Analysis Stops Without any Warning, Re-Analyze is OK!

[Masoudmehr]

Hi,

I am doing parametric analysis of bridge spine models with a large script (over 10000 lines). The parametric study is quite massive and takes several days to analyze. The problem is that analysis stops occasionally without any warning (I don't have convergence problem). However re-analyzing the same model which had stopped, will be OK and analysis continue with no problem. This wastes me much time and loosing over night analysis.
My first guess was lack of memory. But at the time of stop often less than %50 of memory is used. I am using OpenSees-MP-2.4.3, 64 bit and the CPU is 3.5 GHz 8 cores, 16 GB RAM. Also I am using multi-processor with following commands:
set pid [getPID]
set np [getNP]

Any advice is appreciated.

Regards,
Masoud

[fmk]

you have a memory leak .. i suggest you download 2.4.4 and try again (it fixed one memory leak associated with the eleResponse command) .. if that fails i will need to get the script.

[Masoudmehr]

I tried 2.4.4. Problem still exists. I have emailed the script.
Thank you in advance
Masoud

[xantis85]

Hi,
I've the same problem when I'm trying to use Opensees parallel to perform Incremental Dynamic Analyses for several earthquakes. Every time that a single Nonlinear Time Hystory is complete, the RAM memory required however increases unless i close the program. I also tried to erase model and outputs after each NTH from a main script, but in vain.
Did you find any solution?

Thanks,
Marco.

[Masoudmehr]

I haven't received feedback so far. But I tried to clean up model by following changes and it did help considerably:

1.Optimizing fiber sections. Center of RC sections can have coarse fibers.
2.Using steel02 instead of Reinforcing steel material. RS is much demanding due to hysteresis definition. I checked results with an experiment data and adjusted parameters.
3.Avoiding extremely large stiffness for rigid elements and gap/impact elements.
4.OpenSees2.44 works better than its MP version in this matter.

I hope it helps.

[zourguinadjib]

Thank you Mr.Masoudmehr for the orientations.
I'm trying to model a RC bridge under a seismic loading can you send me your email for more orientations
zourgui2004@hotmail.com