Well modeling
Hi everyone,
I use TMVOC to model remediation of soils and groundwaters. I first need to model pumping tests and I was wondering how could I precisely model wells and observation wells.
First, I made the well grid cells "empty" (porosity of 0.999), very permeable (kh,z=10-7m²), applied zero capillary pressures and made the fluids "all perfectly mobile" (kr=1). Other data are similar to the ones of the rock formation where the wells are drilled (density, etc.)
In the saturated zone, it seems to work fine but calculations stops due to problems in the unsaturated zone.
If you have any tips on this subject, please feel free to contact me.
Cheers,
Quentin
7 replies
-
Quentin,
Just a brief tip:
(1) Don't use "all perfectly mobile" (it is unphysical). Use linear functions or Corey.
(2) Set kz=1E-8, but kh equal to that of the formation (and.or set the nodal distance between the well and the formation to exactly zero).
Stefan
-
Thank you very much Stefan,
It works fine now!
-
Hi Stefan,
I just wanted to check with you which parameters I should enter for the linear function for relative permeability in TMVOC (IRP=1):
Could you please tell me if it is correct :RP(1) = Swr; residual water saturation
RP(2) = Sgr; residual gas saturation
RP(3) = Swm; maximum water saturation
RP(4) = Sgm; maximu gas saturation
RP(5) = Sor; residual napl saturationThank you in advance,
Quentin
-
Hi Quentin,
May I suggest that you simply try it out? (I'm giving you this "lazy" answer because I would like the Forum to help users get the information they are seeking from the standard TOUGH2 output file. Moreover, there are several versions of TMVOC, and I don't know which one you have. Finally, I consider PCAP and RELP the responsibility of the user - which is the reason for the instructions in the manual on how to add your own functions).
Here are some ways to find out how IRP=1 in TMVOC works:
(1) If you have the source code, just go to RELP and check the implementation, or code your own linear functions.
(2) Set KDATA=4 and look in the output file what KRGAS, KRAQ, and KROIL are for the given saturation, and compare them with the expected outcome given your assumptions about RP(1)-RP(5).
(3) Set up a test case consisting of a single element and a single small time step; set MOP(5)=9, and you will get the relative permeabilities for each phase once you figure out the structure of the PAR array (which is good to know anyway; see Figure 6.3.1 in the TMVOC manual).
(4) Wait and hope that a good soul from the TOUGH Forum Community gives you the straight answer...
Stefan
-
Thank you Stefan for your answer.
I checked the code and, for the moment, I use the "debuging" relative permeability curve.
I will modify later the Stone I model so that kro is always GE than 0.Just a quick question: what is the justification in setting kh the same as the formation value?
And I do not understand the concept of nodal distance between the well and the formation. Could you please explain it to me, or give some references I could read?Thank you very much for your help!
Quentin
-
Quentin,
Sorry, I don't understand what kh (horizontal permeability?) and the context of your statement "setting kh the same as the formation value".
Usually, set nodal distance from the well element to the interface with a formation element (or from a fracture element to the interface with a matrix element) to zero. This signals to TOUGH2 that it should take the permeability associated with "the other " element (i.e., the formation or matrix element), not the (usually much higher) well or fracture permeability. In essence, it overwrites the weighting scheme (see MOP(11)) to make sure the realistic (lower) permeability is used for such a (serial) connection. It is described somewhere in the manual. Alternatively, you could set the horizontal permeability of the (presumably vertical) well to the same permeability as the formation, and have a non-zero nodal distance (which should be set to a very small value, as there is unlikely to be a pressure loss in radial direction within the well).
Hope this makes sense.
Stefan
-
Hi Stefan,
Thank you for your help, it makes sense now!
Cheers,
Quentin