open TELEMAC-MASCARET The mathematically superior suite of solvers

Hallo,


We are modelling reservoir flushing with spatial distributed graded sediment. We use TELEMAC-3D internally coupled with SISYPHE.

In the simulation furrows occur in the evolution (see figure). We also noticed “streamlines” with higher bed shear stresses. But we did not know where this furrows and streamlines come from.




We did a lot of parameter tests in SISYPHE (skin friction, slope effect, deviation formula, sediment transport formula) and also changed the turbulence model, but we still have this furrows. We also do not know what happens first. Are there first the furrows and the bed shear stress results of the different water depth? Or is there a problem with the bed shear stress and the furrows come later…

Happy or any hint,
Gabi

Hello Gabi,

these furrows are a phenomenon that we notice a lot in our projects. The problem occurs if you have a mesh that is parallel to your flow direction. If I have to guess it'd say that your furrows follow exactly they main stream velocities and the river main channel.
Even though your mesh is not completely structured it has a strong tendency. So I am pretty sure if you make your mesh unstructured in this area you will get ride of the furrows.

For an explanation for this phenomenon we are not quite sure, but a guess is that when your mesh matches the main transport direction the surrounding nodes are not considered as the direction of the velocity (and with that the transport) goes exactly to the next node in the mesh.


best regards
Annalena

Hello Gabi,
to complete Annalena's hint, I'd like to add some comments:
1) probably your result is an admisible solution of a physical phenomena that you are capturing with the model. Do you have any mesurements/observations to verify that?
2) have you tried with another sediment transport formula?
3) have you launch a simulation without coupling the sediment transport/bed evolution model? how does the hydrodynamics results look like?
4) try Annalena hint, with an absolutly unstructured mesh

best,

Pablo

Hallo Annalena and Pablo,

thank you for the hints.

To Pablo's suggestions:

1) Yes, we have measurements. There is more or less a flat bed, bedforms occur in case of flood (antidunes)
2) yes, we have tried MPM, Engelund, Van Rijn -> no big difference
3) + 4) we will try

Many thanks,
Gabriele

Hello Annalena, hello Pablo,

to discuss this topic further since I think it is important:

Positive is that we (Gabi and me) are not the only ones who have big problems with this phenomenon
Negative issues:
1. how to combine the 2 requirements to capture essential features of a river like thalweg, bank edges or other features in the streamwise direction in the mesh and to generate an "absolutely" unstructured mesh?
2. the results are highly mesh dependent or are locally mesh dependent since locally some structured mesh features very likely appear even when trying to generate a completely unstructured mesh. And how to evaluate the mesh quality, the unstructuredness, for SISYPHE?


Best regards,
Clemens

Hi Clemens,
just some comments to your post (I'm back to work on Monday) :

1) to the best of my knowledge, some meshing tools (like Janet) can
be used to construct meshes based on some criteria, e.g. gradient of the topography, etc. And as far as I know, BlueKenue is well adapted to create meshes that capture some of the characteristics of the reach.

2) your second point is really interesting: some similar criteria, as used for mesh adaptation, could be proposed to evaluate the error estimation in a sediment transport simulation. I'll discuss with some colleagues and probably this could be a nice topic to have a student working on.

And, of course, any suggestions or ideas are welcome

all the best,

Pablo

Hello Pablo,
thank you for the answer.
I want to insist a little bit on the topic:
BlueKenue is great in generating meshes incorporating breaklines such as soft lines to capture important features of the river. So when incorporating some soft lines in the streamwise direction in a certain area the mesh generator by default will produce an unstructured mesh but with a structured alignment in this area.
Quoting Annalena:"So I am pretty sure if you make your mesh unstructured in this area you will get ride of the furrows." I want to account for river features but with the side effect that I have some structured alignments. So it seems to me that for SISYPHE simulations "the cat eats its tail".
So maybe also for SISYPHE simulations which are based on meshes produced without any constraints, it is hard to judge which part of the result is attributable only to the structured mesh alignment.

Best regards,
Clemens

Hello,

I am currently generarting a mesh in BlueKenue based on a trapezoidal channel. My concern is that I may face this exact problem, as the geometry is very strucutred itself. My questions: Are there new findings regarding this topic? Or can someone give tips on how to generate an unstrucutred mesh without alingment on nodes in main direction of flow?

Thanks in advance and best regards,
Tobi

Hi Tobi,

Yes this issue will still occur on aligned meshes.