No convergence in NSIN at - non-isothermal simulation T2WELL/ECO2N
Dear TOUGH community users,
I am trying to simulate CO2 injection using T2WELL/ECO2N. In the model, the grid is radial and both wellbore and reservoir are initially fully saturated with water X3=0. At the very beginning after 2 seconds, I get the error message in the output file as shown in the PNG file.
I have read previous posts about this problem, but the suggested solution to change MOP(16) to 6 did not work for me. I am attaching both input and output files. I would very much appreciate your help.
The code actually recovers from the error message you show, and continues running. The outputs at 5 and 10 seconds look pretty normal. But at about 18 seconds, the code begins to have repeated NSIN failures, the timestep is repeatedly cut, and no more progress is made. I think the problem is due to element a 2 trying to go from single-phase liquid to two-phase as the amount of dissolved CO2 exceeds the solubility. Of course, it should be able to do this, but for some reason the iteration in NSIN is not finding good two-phase variables. I have had this NSIN problem in the past, and I don't really know how to fix it, but you can try varying the parameters of your problem and see if anything helps. For example, your problem is pretty cold, so you could try a warmer version (maybe inject at 10 C instead of 5 C). You could try decreasing injection rate. You could try turning off the wellbore aspect of the problem and see if plain TOUGH also has this failure - if not, then maybe there is something problematic about your wellbore specification. You could try running an isothermal version of the problem. I notice your rock densities are pretty high - maybe 10 times more than I would expect. I don't think that is causing a problem, but check that is what you really want. I notice that pressure is decreasing at the injection element with time - is that what you expect? Currently, you are using the default convergence criterion of RE1=1e-5. You could try 1e-6.
You might want to set MOP(5)=3 to show phase changes - this won't change how the code runs, but it can help diagnose problems.