Warning ArpackSOE

[jackjo]

Hi everyone
I am performing an IDA analysis (nonlinear dynamic analyses with increments 0.1g) on a bridge model. This model is analyzed under mainshock-aftershock sequence ground motion. Used ground motion duration is 178.5 seconds. In order to run the IDA analysis for all of scaled ground motions (0.1g, 0.2g, 0.3g ,...) automatically, I wrote a loop command at first of tcl file before wipe. When I run this analysis, analysis ends for the first scaled ground motion (0.1g) successfully, but analysis for the second scaled ground motion (0.2g) goes on until 131.5 seconds, then stops, and displays following warnings. Meanwhile, when I performed nonlinear analysis just for 0.2g scaled ground motion, not for 0.1g, 0.2g, 0.3g, ... scaled ground motions using loop command, I did not see any warning. By the way, when this model is analyzed under mainshock ground motion, by using loop command, analysis for all of scaled ground motions ends successfully without any warning.

WARNING ArpackSOE::ArpackSOE : - LinearSOE - failed in setSize DirectIntegrationAnalysis::handle<> - EigenSOE::setSize<> failedDirectIntegrationAnalysis::eigen<> - domainChanged failedWARNING BandGenLinSOE::BandGenLinSOE :
ran out of memory for A <size,super,sub> <520, 62, 62>
WARNING ArpackSOE::ArpackSOE : - LinearSOE - failed in setSize DirectIntegrationAnalysis::handle<> - EigenSOE::setSize<> failedDirectIntegrationAnalysis::analyze<> - domainChanged <> failed OpenSees > analyze failed, returned: -1 error flag
SUPERLU-MALLOC fails for buf in intCalloc<> at line 149 in file ..\..\..\OTHER\SuperLU_4.1\SRC\memory.c


Please guide me to solve above warnings.
Thanks so much.

[fmk]

the malloc fail could be an indication that there is a memory leak and each time you run a new analysis more memory is being eaten up. you can check this by monitoring your system memory as the program runs.

[jackjo]

fmk wrote:
> the malloc fail could be an indication that there is a memory leak and each
> time you run a new analysis more memory is being eaten up. you can check
> this by monitoring your system memory as the program runs.



Thanks so much dear Dr. fmk. Yes, you're right. I checked it and saw that this warning is because of memory leak. Is there a way to solve it, other than changing system. Because of this warning does not display if the model is analyzed under each scaled ground motion separately (without using loop), is it possible to close tcl file after running the model under 0.1g scaled ground motion automatically and then opening it automatically to rerun the model under 0.2g scaled ground motion, and repeating this process by using commands for solving this warning? In the other hand, is there a remedy to analyze the model under 0.1g, 0.2g, 0.3g, ... scaled ground motions automatically, other than changing system?

[fmk]

it is not necessarily the solver .. if you can send a small script that has a similar model using same element and material types and same analysis options i can run a memcheck on it.

[jackjo]

fmk wrote:
> it is not necessarily the solver .. if you can send a small script that has
> a similar model using same element and material types and same analysis
> options i can run a memcheck on it.

Thanks so much Dr. McKenna, I sent script to your mail.

[fmk]

i have the scripts