Is consecutive multiple analysis possible in OpenSees?

[TWI]

First of all, I am doing the transient analysis, with the constraints handled by 'Transformation', and solved by Newmark method. My goal is to impose a displacement input to one set of DOFs (named as Set S1) first, using 'MultipleSupport + groundMotion (-disp) + imposedMotion'. Analyze it and perform another analysis with differenent choices as follows:
1. Impose load input to another set of DOFs (named as Set S2), using 'Plain + Load'.
2. Impose acceleration to S2, using 'MultipleSupport + groundMotion (-accel) + imposedMotion'.
3. Impose 'UniformExcitation' to the structure.
4. Impose displacement to S2, using 'MultipleSupport + groundMotion (-disp) + imposedMotion'.

All the above inputs are streamed as time series object from a data file.

It turns out only case 4 works. And I certainly hope the rest three cases would also work, so here are my questions:
One, all what I need is to 'restart' the second analysis right after the previous loading path is completed. And I can imagine that, as the displacement is imposed, S1 DOFs are condensed out and with the assignd non-homogeneous displacement moved to the RHS of the equation. So that may be the reason that in case 1-3, the second analysis shows that S1 DOFs are still keeping imposing the displacement specified from previous analysis step. So can OpenSees actually release the condensed S1 DOFs before restarting a new analysis?

Two, in case 4, it seems if I let S2=S1, the second analysis can be executed, but the starting pseudo-time of second analysis is not zero, which means if the first analysis executed 1000 incrments, then the second analysis would read the 1001th entry in the data file to be the starting input. Can I use LoadConst in this case, if I actually do NOT want to hold the previous load condition, but reset the pseudo time clock to be zero to read the data file?

Three, in case 1-3, it seems that the S1 DOFs will continue have the displacement imposed in the first analysis. Would 'wipeAnalysis' actually clear the 'pattern' object in the previous analysis? Looks like NO.

Those are just some preliminary observations. If there is a better way to do what I want to accomplish here, please let me know, thank you.

[zvidrih]

Are you reseting pseudo time after each analysis?

Code: Select all

loadConst <-time $pseudoTime> 

[TWI]

Are you reseting pseudo time after each analysis?

Code: Select all

loadConst <-time $pseudoTime> 

Thanks for the attention.
I understand that I can reset the pseudo time clock by using loadConst, but that would maintain or preserve all the previous 'pattern' objects, which is not what I wanted. I want to execute a brand-new analysis right after a previous analysis, and of course, by brand-new, I mean all the previous defined pattern objects should be wiped out, and only the element and material objects are preserved.
Also, my guess is the DOFs which have been imposed displacement would not be 'recovered' after the analysis, and this could create extra constraint in the following analysis.

[silvia]

you cannot impose accelerations using the multiple=support excitation command.
you can only impose displacements.

[zvidrih]

Now, I am not sure, that I understand what you're trying to do here. Do you want to keep the results (deformed shape) from the first analysis or just run the second analysis like it is the first one?
If what you're trying to do is the second option, then what i usually do is sth like this:

Code: Select all

source geometry.tcl
source analysis1.tcl

wipe

source geometry.tcl
source analysis2.tcl

...

btw, you can use loops
...

wipe analysis command only wipes the analysis parameters, not the results.

I guess you could also use the "reset" or "save-restore" options, but i haven't tried them out yet.

[TWI]

you cannot impose accelerations using the multiple=support excitation command.
you can only impose displacements.

OK. Then the manual should be corrected on this. But still, it seems the DOFs which have been imposed displacement upon during the first analysis did not get released after the analysis is done, and it continues to have the same displacement acting on during the following analysis.

[TWI]

Now, I am not sure, that I understand what you're trying to do here. Do you want to keep the results (deformed shape) from the first analysis or just run the second analysis like it is the first one?
If what you're trying to do is the second option, then what i usually do is sth like this:

Code: Select all

source geometry.tcl
source analysis1.tcl

wipe

source geometry.tcl
source analysis2.tcl

...

btw, you can use loops
...

wipe analysis command only wipes the analysis parameters, not the results.

I guess you could also use the "reset" or "save-restore" options, but i haven't tried them out yet.

What I want to do is to remove the load pattern after the analysis is done, keeping all the material properties as it is and start a new analysis right away with a new load pattern. So the question is wipeAnalysis seems to wipe out the analysis parameters but not the load patterns.

[TWI]

OK. Seems the command 'remove' can do the trick to remove the effect previous load pattern and reset the clock too. Hope the description of this command could be added into the manual.
Thanks everyone paid attention to this post.

[TWI]

OK. Seems the command 'remove' can do the trick to remove the effect previous load pattern and reset the clock too. Hope the description of this command could be added into the manual.
Thanks everyone paid attention to this post.

My mistake. Seems the clock still continues from the previous analysis. I applied 'loadConst' after I removed the pattern object, and now it looks fine.