Convergence issue with 1D model for both PFLOTRAN and TOUGH2
I thought the developers of TOUGH2 maybe interested in this problem as well. I'm using EOS3 to solve the two phase problem. The left end of this model is atmospheric open air boundary condition at 0.01 water saturation. The right boundary is fully saturated water at 10 MPa. we are trying to mimic excavated salt cave 1D direction into intact salt. Air region is fully saturated. DRZ (disturbed rock zone) is partially saturated due to mechanical porosity from excavation and the rest is fully saturated intact salt.
This model is having convergence issue with PFLOTRAN, but it is the same case with TOUGH2.
Let me know if you have any suggestions.
7 replies
-
Hi Heeho,
Without fully understanding what the physical problem is you want to solve, and without trying to make it run, there seem to be many errors and inconsistencies in this input file. For example, you specify unphysical van Genuchten parameters (lambda = m = CP(1) = RP(1) must be in the range 0 < lambda < 1; you specify a value of 4.66; as a result, both krl and krg are 1.0, and Pc=0; the chosen weighting scheme, i.e., MOP(11)=3, is inappropriate; not sure why a semi-analytical heat-exchange solution was chosen for this problem; etc.).
Given all these issues, I'm not surprised at all that TOUGH refuses to converge. I don't know how this faulty input file was translated to PFLOTRAN, and what problems arose there.
This may indeed be a challenging test case that could be analyzed in a code cross-comparison exercise. However, I propose you check your input file carefully, make sure it all makes physical sense, before such a comparison is attempted.
Stefan
-
You're correct, I converted lambda parameter incorrectly.
I've checked input several times to make corrections to input. It is now actually producing results but cannot converge near the end. I cannot tell if I've made a mistake in the input or if the model really cannot converge. Please let me know if you have any suggestions.
-
Never mind. I resolved the problem. DRZ had residual saturation for gas of 0.0. it works with 0.2.
-
Heeho,
Set MOP(11)=2, MOP(16)=4, and RE1=1.0E-06, and it runs through without any problems.
However, I cannot comment on the results, as I'm unsure about what it is you want to simulate. For example, I do not understand why you specified capillary forces in the air. The chosen thermal parameters and tortuosity factors are somewhat questionable as well. Nevertheless, I hope these few pointers help you with your research.
Stefan
-
Do you have Brooks-Corey Burdine model and Val Genuchten Mualem model for both krl and krg in TOUGH2 EOS3?
-
Heeho,
No (see Appendix G and F of the manual for available functions). However, simply go to t2f.f, subroutines RELP and PCAP and add the functions you need.
Stefan
PS: The Brooks-Corey-Burdine as well as the van Genuchten-Mualem and van Genuchten-Burdine model are implemented in iTOUGH2, see Appendix A6 of iTOUGH2 Command Reference.
-
Thank you.