I am new to OpenGeoSys and I have questions about coupling mechanical and porous flow processes.
I want to know if OpenGeoSys support coupling of the PDE at the solver level. What I mean by that is that mechanical and fluid flow PDE are solved together in a single matrix rather than solved sequentially at each timestep.
To be more precise, I am interested in coupling no-linear mechanical processes such as creep/plasticity with Darcy flow in porous medium. The mechanical process lead to damage of of the rock and increase porosity/permeability.
Because, there is a very large contrast in between permeability between undamaged and damaged rock, I have lot of numerical instability problems when not coupling the equations at the solver level. Of course this issue could be solved by reducing the timestep and using a very fine mesh, but I am hoping to be smarter than relying on such brut force approach.
If I am asking this questions, it is because I have so far used software for which this is not possible (FLAC3D) and I am looking for a software solution that would best fit my problem.
I tried to look in the documentation before asking this question but it was unclear to me if this feature was supported.
by default OGS uses the monolithic coupling in the hydro-mechanical (HM) process. So generally that should be what you want. Regarding the damage related permeability: you will have to specify an appropriate material model for the mechanics and also an appropriate permability model. If you search this list for permeability, you get the currently provided permeability models in OGS: OGS: [tag] property
I know the HM process takes the strain derivative of the permeability into account for the jacobian, if the permeability model provides such a derivative which should help the convergence. I know the OrthotropicEmbeddedFracturePermeability does it, but I don’t know for the rest.
But in any way, If convergence becomes a problem NewtonDampening might help: OGS: [tag] damping
Are there limitations on the mechanic material models that can be used is hydro-mechanical (HM) process:
for example I have seem that OpenGeoSys support Lubby2 types of creep models.
Will there be problems for using such model in HM or OpenGeoSys lets you do what you want and it is up to the user to configure the solver properly to get convergence.
I am asking because this type of mechanical model would put me in the case of large deformations.
You’re welcome, in principle you are free to combine what features you want, but not every possible combination has been tested. So there may be some incompatible feature combinations. Here you can see if there are existing benchmarks for any combination of features. just check the box for HYDRO_MECHANICS under process_type and search for permeability under property. Clicking on of the permeability models will show you some existing (and thus tested to be working) examples.
Figuring out good solver settings is up to the user though and probably highly dependent on process and model specifics.
We also have a LargeDeformation process. But not yet a HydroLargeDeformation. Although it might get implemented sometime in the future.