open TELEMAC-MASCARET The mathematically superior suite of solvers

Hi,

For validation purposes I´m running a simple channel where a constant head drop is forced between the ends of the channel. As no friction is considered in walls, I aim to achieve a constant velocity profile across the channel. However, if you take a look at the png file (see attached files), you will see that in the north part of the western boundary, velocities are slightly different than the rest.

I´ve refined the mesh, included friction in the region near the boundary and so far, I haven´t managed to get a constant cross section velocity profile.

Any help will be much appreciated.

Thanks,

Hello,

I see this green color but what you do is not totally clear. At the entrance the profile is given, so I suppose that the left end is in fact the exit ? Then depending on what you do inside the domain (head loss ?) or due to the mesh you may have slight differences at the exit ? Or even an eddy that gets out and causes velocities to get in through the entrance ? For example if you refine the mesh too much in an area, it may cause the local CFL number to raise problems or instabilities. So I would rather look at the underlying mesh, that has perhaps abnormal small elements locally.

Regards,

Jean-Michel Hervouet

Thanks Jean-Michel,

I thought somewhere that creating a sponge was a feasible alternative to avoid discontinuous velocities at boundaries, but indeed it might be a mesh problem.

Thanks,

Alberto