Help Needed: EOS3 Non-convergence in Compressed Air Injection into Aquifer
Help! TOUGH2 EOS3 simulation of compressed air injection into aquifer fails after few steps. Adjusting parameters/time steps didn't work. Seeking root causes & debugging tips!
4 replies
-
Looking at the output, the run stops because the P exceeds the allowable limit of 1000 bar in eleme 41717. This means that probably the assigned injection rate of 20 kg/s is too high for the rock petrophysical properties. Either reduce the injection rate or reconsider the petrophysical properties.
Element "41717" belongs to rock domain UNDER with very low permeability (1.E-20 m2). Injected air should not enter into the UNDER rock domain, as AQUIF(er) permeability is much higher (1.E-12 m2).
I guess you should reconsider the option about the interpolation of permeability at the interface. Using MOP(11)=2 should be more appropriate than using 0 as you did.
I suggest to activate the printout of phase changes ( MOP(5) ) for further diagnostic purposes.
Regards,
Alfredo
MOP(11) = 0 *** CHOOSES EVALUATION OF MOBILITY AND ABSOLUTE PERMEABILITY AT INTERFACES.
= 0: MOBILITIES ARE UPSTREAM WEIGHTED WITH WUP. (DEFAULT IS WUP = 1.0). PERMEABILITY IS UPSTREAM WEIGHTED.
= 1: MOBILITIES ARE AVERAGED BETWEEN ADJACENT ELEMENTS. PERMEABILITY IS UPSTREAM WEIGHTED.
= 2: MOBILITIES ARE UPSTREAM WEIGHTED WITH WUP. (DEFAULT IS WUP = 1.0). PERMEABILITY IS HARMONIC WEIGHTED.
= 3: MOBILITIES ARE AVERAGED BETWEEN ADJACENT ELEMENTS. PERMEABILITY IS HARMONIC WEIGHTED.
= 4: MOBILITY * PERMEABILITY PRODUCT IS HARMONIC WEIGHTED.
-
Thank you for your answer, which helped me resolve the interface issue. I greatly appreciate your efforts. After fixing the interface problem, I had to adjust the injection rate to 1 kg/s to achieve convergence. The simulation converged earlier than expected, completing in just over a day—much faster than I anticipated. Can I only extend the runtime by modifying the injection rate? Additionally, as a TOUGH2 beginner, I checked the manual but couldn't find how to activate the phase change printout (MOP(5)). Could you advise on the setup? Finally, I wish you good health and success in all endeavorsimage
-
After reducing the injection rate, my simulation still terminates prematurely—it only runs for 16 days. Could there be an issue with my initialization setup? I sincerely appreciate your guidance
-
About printing of phase transition messages, you may use MOP(5)=3. I suggest to try the effect of different choices by just running short simulations when you have doubts on some input option.
Your run still stops because the P goes too high above 1000 bar which is the limit of IAPWS-IF97 correlations for liquid water. Thus, you are still injecting too much for the model petrophysical properties assigned. To note the highest P is in eleme 51515 located at a deeper elevation (-938 m) than the injection element 91515.
+++++++++ CANNOT FIND PARAMETERS AT ELEMENT *51515* XX(M) = 0.100000E+09 0.119324E-01 0.384482E+02
+++++++++ REDUCE TIME STEP AT ( 39, 1) ++++++++++++++++ NEW DELT = 0.431213E+00You are using the same petrophysical properties for all the rock domains, while one may expect different properties for the aquifer and the boundary domains UPPER and UNDER with much lower permeabilities.
ROCKS----1----*----2----*----3----*----4----*----5----*----6----*----7----*----8
AQUIF 2 2.600E+03 2.000E-01 1.000E-12 1.000E-12 1.000E-11 2.510E+00 9.200E+02
1.000E-10
7 5.000E-01 1.000E-01 1.000E+00 1.000E-01
7 5.000E-01 5.000E-02 1.000E-05 1.000E+05 1.000E+00
UPPER 2 2.600E+03 2.000E-02 1.000E-20 1.000E-20 1.000E-20 2.510E+00 9.200E+02
1.000E-10
7 5.000E-01 1.000E-01 1.000E+00 1.000E-01
7 5.000E-01 5.000E-02 1.000E-05 1.000E+05 1.000E+00
UNDER 2 2.600E+03 2.000E-02 1.000E-20 1.000E-20 1.000E-20 2.510E+00 9.200E+02
1.000E-10
7 5.000E-01 1.000E-01 1.000E+00 1.000E-01
7 5.000E-01 5.000E-02 1.000E-05 1.000E+05 1.000E+00
WELL 2 2.600E+50 2.000E-01 1.000E-12 1.000E-12 1.000E-11 2.510E+00 9.200E+02
1.000E-10
7 5.000E-01 1.000E-01 1.000E+00 1.000E-01
7 5.000E-01 5.000E-02 1.000E-05 1.000E+05 1.000E+00The parameters used for RELP with VG model need to be checked.
Below the RELP for aqueous and gas phases using the parameters given in ROCKS. It seems that the overal mobility of phases at around SL=0.67 is very low. REPG is almost zero up to SG=0.2. I would consider to increase the mobility of both phases at least in the AQUIF domain, unless you have experimental data supporting the choice of parameters made.
Regards,
Alfredo
-
