“Pseudo-dynamic Analyses”experiments in OpenSees ?

[burnningcat]

Dear Sir,

I need to use the OpenSees to simulate the “Pseudo-dynamic Analyses”experiments, and I find that it is quite difficult, so I kindly ask if somebody have successfully doing this before and how?

In the “Pseudo-dynamic Analyses”experiments, instead of using the shake table, the displacements are applied in each floor by the jacks, as describe in the following paper:

“Computational aspects of a seismic performance test method using on-line computer control”, Pui-Shum B. Shing &Stephen A. Mahin
in the website of
http://onlinelibrary.wiley.com/doi/10.1 ... 6/abstract
It seems that to model this “Pseudo-dynamic Analyses”, in any time-step, a specified displacement vector in each floor level should be applied. This is very difficult to model in OpenSees. Because OpenSees use a fix Load Vector or a fix Displacement Vector in the analysis. So I kindly ask if somebody have successfully doing this before and how? Thank you very much!

[hsafti]

I was not able to open the paper but I guess your model can be implemented in OpenSees. You need to define an SP constraint to the node attached to the jack. The SP constrains should be introduced via a pattern command and a time series command (look up manual)

Also, I guess this is a .exe user question!

[burnningcat]

Thank you very much for your kind reply. And would you please give a short example about how to define this pattern for the SP constrain. For example, if we need to apply the displacement-history in a three storey building. How can we define the load pattern for the floor displacement for the three floors? Many thanks to you!

[hsafti]

First you'll need to define a time series (pdf manual page 304):

Series -time {list_of_times} -values {list_of_values}

or via files:

Series -fileTime $fileName1 -filePath $fileName2

The sp command (page 307)
sp $nodeTag $DOFtag $DOFvalue


The pattern command (page 305)
pattern Plain $patternTag (TimeSeriesType arguments) {
load (load-command arguments)
sp (sp-command arguments)
eleLoad (eleLoad-command arguments)
}


if you want to apply only horizontal disp. to node 58 (for example):

set MySeries "Series -time {0 1 2 3 4 5 6} -values {1 -1 1 -1 1 -1}"
pattern Plain 1 $MySeries {
sp 58 1 0.1
}

[burnningcat]

Dear hsafti, really thank you for your kind reply and your detailed help!! But I still have a problem. Because I need to appy the displacement in different floor level. For example, if we need to apply SP to two nodes, and each node has a different time series. So hwo to modified your above example? Many thanks

in your example:

if you want to apply only horizontal disp. to node 58 (for example):

set MySeries "Series -time {0 1 2 3 4 5 6} -values {1 -1 1 -1 1 -1}"
pattern Plain 1 $MySeries {
sp 58 1 0.1
}


why in this line sp 58 1 0.1

is using 0.1 but not 1?

[hsafti]

The "0.1"is a factor to be multiplied by the values vector ... so it is arbitrary.

For more than one node (using the same MySeries) you just add another sp command to the pattern.

pattern Plain 1 $MySeries {
sp 58 1 0.1
sp 102 1 0.1
sp 107 2 0.2
sp ..
.
}

If you need different time series for each node, you have to make different patterns for each node.

pattern Plain 1 $MyFirstSeries {
sp 58 1 0.1
}

pattern Plain 2 $MySecondSeries {
sp 102 1 0.2
}

[smith88]

burnningcat wrote:
> Dear Sir,
>
> I need to use the OpenSees to simulate the “Pseudo-dynamic
> Analyses”experiments, and I find that it is quite difficult, so I kindly
> ask if somebody have successfully doing this before and how?
>
> In the “Pseudo-dynamic Analyses”experiments, instead of using the shake
> table, the displacements are applied in each floor by the jacks, as
> describe in the following paper:
>
> “Computational aspects of a seismic performance test method using on-line
> computer control”, Pui-Shum B. Shing &Stephen A. Mahin
> in the website of
> http://onlinelibrary.wiley.com/doi/10.1 ... 6/abstract
> It seems that to model this “Pseudo-dynamic Analyses”, in any time-step, a
> specified displacement vector in each floor level should be applied. This
> is very difficult to model in OpenSees. Because OpenSees use a fix Load
> Vector or a fix Displacement Vector in the analysis. So I kindly ask if
> somebody have successfully doing this before and how? Thank you very much!