Error Message

Yogendra Narayanan Suresh Kumar 

There is a blank line in the ROCKS block that should not be there, between the line starting with 03 and the line starting with 12.  Try removing that line.  In general, when you are in a block, like the ROCKS block, TOUGH expects things in a certain order and format, and extra blank lines cause problems.  The ROCKS block should end with a blank line, and yours does that correctly.  In the input file, in between blocks, you can insert comment or blank lines as you like, but not within a block. 

@Yogendra Narayanan Suresh Kumar

When I try the infile you posted as INFILE (1) (I renamed it INFILE), TOUGH3 ran 343 steps to 118.9 seconds and quit with a convergence error because the pressure got too big and went out of range of the EOSS (0.6e8 Pa). There were no errors in INFILE, the pressure just got big because you are injecting CO2 into a finite model.

If you have the executable in the same folder as your input file, be sure to run the command ./tough3-eco2n to get the local version of the executable and the local version of INFILE.

Yogendra Narayanan Suresh Kumar You are injecting CO2 at one end of your core, and I think you need to allow fluid to flow out of the other end of the core, to actually model a core-flooding experiment.  You could accomplish this by holding the pressure fixed at its initial value at element A11 1.  The simplest way to do that in TOUGH is to set the volume of element A11 1 to a huge value (say 1.E30 m3).  With this huge volume, any fluid that flows into element A11 1 "gets lost", and the pressure at element A11 1 does not change.  Then there will be a pressure gradient between the CO2 injection element A1112 and the constant-pressure element A11 1, and fluid (your initial water and injected CO2) will flow through the core.  You could also hold the pressure at element A11 1 fixed at some other value than the rest of the core starts at, by specifying it in the INCON block, if your core flooding experiment specified outlet pressure is held fixed at a certain value.  

By varying the parameters of your relative permeability and capillary pressure curves, you should see different saturation conditions.  Note that the curves you are specifying in the ROCKS block (IRP=7, ICP=7) are not hysteretic, contradicting what your title line suggests.  I strongly recommend continuing with non-hysteretic curves until everything is working perfectly in your model.  Hysteresis adds a lot of complication, and in my opinion it should not be the first thing you try.

Yogendra Narayanan Suresh Kumar 

You asked the code to run 1000 time steps (PARAM block, first line, columns 5-8) and it did.  Then it ended normally. It reduced the time step twice (as you see in the message on the screen), but this is not an indication that the run failed.

The main place to look for information on your run, besides the messages printed to the screen, is the file OUTPUT.  This shows that the simulation ended with no error message.  When you are a new TOUGH user, it is helpful to read through the entire OUTPUT file.  But I usually go to the bottom of the file and then search backwards for the string "OUTPUT DATA".  This gives you the state of things when the simulation ended.  Another tip is to start from the beginning of the OUTPUT file and search for the string " ST " repeatedly.  This will show you the time step progress.

I tried increasing the number of time steps requested to 9999, but the simulation did not reach your requested end time of 86400 seconds.  Then I noticed that your maximum time step length is set to 3600, right-justified with no decimal point.  TOUGH interprets this as 0.36 seconds.  I changed this to 3600.0, and the simulation took 70 steps to reach 0.625 days, at which point it stopped because it recognized that nothing was changing in the model - that is, you had reached steady state. I think this makes sense physically.  I think your model is working properly. 

Try adding the decimal point (make sure 3600.0 fits in the proper columns 31-40), and see if it works for you.

Yogendra Narayanan Suresh Kumar  In general, I expect that when permeability is decreased for a constant mass injection rate problem, pressure will increase, CO2 density will increase, and you would therefore expect gas saturation to decrease, since denser CO2 takes up less room.  I checked your problem by cutting permeability by a factor of 10, and I was surprised to see gas saturation increase, just as you found.  I don't know why this is occurring, but there are other things that change when pressure increases, such as the interplay between imposed pressure gradient and capillary pressure, the amount of water that evaporates into the co2 phase, and the amount of co2 that dissolves into the aqueous phase.  You can turn off capillary pressure by changing ICP=7 to ICP=8.  Note that for your problem, density difference for a factor of 10 or 100 change in permeability is very small.  I tweaked your problem by increasing overall pressure and decreasing permeability so a big pressure gradient and a big density difference occurred (basically I converted from gas-phase CO2 to supercritical CO2, where I have more experience).  Even then, gas saturation was not smaller for lower permeability.  I do not know why this is.  It looks like more water is evaporating into the gas phase at lower permeability, which could increase gas saturation, but the effect seems too small.  If someone else reading this post has any ideas, I would like to hear them.

When you get counter-intuitive results from TOUGH, it is good to vary different parameters to see if you understand the changes you get.  Remember that with multi-phase, multi-component flow there are coupled non-linear processes at play, and you might not have good intuition about how they interact.  Your problem is good for testing because it is a simple 1D column.  Sorry I cannot be more help.

Yogendra Narayanan Suresh Kumar That makes sense to me.  Changing porosity generally just changes the time scale on which flow processes occur.  So if you are just looking at steady state results, then I am not surprised that you get the same gas saturation for different porosities. 

Yogendra Narayanan Suresh Kumar The only way I know how to do that is to co-inject water vapor with the CO2, so the gas phase that forms is already saturated with water vapor, so no evaporation of in-place water occurs.  Note that the amount to co-inject will depend on the pressure conditions.

Yogendra Narayanan Suresh Kumar  Definitely.  You can try changing the various parameters and see what effect it has.  See the TOUGH user's guide for a description of what the different parameters are.  For IRP=7, the two that seem most important to me are RP(1)=m and RP(2)=Slr. If you set RP(4)=0, you should get very different results, because there is a flag that says if RP(4)=0, use krg=1-krl.  Even though they are not very realistic physically, you might want to try linear relative permeability (IRP=1) just because it simplifies the problem.

Stefan Finsterle Thanks for an elegant explanation.  I guess after all these years, I am still biased to  think in terms of liquid-saturated systems. I agree with you that gas saturation will increase or decrease as needed to produce the effective conductivity and capillary pressure that the overall system demands. 

But I think your answer is incomplete.  What about the pressure gradient?  Flow rate in a 1D system with a constant-rate mass source at one end and a constant-pressure boundary at the other end should be equal to effective hydraulic conductivity times pressure gradient.  If intrinsic permeability is decreased, pressure at the injection element could increase, in addition to saturation increasing, to maintain gas flow equal to injection rate.  If imposed pressure gradients are small compared to capillary pressure gradients, I imagine that saturation changes will dominate, but I don't think that will always be the case.