running multiple time histories

[STKO Team]

Dear user,

Here is a working version of your file that can do parametric analysis with OpenSeeMP.
It can handle different dt and num_steps even in the same X-Y pair.

Modelbi.zip

(251.91 KiB) Downloaded 528 times

Since the Monitor's name is parametrized, so that each process writes monitor files on different files, you need to first download the latest version of STKO (2.0.2).
Then, when the monitor starts, you will find a new feature, to visualize multiple plots at the same time.

Keep in mind (as I will explain in today's webinar), that a parametric analysis with OpenSeesMP should be run manually:

• Write TCL files from STKO (do not run it)
• Go into the directory where the TCL files are
• Open a windows command shell
• Type C:\OpenSees-Solvers\openseesmp.bat .\main.tcl 25 (25 because you have 50 GM, so 25 pairs)

Keep in mind that if you have less than 25 processors, your PC will slow down a lot...

Another note:
Since you found the bug in the Infill Masonry Wall Element, we just fixed it and included in the 2.0.2 few minutes ago. If you downloaded version 2.0.2 already, re-download it now. It will contain the bug fix

[Zeeshanbhat]

Dear Team,

Thank you so much.

Regards,
Zeeshan

[STKO Team]

You're welcome! Happy Modeling

[elena.miceli]

I am trying to perform a parametric analysis changing the seismic input on a RC frame, and I am trying to follow the example in "Modelbi.scd".
Everything works fine until it runs the "Analysis run (parametrized)" CustomCommand. In particular, I receive the following error "invalid command name "CustomFunctionCaller"
while executing
"CustomFunctionCaller $increment_counter $time_increment $time $num_iter $last_norm $perc $process_id $is_parallel""
It seems like it doesn't find the command "CustomFunctionCaller" probably because is part of an old version of STKO.

Can you help me please?
Thanks in advance

[STKO Team]

Try this:

Modelbi.zip

(256.43 KiB) Downloaded 550 times

The logic of the custom script is slightly different now, and should go before the real analysis step

[Gabriele.Neri]

Dear stko team,
I'm trying to run a parametric analysis according to the example in "Modelbi.scd".
I made two differente files, in order to make some attempts:

In the first file "timehistory mod" I run just one timeseries and I get in the output the number of steps equal to the seismic input. The seismic input used here it's equal to the first of the thirty seismic inputs used in the second file.

In the second file "0.1g" I run 30 timeseries simultaneously and the number of step obtained in the output is smaller than the number of the seismic inputs.

Can you help me?

Thanks in advance

[kesavapraba]

Hi,

In the second file "0.1g" I run 30 timeseries simultaneously and the number of step obtained in the output is smaller than the number of the seismic inputs.

Your exact question is not clear to me. Specifically, why you compare the number of steps in accelerogram (seismic input) with the number of accelerogram. Can you please explain your issue in a more detailed way?

[Gabriele.Neri]

[/quote]
Your exact question is not clear to me. Specifically, why you compare the number of steps in accelerogram (seismic input) with the number of accelerogram. Can you please explain your issue in a more detailed way?
[/quote]

Hi, yes of course!
In the file "timehistory mod" everything is ok.

in the file "0.1g" I made a parametric analysis, the steps are the following:

- I created TCL files with the Write Input File command
- I opened a command shell in the directory in which I generated the TCL files
_I launched OpenSees with the number of processors = number of parameters, that are 30 in this case
(C:\OpenSees-Solvers\openseesmp.bat .\main.tcl 30)

The number of steps for each seismic input is in " GMNumPoints.txt " file.
(for the first there are 2999 steps, the second 1999 steps and so on...)
In the output I should obtain a number of steps = number of steps of each seismic input, (that is 2999 steps for the first input, 1999 for the second and so on until the last seismic input), but it doesn't happen. I obtain less steps in the output than the number expected. I can't understand why.

Thanks in advance

[kesavapraba]

Hello, I understand your question clearly now. I did a parametrized analysis of your model using only six accelerograms, and it worked well. To identify which processor is handling each accelerogram, refer to the figure below. Please note that the order of information (no. of steps) displayed on the command prompt screen may be shuffled differently. Additionally, I included a monitor plot in parallel mode to validate the results - please find the attached model with the six records.

[Gabriele.Neri]

Hello, I understand your question clearly now. I did a parametrized analysis of your model using only six accelerograms, and it worked well. To identify which processor is handling each accelerogram, refer to the figure below. Please note that the order of information (no. of steps) displayed on the command prompt screen may be shuffled differently. Additionally, I included a monitor plot in parallel mode to validate the results - please find the attached model with the six records.

Hi,
the issue is with the drifts in the output. I make an example just for the first seismic input, because the issue is the same for the others.
In the file " GMNumPoints.txt " there are 2999 lines , while in the corresponding drift file in output I obtain just 2900 lines, so 99 lines are missing.
I Attached the images in order to be clearer, meanwhile thanks for the support!