unloading/reloading issue with Hysteretic material

[giovannimilan]

Hi all,

I am running Time-Histories on SDoF model with 1 transnational spring with the Hysteretic material (see source code below by signature).

I've realized something strange in the output hysteretic curves:

-in the non-linear range, after the unloading, the reloading shows "jump" of forces (almost vertical lines as infinite stiffness) - it is shame i cant upload figures here.

I havent found this info explicitly in the documentation, but I assume that the model is a modified takeda and hence in the unloading/reloading if performs the following check on the stiffnesses:

S = max(S_sec, So (dy/dm)^beta) from (https://dianafea.com/manuals/d95/MatLib/node402.html).

I've done some hand calcs on the ouput hysteresis, and indeed it looks like that the unloading follows the "So (dy/dm)^beta" stiffness even if it is lower than the "S_sec". and consequently, the material suddenly realizes that the stiffness is too small, and so it jumps back abruptly (to the force that it would have predicted if it were to follow S_sec).

I've used this model extensively before with different backbones and was not causing issues. Indeed I've tryied to use a different backbone (but same pinching, damage, and beta), and does not cause the same odd behavior. I've tried with several different EQ recs, and with static displacement control.

Is it really an unlucky combination of the input param's that triggers the issue?

Any idea of what could have gone wrong?

Thanks a lot,
Gio

PS: I am using V 3.1.0 64-bit.

#source code

# build the backbone

set s1p 250.
set e1p 0.002
set s2p 500.
set e2p 0.01
set s3p 500.
set e3p 0.036

set s1n [expr -$s1p]

set e1n [expr -$e1p]

set s2n [expr -$s2p]

set e2n [expr -$e2p]

set s3n [expr -$s3p]

set e3n [expr -$e3p]

#uniaxialMaterial Hysteretic 2 $s1p $e1p $s2p $e2p $s3p $e3p $s1n $e1n $s2n $e2n $s3n $e3n 1 1 0.0 0.0 0.75