PIMY Backbone and CSS test simulation

[aziz_barzuk]

Dear All,
1) I need to simulate a CSS test with an initial isotropic stress state of p' magnitude.
I have to use a quadUP or brickUP element to controll the pore pressure.
I reached to perform only an initial isotropic stress by imposing in the stage 0 nodal forces. So in the end of the stage 0 I obtained a isotropic stress state.
After the stage 0 I need to switch to a displacement controlled analysis because I have to perform a simple shear deformation with an imposed shear deformation gamma in a cyclic way.
I tried some solution but I miss the initial stress state, considering that I need to change the fixes configuration and seems that the load is applied form an unload state on every stage from the stage0.

I saw also that I need the InitialStateAnalysisWrapper but I don't get how to use it well.
The question is also how to obtain a stage 1 stress state that starts exactly from the one of the previous stage (stage 0).
Because seems that it takes a wile in every stage to reach the previous stress state.
It is quite hard to perform everything in a only one stage by imposing a time line variable load because the contraints change from a stage to another.

2)Using the examples available I try to obtain the backbone curve (recorder Element –ele $eleNum -file $fName -dT $deltaT material $GaussNum backbone $p1 <$p2 …>) and it changes not only imposing the $p1 in the recorder command but also it depends by the stress state impose in the model. So it is not so clear to me what is the p' that is considered. Another information is that if I try to obtain the curve using some cyclic simulation (with no initial p' and no weight) I have a stiffer curve.
Then I can't reach the same result by using the hyperbolic curve equation.
So I wanted to know better how this curve is implemented and how to use it.

Thanks in advance

[aziz_barzuk]

################ UPDATE ######################################
I'm doing a model with stko application and I reach this problem.
In particular I obtain a shift of the sigma x, and sigma y in the passage from the initial stage (initilized with the initialState wrapper) and the application of the shear (in terms of displacement with a trigonometric sp function).
The thing don't change in comparison with the static or transient solver.

In the image below (Displacements Ux - time at the top node of the squared element) it can be seen that the shift is at the time 0.1 with a strange jump. (please don't consider the image title)

https://drive.google.com/file/d/1APGS0O ... p=sharing

https://drive.google.com/file/d/1cixquz ... sp=sharing

How can I avoid that jump ?
How can I maintain a constant isotropic pressure considering that with very small shear deformations the plasticity doesn't occur ?

The file developed with stko is annexed to this thread.

[STKO Team]

Something like this:

PIMY.png (85.33 KiB) Viewed 2416 times

Check this model (

iso_stress_dir_sh.zip

(129.52 KiB) Downloaded 124 times

), look at the changes, and understand why they are needed to obtain this state of stress.

• Initial state analysis is not needed... that's for construction stages
• If you plan to do a shear along the X direction you cannot fix X on more than one point. Furthermore, to fix RBM in a 2D problem you need to fix 3 DOFs, 1 X and 2 Y at a certain distance. Removing the constraint and using force from reaction is not the solution.
• I don't know how you want to constrain the pressure in this mode, so I just fixed it at the bottom. Adjust it as needed
• When you apply shear on top, you cannot use SP constraint, otherwise, you are avoiding any relative disp between these 2 nodes in the X direction, which is not fine with nonlinear material that may dilate under shear
• Still, on top, to avoid rotation (i.e. no rocking, just plain shear) use an EqualDOF in the Y direction -> fixes the rotation about the Z axis
• Finally, because you cannot use a triangle, you need a way to tell the bottom-right node that it has to displace in X (which in this model represents Eps11) of an amount equal to the displacement difference in the 2 top nodes (still Eps11 due to dilatancy). There is no built-in constraint in OpenSees to do that, so we developed a new element (see the ConstraintEquation analysis step).
  It's not yet in the official repo, because we still need to do the documentation. It will be available soon. In the meantime, you can use our OpenSees repo and compile it, or use the OpenSees provided by Prof. Tarque