GSLIB+Anisotropy permeability

Ramin,

Did you set MOP(7)=9 and check near the top of the output file? That's where you find all three permeabilities. In some (but not all) EOS modules, the X-permeability is also printed to each output block. You may also use the iTOUGH2 option OUTPU (see Section 3, Table 4 of the Enhancement manual, http://tough.lbl.gov/assets/files/02/documentation/TOUGH2-In-iTOUGH2_Enhancements.pdf) to print all three permeabilities.

Also, I'm a bit confused about you mentioning GSLIB. Note that the anisotropy factors to be specified in GSLIB create anisotropy of the larger-scale permeability structure (i.e., on a scale larger than that of the individual TOUGH grid blocks), whereas the 3 directions you specify in block ROCKS account for anisotropy on the scale of or below that of an individual grid blocks. If you specify anisotropy in ROCKS, they will be applied in addition to the structural (larger-scale) anisotropy you create with GSLIB.

To further clarify: GSLIB generates structural anisotropy, typically as the result of a depositional process, for example, elongated clay lenses embedded in a sandy matrix, whereas ROCKS anisotropy specifies local anisotropy, for example, that generated by the small-scale fabric of the clay particle arrangement within the clay lens, which tends to have a preferential orientation.

Hope this helps,   

Stefan

Dear Ramin,

I likely have misunderstood what data you have and what you want to do, but it seems I have to go back to my previous reply:

(1) anis1 and anis2 refer to larger-scale anisotropies, so printing out all gridblock-scale three permeability values for all elements does not help at all in determining anis1 or anis2, even if you want to use them in a post-processing step to calculate an effective permeability (unless you use full tensors etc.) and then compare them to your effective, "measured permeability" (I'm not clear on what scale they are). 

(2) Since you specify the gridblock-scale anisotropy in block ROCKS, i.e.,

a_31 = PER(3,mat)/PER(1,mat)

and since this ratio remains constant even if you apply GSLIB, there is no need to print out all three permeabilities for all elements. You have the element-by-element permeabilities PER(1,iel), which are now heterogeneous as you used GSLIB. The other permeabilities are then simply given by, e.g.,

PER(3,iel) = PER(3,mat) * (PER(1,iel) / PER(1,mat) = PER(1,iel) * a_31

(3) If you still want to get all three permeabilities, just follow the instructions I gave you in my previous reply!

(4) Assuming I misunderstood your data and objectives, consider taking advantage of iTOUGH2-GSLIB's capability to directly estimate anis1 and/or anis2 by calibrating your model agains appropriate measurements of state variables (i.e., not "measured" parameters), such as flow rates, saturations, tracer concentrations, etc. For a similar application, see Finsterle and Kowalsky, Joint hydrological-geophysical inversion for soil structure identification, Vadose Zone J., 7:287–293, doi:10.2136/vzj2006.0078, 2008.

Good luck!

Stefan

Ramin,

If you have no data, there is no way iTOUGH2 (or anything/anybody else) can give you an answer. If you do have data (either measured permeabilities or data on the system state), iTOUGH2 can help you, using the various approaches I described in my previous posts.

Final suggestion: If you have "no idea on horizontal correlation length", make an assumption! Then change it in a sensitivity analysis and see how much it affects the results you are interested in. If the horizontal correlation length turns out to be an influential parameter, go to the field and collect informative data.

Best,

Stefan