Bug Spotted in Bilin material

[thomas.ulrich]

Hello,

When I do a restore in a model with Bilin uniaxial material, Opensees raises an error (saying that UniaxialMaterial has no zero class or something like that).I think that is a small bug that can be corrected by modifying the sources like that:

SRC/material/uniaxial/snap/Bilin.cpp
replace UniaxialMaterial(XXX, 0) by UniaxialMaterial(XXX, MAT_TAG_Bilin).

cheers,

Thomas.

[thomas.ulrich]

Hi,

It seems that it has been corrected first, but then that the revision 5253 has put back that bug.
Is it a mistake? Then could you correct it please?

In the same revision, the following condition has been modified:
if((flagstopdeg==0)&&(flagdeg==1)) { (line 850)

Could you explain what was wrong before? Which degradation mechanism will be affected?
(because it seems that the degradation mechanisms were well functionning???).


Thomas.

[thomas.ulrich]

Hi,

We have been using the Bilin uniaxiamaterial in our research, and it presented a correct behaviour, but since revision 5341, there is a bug:
when we impose a small accelerogram (one of the first stage of our IDA) on our model, the Inter-story drift "explodes" to very high values!
Since I commited the lattest revision with the previous version of Bilin, the bug is gone.
So there must be some bug introduced in the lattest version of Billin!

Thomas.

[fmk]

do you want me to make a copy of the 5274 version of the code and call it Bilin1 .. that way when they fix the problems they won't mess you (and others) up?

[thomas.ulrich]

Since with svn, we can access the previous version of the file, I could work it out myself.
For example, I'm now using that code to have a reliable version:

svn co svn://opensees.berkeley.edu/usr/local/svn/OpenSees/trunk OpenSeesIntel/
svn export -r 5340 svn://opensees.berkeley.edu/usr/local/svn/OpenSees/trunk/SRC/material/uniaxial/snap/Bilin.cpp OpenSeesIntel/SRC/material/uniaxial/snap/Bilin.cpp
svn export -r 5340 svn://opensees.berkeley.edu/usr/local/svn/OpenSees/trunk/SRC/material/uniaxial/snap/Bilin.h OpenSeesIntel/SRC/material/uniaxial/snap/Bilin.h

I just wanted to point out the bug. Maybe you should see that with the people who made the latest corrections.

Thomas.

[f.ribeiro]

The changes made in the last commit were mainly related to the update of model parameters due to cyclic deterioration. Additionally, other minor bugs, namely in the post-capping stiffness, were corrected. However, from the validation examples we used, not large differences are expected. Can you post the example in which you are having problems for me to test it?

[f.ribeiro]

There was a minor bug in the update of the unloading stiffness due to deterioration. The code with revision 5531 contains the latest bug fix so that stiffness deterioration also works for the case when the Bilin material is used in zero-length springs with the optional n-factor. Please use this latest version.

F. Ribeiro

[thomas.ulrich]

Hi,
I tried revision 5533 of the code, and I'm now getting an error when I'm executing a save statement.
I precise that the problem doesn't occur with the release 5340.
If you want, I can send you my input file.

Thomas.

[fmk]

thanks .. please send it to me
fmckenna AT berkeley DOT edu

[fmk]

can you confirm whether the material works from your point of view in terms of stresses, strains and tangents in a script that does not use the save.

[thomas.ulrich]

I just run a dynamic simulation with the new model, and I'm getting exactly the same displacement at the one of teh top nodes of the structure as what we obtained in the previous releases (and these previous results seemed good). Once save and restore functions will be restored, I will be rapidely able to carry more tests.

[fmk]

i have updated the code .. version 5537 .. to fix send/recvSelf i also had to modify the setTrialStrain function to deal with the case where
the the difference between the new trial strain and previously committed one was < DBL_EPSILON .. hope this has not effect the code ..
it shouldn't but you never know.

[thomas.ulrich]

Dear fmk,

thanks for the update. There still a bug thought. I sent you a script that triggers the new bug.

Basically, what I'm doing is:
Building the model
Static analysis
Save 0
Dynamic analysis
Restore 0
Crash

Thomas.

[fmk]

Thomas,

i cannot repeat the bug on any of the machines that have the latest code. Can you update your source code again, clean it out and the build again.

if that fails, can you tell me which operating system you are using and which compiler to see if i can get access to something similar.
thanks
frank

[thomas.ulrich]

You were right, something went wrong during my compilation of the new rev.
A "make all" was not enough, I had to do a "make clean" to make it work properly.

Thank you.