recorders not generated [OpenSees] - azing, March 30, 2021

[JustinBonus]

Hi,

I am using OpenSees model of the 20-story SAC building in 2D as an input for WE-UQ. The model has a lot of elements and nodes, so I used the "user defined" option in the EDP section. If I use the "standard wind" option, Dakota will give me an error saying that "expected function values are not equal to the values found".

When using the user defined option, I modify the "recorder" and the "postprocess" files found on the website for the 9-story SAC building to just give me the results for the displacement at some node at roof level of the building. But no matter how I change the recorder and postprocess files, when I run the program, no disp.out files are generated for me in the work directories. Accordingly, I get zero and NAN values for the forward propagation analysis results. Could there be another reason for the zero and NAN results? I am using a stochastic wind event.

This is what I have in the recorder file:
recorder EnvelopeNode -file disp.out -node 1123 -dof 1 disp

This is what I have in the postprocess file:
set dispIn [open disp.out r]
while { [gets $dispIn data] >= 0 } {
set maxDisp $data
}

create file handler to write results to output & list into which we will put results

set resultFile [open results.out w]
set results []

foreach edp $listQoI {
set splitEDP [split $edp "_"]
set response [lindex $splitEDP 0]
set tag [lindex $splitEDP 1]
set dof [lindex $splitEDP 2]

if {$response == "Disp"} {

set nodeDisp [lindex $maxDisp 0]
lappend results $nodeDisp
} else {
lappend results 0.0
}
}

send results to output file & close the file

puts $resultFile $results
close $resultFile

• azing, March 30, 2021

[JustinBonus]

Azing,

Would you be so kind to share/attach your OpenSees input file so we can run it ourselves for debugging?

Thanks,
Peter

• pmackenz, March 30, 2021

[JustinBonus]

That would be great. I appreciate it. I have attached the code to the message.

Thanks,
Azin

• azing, March 31, 2021

[JustinBonus]

For this model, I use the nodes located at the bottom of the panel zones, throughout the left column line as centroid nodes as well as response nodes. Therefore, not all the nodes on the left column line are included in loading and response monitoring (e.g. nodes on the panel zones are not included). Could this the problem?

I used the WE-UQ three-story example, and eliminated one of the nodes to see what happens. i.e, instead of 1 2 3 4 as centroid nodes, I wrote down, 1 2 4, and modified the recorder and postprocess files accordingly. So with this conditions, no recorders were generated for me. That is why I assumed that either all the nodes on the column line should be included in the centroid and response sections, or I am modifying the recorder and postprocess files incorrectly.

• azing, April 01, 2021

[JustinBonus]

When I use the nodes on the leaning column as the "centroid" and "response" nodes, the problem is resolved since I don't have panel zones on the leaning column, and I don't need to skip any nodes in between, which was the case when I was using the left column line as the loading and response line.

• azing, April 06, 2021

[JustinBonus]

Hi,

is this resolved or do you need more help?

frank

• fmk, April 06, 2021

[JustinBonus]

Thank you, It is resolved.

Azin

• azing, April 06, 2021