Drift recorder with specified -dt does not write sometimes

[aodonne1]

Hello,

I am performing close to 5000 dynamic analyses in a "for" loop and for each analysis the only thing that changes is the input ground motion. I have a very simple 2-D model with 2 dof from which I am outputting maximum story drifts and floor accelerations using the EnvelopeDrift and EnvelopeNode commands. I would like to output residual drifts as well, however due to Hard Drive capacity (I am running 24 batches of 5000 analyses) and post-processing constraints I would like the output files to be reasonably small (i.e. not the entire time history). From the suggestion of some previous posts on this message board I set about doing this by specifying a drift recorder with a -dt that is equal to [(the duration of the ground motion) minus (2 times the analysis dt)] or for all of you who really enjoy .tcl :

set dt_write [expr (dt_GroundMotion*numGroundMotionPoints)-2*dt_analysis]

This tactic works perfectly for the first 99 out of the 5000 analyses but starting with the 100th analysis the drift recorder with specified -dt no longer writes a file.

Here's what I've tried so far with regard to troubleshooting:
1.) Setting -dt equal to the analysis timestep such that it should be the same as the output file as if -dt had not been specified (nope!)
2.) Turning on the -time option to output the time vector as well (nothing!)
3.) Setting the output format to -xml rather than -file (nada!)
4.) I checked the location of "wipe" to see if that would be causing any issues but it is properly located (so not that either!)

I am at a loss and would appreciate any help. Here is the recorder line as it exists in my code. As I mentioned, it works perfectly the first 99 times so I don't think anything is wrong with the syntax but I could be wrong.

set DTwrite [expr ($gm_dt*$num_gm_data_pt)-(2*$gm_dt)];
recorder Drift -file [concat $name/drift_resid.out] -time -dT $DTwrite -iNode 1 2 -jNode 2 3 -dof 1 -perpDirn 2;

Also, to preemptively address any questions or concerns, for this analysis $gm_dt is equal to the analysis timestep so that's not the issue either.

Thank you,
Andrew

[fmk]

probably has nothing to do with the specific command and mor to do with some memory problem in the code, i.e. some memory is not getting cleaned up properly. you will have to email the script for me to see if i can track it down.

[aodonne1]

OK Frank, I would appreciate that very much. How would you like to go about getting me your e-mail address? I assume you probably don't want to post it to the message board, haha.

[fmk]

i removed address

[aodonne1]

Thanks Frank, you should have an e-mail in your inbox with my scripts. If you have any question please don't hesitate to ask.

Thanks,
Andrew