I am trying to derivate the test Drainage_LiquidFlow into a 3D model with an injection well in order to see the pressure evolution.
I have initialized the model pressure with a hydrostatic pressure from the top. I defined a polyline to make an injection.
When i run the simulation i have an error “Failed during Eigen linear solver initialization”.
Can i have some clue to understand and sove this error ?
Thanks in advance
Grid.zip (2.5 MB)
I just had a brief look. It seems that SparseLU is not able to solve the system. The root causes could be manifold. Therefore, just some tips:
- For more complex 3d problems, I suggest using VTK submeshes instead of gml files for boundaries and source terms. They are easier to handle and check than the gml files.
- If something does not work, I recommend increasing complexity stepwise. E.g., deactivate source terms; keep a lot of things fixed in the beginning.
- If you are in SI system, a time steps from 1s to 9s seem very small to me. Small time steps are not always better than bigger steps → this depends on the grid…
- The bigger the system, the more I would recommend using a different linear solver, like PardisoLU. It is also more tolerant and can solve systems that are not solvable by SparseLU anymore. However, it requires OGS compiled using MKL.