Posted Wed, 18 May 2022 13:56:12 GMT by Davide Furlanetto Università di Padova - Dipartimento di Ingegneria Civile, Edile e Ambientale
Goodevening everyone, I'm looking for some help about a project that I'm developing at the university.
I have to calibrate the conductivity field in a tridimensional domain using FePEST. I'm using pilot points in some of the mesh layers and interpolating with ordinary Kriging, using an exponential anisotropic variogram. Once the calibration ends, I export the K field in FEFLOW and what I see is a "blotchy"/"patchy" (sorry I don't know how to explain) distribution of the conductivity, while I expected a more "continuous" K field. Particularly this occurs in the horizontal layers, while in vertical the distribution is more plausible. I attach the image of a slice and of a section to show what I mean.
Initially I though this could be related to the Kriging options but I tested various correlation lengths and the problem persists anyway.
Any idea about the reason for which this can happen? Any suggestion?
Thank you very much in advance!

Posted Wed, 25 May 2022 11:57:23 GMT by Carlos Andres Rivera Villarreyes Global Product Specialist - FEFLOW
Hi Davide,

You do not need to wait until the end of the optimisation to see the results. Even before you start the run, you can go to Edit -> Save Parameters in FEM. This will trigger the interpolation using the settings behind the Parameter Definitions, so you can check whether the resulting interpolation fulfil your expectations.
The "patchy" distribution is either attributed to the given range in the variogram, the number of pilot points and/or a combination of these two. If I were you, I would simply increase the number of pilot points (as a test by a factor of 2) and check the results. As a way to debug, FePEST writes in the GUI the average pilot point separation, which should go in accordance with the variogram range.

You must be signed in to post in this forum.