0

Infinite Acting Aquifer

I have a TOUGH3 model (EOS7C) that is showing signs that the model extent is not wide enough in the East and South directions (pressures increasing over time). I don't have any geologic data to extend the grid, so my options are to manually take averages and "make-up" subsurface data (which seems to be tedious and causing a bit of a headache) or I've heard it rumored that I could set up an "infinite acting aquifer" along the boundaries where my model is showing signs of boundary condition influence.

I have looked through the TOUGH3 documentation for this, but couldn't find what I think I need, so I was curious if anyone has experience setting up an "infinite acting aquifer" and could give a summary of what this looks like in tweaking any respective TOUGH3 blocks.

Thanks!

6 replies

null
    • Mikey_Hannon
    • 1 yr ago
    • Reported - view

    Hey Erik! What you'll need to do is connect additional elements at the lateral extents of your model elements having "infinite" (typically 1.0E+50) volumes. The conditions (i.e., primary variables) of these elements will be maintained at the values they are initialized at. Typically, the process is to determine what the conditions of those elements need to be by running a steady-state stage of your simulation where the elements have finite volumes. Then you increase the volumes of the boundary elements to 1.0E+50 before initiating the transient stage of your simulation. If you want to make sure the conditions of these elements maps to the edge of the active nodes connecting to them, you would need to reduce the D1/D2 values in the CONNE block as well, so that the distance from the boundary element central node to the interface is something very small (typically, 1.0E-10).

    If you're doing this in TOUGH3, be careful to remove the MESHA and MESHB binary files that are created from previous runs if you make any changes to the ELEME or CONNE blocks.

    • Erik_W
    • 1 yr ago
    • Reported - view

    Thanks Mikey for your quick response. So, just to clarify what I think you're saying. In the attached image, my model is in the red box and the larger cells to the right and down were the manual grid I started on. What I believe you're saying is that I need to add two cells (one to the east and one to the south) and make them as the one to the right as wide as necessary to achieve that 1E50 and the one on the bottom as long as necessary to achieve that 1E50 (since the other parameters are determined by the current model).

    And secondly, I'd need to tweak the CONNE block manually as you've noted above.

    Am I understanding that first part correctly?

      • Mikey_Hannon
      • 1 yr ago
      • Reported - view

      Erik W, you've basically got it right. You won't necessarily need to adjust the dimensions of the additional cells you include, although that would work. In that case, though, it'd be very important you adjust the D1/D2, because the distance from the interface to the center node of your boundary elements will be practically infinite as well.
      Alternatively, you can just create elements with the same dimensions as the one they're connecting to and hard-code the volumes of those elements to 1.0E+50 in the ELEME block.

      By the way, is this a 2D model, or just a single layer of a 3D model?

      • Erik_W
      • 1 yr ago
      • Reported - view

      Mikey Hannon It is a 3D model. That screen-shot is just the plan-view of a single model layer.

    • Finsterle GeoConsulting
    • Stefan_Finsterle
    • 1 yr ago
    • Reported - view

    Erik,

    You could use program AddBound, which does essentially what Mikey proposes. Check it out at 

    https://www.finsterle-geoconsulting.com/download

    The usage of AddBound may not be totally intuitive. You have to select a subdomain (by defining the coordinates of a bounding box) that captures the last column (or row) at the boundary of your red domain. All these elements will then be connected to a single, large, "blue" element. 

    Be aware, however, that this approach will give you a constant pressure/temperature/concentration boundary at the interface between the red and blue domains (or at the nodal distance from the blue element to the interface you can select in AddBound). This is different from an "infinitely-acting" reservoir. You replace a Neumann no-flow boundary with a Dirichlet boundary, i.e., you are not simulating "free" fluid flow to a large half space. Your proposed extension with a discretized subdomain (you could use gridblocks whose size increases relatively fast as you get farther away from the red domain) would be a better representation of an infinitely-acting aquifer (but then again, all aquifers I've ever "seen" are not inifinite...).

    Cheers,

    Stefan

      • Erik_W
      • 1 yr ago
      • Reported - view

      Stefan Finsterle thanks for the recommendation of this additional tool. I'm still getting TOUGH under my belt, so I'll try that route first, but it's good to have recommendations of other tools to solve problems.

Content aside

  • 1 yr agoLast active
  • 6Replies
  • 34Views
  • 3 Following