Initial stress material

[armmlk]

Hi,

I recently came across SuanPan and it seems like a great platform. I've been experimenting with modeling a reinforced concrete cross-section, following the example provided here:
https://suanpan-manual.readthedocs.io/en/3.7/Example/Structural/Statics/rc-section-analysis/

I'm trying to extend that example to include prestressing steel. In OpenSees, there's a material model called InitStressMaterial (https://openseespydoc.readthedocs.io/en/latest/src/InitStressMaterial.html) that wraps a uniaxial material with an initial stress state which can be useful for this kind of problems.

Is there anything similar available in SuanPan, or perhaps another material model or workaround for modeling prestress?

Thanks in advance

[TLCFEM]

Currently there is no similar thing to achieve this.
But I think this is a good thing to add.

InitStressMaterial performs a local iteration to convert initial stress to initial strain and adds it to the underlying material.
InitStrainMaterial directly applies the initial strain to the underlying material.
The main problem of this approach is that, no matter how large the initial strain is, it is applied in one step, it could be excessively large such that the underlying material cannot converge, or the error is large.
And the displacement is not synced with the material at the beginning of this 'initial' state.
Furthermore, do you mind sharing your experience with InitStressMaterial?
I presume when you check the results, for example, displacement, the first point (before applying initial stress) will be zero, the second point (after applying) will abruptly jump to the prestressed state, no matter how small the step size is.

I plan to add a Prestrain wrapper in the coming days.
The interface will look like this.

material Prestrain <material_tag> <underlying_material_tag> <amplitude_tag> <maginitude>

Just like conventional loads, the initial strain will be applied gradually following the curve defined by the attached amplitude.
By such, the numerical performance would be better (less convergence issue and controllable even if the prestrain is large).
The corresponding results will vary smoothly instead of abruptly.

What do you think about this?

[TLCFEM]

I've created an issue #265

[armmlk]

Nice!
You are right about the sudden application of the strain in OpenSees, and one has to create a script to update the initial stress parameter in order to smoothly apply it. With your proposed way, one does not need this workaround.

[TLCFEM]

#266 looks fine to me and appears to be working.
The initial strain to be applied can be precomputed for whatever uniaxial material models.
I'll merge it and you are very welcome to test it.

You can compile the application locally, or I can send you a copy if you leave your email address and the preferred flavor here.