only the steady-state part is required...if you refer to the details of the procedure available for this to produce steady-state response at different time, will be too much help for me....

[raback]

Hi,

As I understand it you're looking for one cycle in transient analysis where the initial transients have died out. In Elmer the only way to reach to this currently is to solve the transient system until each period looks the same. We have no shortcut to ignore the transient system. The methods outlined in test case "TemperatureCyclic" make determination of converged state faster and easier.

I could imagine a method where you for a basis using the eigenmodes and use that as ansatz for harmonic wake. This would give a smaller system where the amplitudes for the steady-state condition could be solved. We have no such method.

-Peter