Frequency response problem

[bjorn_fallqvist]

I'm running a frequency response analysis using the modal results from an earlier scenario. Everything seems to run fine, but the analysis seems to more or less stop at this line:

 **** END   TYPE-2 DATA INPUT
 **** Writing stress output file for Superview ...

It has been stuck here for more than 8 hours now. It's a big model, but not huge (maybe around 500 000 DoF) so it shouldn't take that long. I don't get an error message or anything. There are no solvers running as far as I can see in the task manager either.

If I close the analysisserver.exe application, I can access results (but the load cases are given as "no frequency avbailable", and there seems to become some file conflict because I cannot re-run the design scenario, copy the model or do anything with it without restarting the computer.

Suggestions?

Thanks

Björn

[xli]

How many frequencies you did in the modal analysis?

[bjorn_fallqvist]

Hi,

I calculated ten modes.

[xli]

Bjorn,

Half millions DOFs and 10 modes should not be too large for stress recovery processor to run. More than 18 hours on a PC with standard cconfiguration today (say ~8 GB RAM, 2.x GHz) showed there might be something else going on. Such as: are you running other big jobs while you are doing this analysis? Or do you have a 'bad' hard disk etc. Or it could be a program defect like some odd thing is recently introduced in this processor regarding to I/O performance etc. It is really interesting. In case you run this model case again, if you can post the log files, modelname.ds_data\ds#\ds.l, and odelname.ds_data\ds#\ds.log for this long job; and a snapshot of memory/processes of your machine at time it runs 18+ hours not finished, that would be a great help us of diagnosis to the problem. Thanks a lot.

-xli

[bjorn_fallqvist]

Hi,

I'm running on a reasonably high-performance computer (quad-core 2,4GHz with 24GB RAM) which is rather new, so I don't think hardware is the problem. I left the analysis running overnight without any other jobs, so they shouldn't interfere. It seems after all that we don't need to do this analysis, but if I have time I will run it again and post the log files.

Thanks

Björn