Error Message

Yogendra Narayanan Suresh Kumar 

Please check the alignment of the first line in the PARAM block, in particular your MOP entries.  They do not seem reasonable to me.  If it is not just a simple alignment mistake, we should discuss your MOP choices further, as several are not typical.  I always recommend MOP(1)=1 and MOP(5)=3 and MOP(7)=1 to aid in debugging.

@Yogendra Narayanan Suresh Kumar

At first I thought your boundary elements were not big enough.  If you want them to act like constant-pressure boundaries, the volume should be even bigger, something like 1.E30 m3. But then I noticed that only the pressure at the injection element is too high, and the pressure at the outer boundaries is unchanged from its original pressure.  Maybe your permeability is too low to enable the pressure response to injection to spread out through the model.  Try higher permeability or lower injection rate or both.  You could also try less-interfering relative permeability curves by increasing RP(1)=0.457 to something like 0.8.

@Yogendra Narayanan Suresh Kumar Stefan Finsterle kenny

Hm, I seem to be stuck.  I have tried a bunch of things, and the code always fails, usually in a really odd way, where it starts trying to do precipitation (we are using ECO2N).  There is no salt in the problem, so there is nothing to precipitate, so the code is apparently confused.  Pressure and temperature are in the normal range when this happens.  One clue about confusion is that the precipitation is in the very first grid block, and this is not a place where anything very interesting is or should be happening. Another clue about confusion is that when I use a different executable (for Ubuntu instead of for Windows), the code quits in the same place , but instead of precipitation, I get NaN (not a number) for pressure. This is the exact same input as Yogendra provided, and his run quits without precipitation or NaN, just a steadily decreasing time step.

In the olden days (TOUGH or TOUGH2), I would have thought this is a dimension problem, where variables are being over-written when some other variable gets too big for its assigned range, but I don't think that should happen with TOUGH3.

I found a lot of minor issues in the input file that I modified, but nothing helped.

Here is a quick summary:

This is a two-dimensional (x,z) problem, with a uniform rectangular grid: elements are all 2-m tall and 5-m wide, and there are 25 elements in the z direction and 100 elements in the x direction.  scCO2 is injected into a water-filled medium mid-way along the left edge and all the elements along the right edge have a huge volume, to keep pressure fixed.

Some possible problems, and what I did to fix them:

• No gravity equilibration was done, so pressures held fixed along the right edge are all the same, and not in hydrostatic equilibrium (I turned off gravity)
• Injection rate was large compared to model pore space and requested run time - the injected CO2 would fill the entire pore space within the first 4 days of a 10-year simulation (I tried cutting down injection rate by a factor of 100)
• The relperm and Pcap functions specify Sls=0.9, which I have no experience with, so I changed it to Sls=1.0
• Rock compressibility was kind of small (4.5e-10) so I increased it by a factor of 10
• Rock permeability was high (~1e-11) so I decreased it by a factor of 10
• Capillary pressure strength was weak (v.g. alpha~1e-2) so I decreased alpha by a factor of 100
• Problem was isothermal, so I tried making it non-isothermal (the opposite of what I usually do to debug, but I was getting desperate)

I hope someone can figure this out.  Thanks, Chris