0

Convergence becomes difficult after CO2 breakthrough (ECO2N)

Hi, I want to know why the convergence becomes difficult (so the time step becomes very small, or simulation terminates) near or after phase breakthrough. I use ECO2N module for the sub-core scale problem in which I not only try to see the flow behavior before breakthrough and after breakthrough.  I have done many tests that show the problem. 

Could anyone tell me the reason in principle and how to avoid the problem or improve simulation? 

I use a single material, inactive cell at outlet face and already small enough time steps. I use Tough2-ECO2N and Tough2MP-ECO2N for the simulations. 

In reality, the flow should become easier soon after Breakthrough. Should it be caused by the relaxation of capillary pressure--a dramatic change in pressure within a time step?

Thank you!

4replies Oldest first
  • Oldest first
  • Newest first
  • Active threads
  • Popular
    • YZ
    • Eomi
    • 4 yrs ago
    • Reported - view

    The problem appears when the grid size is small. If I use large size grid, no this problem. But I still don't know how to avoid the problem for find gird.

    Like
  • I don't think a general answer can be given without digging into the details. Can you check whether the simulation tries to cross the saturation line (which is not possible in ECO2N)? Look at the elements/connection that cause the convergence issue and time-step reduction, than check its thermodynamic state.

    Like
    • YZ
    • Eomi
    • 4 yrs ago
    • Reported - view

    Stefan, thank you for your reply. As seen from the following printout near the breakthrough time, the changes in pressure seem suddenly turn to large. I guess this may be relevant to the avalanche of capillary pressure (no kinetic parameters controls the relaxation) when the CO2 pathways became suddenly fully connected through the modeling sample.  

      TOTAL TIME     KCYC   ITER  ITERC   KON        DX1M         DX2M         DX3M          MAX. RES.      NER    KER        DELTEX
     0.115117E+05 129987      91055868     2       0.63165E+01  0.96148E-07  0.14692E-03    0.52799E-09     136     3       0.10000E+00

    changed to 
     0.119846E+05 139986      51150662     2       0.25092E+02  0.15628E-09  0.21620E-06    0.89070E-05     157     3       0.16206E-03

    I guess if I use a weak capillary pressure function or zero capillary pressure. This problem (phenomena) will be disappeared. I will later present the testing results.

    Like
    • YZ
    • Eomi
    • 4 yrs ago
    • Reported - view

    I have confirmed that the convergence becomes better if the capillary pressure function is lowered. The convergent problem near breakthrough occurs when the accumulated strong capillary pressure tends to be relaxed after facing a zero capillary pressure (due to 0 saturation) at the INACTIVE outlet. Thus for rock with a large capillary pressure, the simulation "naturally" becomes difficult near breakthrough and there is no solution (?) if mesh size is too fine. 

    Like
Like Follow
  • 4 yrs agoLast active
  • 4Replies
  • 449Views
  • 2 Following