Run ends early
Hi,
I was working on a larger block mesh using a doublet configuration for an aquifer, over 10 years. When I made the mesh finer, the simulation timed out and quit after 16 hours. The overall size is the same, but the blocks are smaller. Why would the simulation be ending at 16 hours if it should run 10 to 30 years? Under what circumstances should I expect this in the future? Is there a prompt/keyword for the INFILE that would help me avoid this problem?
5 replies
-
Hi Nicholas.
From your mesh file it seems you are using a single horizontal layer (NZ=1).
The infile reports initial conditions only for the injection element: P=5.86E6 Pa, T=12°C and Xair=0. If you do not need to input any air, why using EOS3 instead of EOS1?
No air at all assigned may give some convergence problem to EOS3.
GENER: you have not assigned the enthalpy for the injected water, as you want to fix the injection T at 8°C using INJ1 rock domain with infinite specific heat.
Looking at the specifications of your problem I guess you should not have two-phase flow in the system, unless your extraction rate is too high for the rock permeability assigned. If P declines too much at the production block, as there is no air and the T is very low (8°C), you would have two-phase conditions at P close to zero (Psat at 8°C).
This might explain why your problem runs with large blocks and not with a finer grid.
I think you should check if your production rate is too high for the aquifer permeability assigned, and in this case you need to reduce the well doublet rates. The aquifer should stay in liquid conditions and the PCAP function used should not have any role in the simulation.
Regards,
Alfredo
-
Thanks for the response, Alfredo. This is a cooling process with an injection temperature of 12C. My understanding is that the enthalpy of the fluid need not be assigned for the EOS given a fixed mass flow rate and temperature.
I have assigned a lower mass flow rate. That did nothing.
I ran the INFILE in EOS1. That did not work.
Admittedly, I am new to TOUGH (like a few years with no formal training, lots of irritation, and an abundance of admiration for people with the patience and time to learn from sources unknown), but this failure to converge does not make much sense. -
Nicholas,
You should check why the simulation is not converging, looking at messages, at the element causing the non convergence and at its thermodynamic state, at the equation causing the non convergence, etc.
Usually this allows to understand what is not working and fix it.
Alfredo
-
Hi Alfredo,
I am not aware of any error handling in TOUGH. Could you point out where there would be messages indicating errors? Attached is a recent output file, if that helps.
Best wishes,Nick
-
Nick,
I apologize, but I missed your post. I imagine is too late for a feedback...
Any way, in your run the time step is always cut because of gas phase appearance in eleme C1f21 driven by a negative PX. I guess this is the extraction well block (no GENER info are printed because of premature run stop). As your aquifer T is only 8°C, and there is no air, two-phase conditions would occur at the very low saturation T of water at 8°C
A1j 3( 9, 2) ST = 0.511000E+03 DT = 0.256000E+03
...ITERATING... AT [ 10, 1] --- DELTEX = 0.512000E+03 MAX. RES. = 0.179135E-01 AT ELEMENT A1j 2 EQUATION 3
$$$$$$$$$$$$$$$$ GAS PHASE EVOLVES AT ELEMENT *C1f21* $$$$$ XAIR = 0.000000E+00 PX = -.671870E+05 PG = 0.107205E+04
...ITERATING... AT [ 10, 2] --- DELTEX = 0.512000E+03 MAX. RES. = 0.763317E+02 AT ELEMENT C1f21 EQUATION 1
The code is able to initialize the two-phase conditions but during next iteration another element has P>1.E+9 Pa, which is the limit for the IAPWS-IF97 correlations for the liquid water field. This drives the time step reduction.
+++++++++ CANNOT FIND PARAMETERS AT ELEMENT *C1d13* XX(M) = 0.102317E+09 0.000000E+00 0.813613E+01
TOUGH STATUS: Failed in EOS. IGOOD= 5
Subsequentlt, many other elements experience a phase transition to two-phase conditions, but time step is cut because some element goes to extremely low negative P.
Looking at the C1f21 element P at the converged step for which printout is available, you can notice that PX=1,201 Pa, which is really very low. It seems you were extracting too much for the aquifer capabilities.
Regards,
Alfredo
ELEM. INDEX PRES TEMP SAT_G SAT_L X_WATER_G X_AIR_G X_WATER_L X_AIR_L REL_G REL_L PCAP_GL DEN_G DEN_L POR
(PA) (DEC-C) (-) (-) (-) (-) (-) (-) (-) (-) (PA) (KG/M**3) (KG/M**3) (-)
C1f21 9561 0.12010E+04 0.8000E+01 0.0000E+00 0.1000E+01 0.0000E+00 0.0000E+00 0.1000E+01 0.0000E+00 0.0000E+00 0.1000E+01 0.0000E+00 0.9865E-02 0.9999E+03 0.2000E+00