Gravity and periodic boundary conditions for relperm upscaling about opm-upscaling HOT 9 CLOSED

use the original relperm upscaling code

Just out of curiosity, do you have an idea of what time point corresponds to the original relperm upscaling code? I.e., whether it was based on sources from one year ago, three years ago, five years ago or ten years ago?

from opm-upscaling.

We can double check tomorrow, but I believe we went back to the sang based code to get a reference solution.

from opm-upscaling.

Just checked. We seem to have incorrect results all the way back to May 2012. Louis can send you a test case.

from opm-upscaling.

We seem to have incorrect results all the way back to May 2012.

Okay, that does at least narrow the search region a bit.

Louis can send you a test case.

Yes please.

from opm-upscaling.

There's a wonderful dichotomy in the discretisation of the gravity term in the upscale_relperm utility (and maybe others, I haven't verified this). While the grid constructor assumes that all lengths and depths are given in units of metres the gravity term discretisation assumes that the depths are given in units of centimetres.

I think I know the historical reason for this assumption. All of the models used in the original development of the upscaling utility used centimetre length scales. Would you mind, as a quick check/hack, to run the tests using

-gravity 981

rather than 9.81 and see if you get better results this way?

from opm-upscaling.

Today I finally had occasion to look into this more closely. My preliminary conclusion is that the gravity handling is fundamentally broken (not to mention, explicitly turned off in the low-level upscaling code, despite being actively queried in upscale_relperm.cpp).

Moreover, some parts of the code assume centi-metre length scales, while others use metres. That, in turn, affects pressure gradients and flow rates--at least when the flow is driven by pressure drops.

Fixing this will require days of implementation work--not to mention substantial testing efforts--and will affect code in other OPM modules.

from opm-upscaling.

I have finished my initial implementation of this feature so I'd like to enlist some testing help to assess the state of completion. At this time I'm a little reluctant to create a PR because there may still be a lot of work remaining depending on the testing outcome. Therefore, if you would like to help test, please use the following commands to access my branch

git checkout -b bska_capillary-gravity-upscaling
git pull git://github.com/bska/opm-upscaling.git capillary-gravity-upscaling

from opm-upscaling.

Excellent! I can confirm that the branch builds (when merged on top of master, to account for the opm-parser API change). There is a test failure:

14/16 Test #14: upscale_relperm_BCf_pts30_surfTens11_stoneAniso_stoneAniso_27cellsAniso ............................***Failed    0.18 sec

This is expected since this test essentially is a regression test, and the fixed bugs should be expected to change the results for some of the tests.

from opm-upscaling.

This was fixed a year ago, closing.

from opm-upscaling.