Reg. the sign convention of 'InitialEffectiveStressField'

Dear Prof. @Thomas_Nagel,

I was solving the HM coupled problem of a footing on the homogeneous soil media using the MCC model. I have used the InitialEffectiveStressField option in .prj file to simulate the initial stress state.

  <parameter>
      <name>InitialEffectiveStressField</name> <!--effective stress!-->
      <type>Function</type>
      <expression> sxx </expression><!--xx-->
      <expression> syy </expression><!--yy-->
      <expression> szz </expression><!--zz-->
      <expression>0.0</expression><!--xy-->
  </parameter> 

However, I encountered an issue. When I applied the standard negative sign convention for compression in the initial effective stress components, the pressure bulb (effective plastic strain contours) beneath the footing appeared incorrectly, as shown below:

On the other hand, using a positive sign convention resulted in a more accurate Prandtl-type pressure bulb:

I’m unsure why the negative sign convention—typically used in OGS—leads to this discrepancy. I would greatly appreciate your insight into what might be causing this issue.

Thank you for your time.

Thanks
Pavan

Hi Pavan,
this should all work with the negative sign convention … the way to check if everything is done correctly is to leave out the top load first and see if everything is in equilibrium with the body forces (gravity) … so no additional displacements, and proper neutral and effective stress gradients … note that by now you can also enter a initial total stress field, so that the effective stres is automatically calculated from that and initial pore pressure … simplifies things a bit …
Th
PS: Which of the MCC models to you use in OGS?

Thanks, Prof. @Thomas_Nagel, for the prompt response. I’ll try the solution you have suggested and get back to you. By the way, I use the ModCamClay_semiExplParaInit code, which uses linear elasticity. Because the non-linear elastic MCC code ModCamClay_semiExpl_absP still has some convergence issues and does not work for all types of problems.

And the ModCamClay_semiExpl version? Should be classical incremental formulation …

Another note: check whether your initial stress takes you outside of the curve specified by the initial preconsolidation pressure. In that case, some more thoughts might be required for initialization.

You are correct, Prof. @Thomas_Nagel. I use the ModCamClay_semiExpl code with the classical incremental formulation, and it seems to work for all kinds of plasticity problems I deal with. However, the ModCamClay_semiExpl_absP doesn’t seem to work at all other than the triaxial test example in the benchmark section. Therefore, whatever modifications I perform to extend the MCC framework are on the ModCamClay_semiExpl code.

Prof. @Thomas_Nagel, I have run the bearing capacity problem indicated above with and without initial effective stress using the InitialEffectiveStressField code line first. It seems that the difference in failure envelopes obtained is different in both cases because of the activation of initial effective stresses. The case with activated initial effective stress (i.e., compression) could not form a failure envelope as per classical Prandtl theory. However, with zero or positive initial stresses (i.e., tension), the Prandtl failure envelope is formed similar to earlier. I am not sure why the Prandtl failure surface is difficult to form under geostatic stresses.

However, as suggested by you, instead of using the InitialEffectiveStressField code line, if I use gravity initialization before load application (e.g., t=0 to t=1s), I observed the accumulation of plastic strains and could not finish the gravity initialization step for a solid density of 2579 kg/m^3.

As I would like to extend the process of placing such footing loads on the slopes, I thought of verifying it on flat ground first. It would be really helpful if you could suggest something to resolve this issue.

Can you post a result where you:

  • specify hydrostatic pore pressure
  • specify initial effective or total stress increase with depth
  • apply no external load
    Results I’d like to see are the vertical stress, pore pressure, plastic strain and displacement.

Thank you very much for your support and help, Prof. @Thomas_Nagel. I tried to simulate the above-mentioned procedure step by step. Firstly, I have activated gravity to generate total stresses (t=0 to t=1s). Secondly, I have activated the Biot’s coefficient to generate effective stresses (t=1s to t=2s).

However, the gravity initialization itself failed around t= 0.47s due to the generation of plastic strains. I am attaching the files (.prj, .pvd) for your kind reference.

https://drive.google.com/drive/folders/1RYmCRTVxz9YdbJ5s4IBF1tyCOJgsoZ3J?usp=drive_link

I’ve made some small changes and it seems to run. The screenshot shows the initial state (looks good; stresses on right axis, pressure and pre-consolidation pressure on left).

You can see how the top has an OCR > 1.

Here are the changes (you can ignore time stepping changes):

Essentially I also ramp up the liquid density, leave Biot 1 all the time and added all possible output.

Thank you very much, Prof. @Thomas_Nagel. It worked. However, I have one question. Is there any way to reset the displacements to zero after geostatic stress initialization?

Sure. Do a restart with zero initial displacements. Otherwise, you can do this via a filter in paraview just as part of the post processing (temporal array something or other).