[Opm] Use external mesh with Flow OPM and unstructured meshes

[Atgeirr Rasmussen]

Hi Antoine,

The trick you used does not tell OPM in any way that the grid is tetrahedral or hexahedral, it just provides the necessary data as a soup of volumes and connections, and not as a proper grid. The VTK output assumes that we have a proper grid.

Did the simulation produce expected results?

About tetrahedra: The Dune ALUGrid class provides this. However, the two-point flux approximation (TPFA) method used in Flow is not consistent in general (only on K-orthogonal grids), so it is likely to be a bad discretization method for your tetrahedral grid! (If you were to make Voronoi-cell (or PEBI) grids on the other hand, it should work very well.)

Atgeirr
________________________________
From: Opm <opm-bounces at opm-project.org> on behalf of Antoine B Jacquey <ajacquey at mit.edu>
Sent: 16 November 2020 15:24
To: opm at opm-project.org <opm at opm-project.org>
Subject: Re: [Opm] Use external mesh with Flow OPM and unstructured meshes

Hi Atgeirr and the OPM community,

We tried to use the trick you mentioned using MRST to export the grid to OPM.
We encountered some issues related to the geometry. In MRST we have an unstructured grid (we tried both hexahedra and tetrahedra) but the VTK output from OPM just looks like a line.
The DX.txt file generated by MRST looks ok but the DY.txt and DZ.txt are just lists of 1s.

Here is the grid we built in MRST:
https://eur03.safelinks.protection.outlook.com/?url=https%3A%2F%2Fdrive.google.com%2Ffile%2Fd%2F1l0584yAr8eYLAdtkVPtiuZzJaxHs5pNB%2Fview%3Fusp%3Dsharing&data=04%7C01%7CAtgeirr.Rasmussen%40sintef.no%7Ccf6be10f53d44cbfd48308d88a3b6fc2%7Ce1f00f39604145b0b309e0210d8b32af%7C1%7C1%7C637411335182488365%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C1000&sdata=UkRCDgscFclkRs68F2ol7oUMJcYWBkYJhRc3GRUiOmw%3D&reserved=0 <https://eur03.safelinks.protection.outlook.com/?url=https%3A%2F%2Fdrive.google.com%2Ffile%2Fd%2F1l0584yAr8eYLAdtkVPtiuZzJaxHs5pNB%2Fview%3Fusp%3Dsharing&data=04%7C01%7CAtgeirr.Rasmussen%40sintef.no%7Ccf6be10f53d44cbfd48308d88a3b6fc2%7Ce1f00f39604145b0b309e0210d8b32af%7C1%7C1%7C637411335182498305%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C1000&sdata=w8QvsLDAayj%2BKz8pSyhnAyRZI%2Fr%2F4R%2BX4wcsDi%2BaebA%3D&reserved=0>

And that’s the output (VTK) we get from OPM (it’s just a line with broken connectivities):
https://eur03.safelinks.protection.outlook.com/?url=https%3A%2F%2Fdrive.google.com%2Ffile%2Fd%2F105WrVKZCswe7fJV_zzAqYZmaT4wWjQr9%2Fview%3Fusp%3Dsharing&data=04%7C01%7CAtgeirr.Rasmussen%40sintef.no%7Ccf6be10f53d44cbfd48308d88a3b6fc2%7Ce1f00f39604145b0b309e0210d8b32af%7C1%7C1%7C637411335182498305%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C1000&sdata=PZYivdpbihq9cLx0YhOMBJ5rr06vFeZs8jbevpXM0xg%3D&reserved=0 <https://eur03.safelinks.protection.outlook.com/?url=https%3A%2F%2Fdrive.google.com%2Ffile%2Fd%2F105WrVKZCswe7fJV_zzAqYZmaT4wWjQr9%2Fview%3Fusp%3Dsharing&data=04%7C01%7CAtgeirr.Rasmussen%40sintef.no%7Ccf6be10f53d44cbfd48308d88a3b6fc2%7Ce1f00f39604145b0b309e0210d8b32af%7C1%7C1%7C637411335182498305%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C1000&sdata=PZYivdpbihq9cLx0YhOMBJ5rr06vFeZs8jbevpXM0xg%3D&reserved=0>

We are not sure if we missed something in the process or if we did a mistake somewhere.

We would like to use a mesh built from tetrahedra (similar to the one built from MRST). Is this type of element supported by the OPM Flow solver? And do you think by using the trick of the MRST export we can rebuilt tetrahedra in OPM?
If we choose to code our own grid class or link to another grid class from Dune, do you know if tetrahedra are also supported in any of the Dune classes?

Thanks In advance for you comments and advice.
Antoine