open TELEMAC-MASCARET The mathematically superior suite of solvers

Hi,

I am having some trouble getting nice results from a sysiphe run coupled with telemac. A tidal varying flow is imposed on an island which is partly protected from erosion by editing the noerod.f file. The simulation gives nice results for the majority of the domain but there are some instabilities right next to the non-erodable bed (see figure).



We have tried almost every possible option to lower the wiggles but nothing really helps:

- Advection schemes: 1,13 and 14
- Option for non-erodable beds : 0,3 and 4 with finite volumes
- Coupling period: 1,5,10
- Teta: 0.55,1
- Beta: 1.3,1.6
- Friction angle: 40,35,25,15
- Time step of telemac : 1s,3s

In the attachment you can find the used .cas files. We really want to get those wiggles solved because that zone is the one we are most interested in.

Thank you for your response

Kind regards

Tim

Hi Tim,
thanks for your post.
What about your mesh density near those zones?. Sometimes you would need fine elements over these transition areas.

I hope it helps.

Pablo

Thanks for the quick reply!

The mesh has a edge length of 10 m in that zone and the profile posted above has a width of 400 meters, but I can try to make it even more dense. The problem is though that the timestep will have to decrease again and we would like to keep the model fast enough.

Kind regards

Tim

I have tried to run the simulation with a mesh size half of the original one. This is the result:





As you can see the instabilities remain. Every peak is a cell wide.. I am running out of possible options to get rid of these wiggles

Hopefully someone can help me :)

Grts

Tim

Hello,

Please note that with the finite element option, the options for treating the non-erodable bed are not active, because the method naturally ensures that the bottom is not eroded. With this option the Exner equation is treated as an advection of the sediment layer, and this advection is mass-conservative and would be monotonous with a divergence free advection field, which is not the case. Though the peaks are cell wide, which is certainly a numerical effect, I am not sure that we can find a correct scour if we do not have an effective sliding effect due to the slope. Depending on your local mesh size, the physical maximum slope is perhaps not reached. Setting an artificial maximum slope would perhaps solve the matter (and would possibly act only in this zone if it has the largest slope). Otherwise we would need all the files to reproduce the case and try different things: use a more diffusive advection scheme (the contrary of what we are generally looking for...) or smoothing the bedload itself, like with the option for smoothing the bottom.

With best regards,

Jean-Michel Hervouet

Thank you for your reply mr. Hervouet!

We have tried using the finite volume method and changing the treatment of non erodable beds to 4 but we get very unrealistic results:





The 2 graphs have the same morphological simulation time, the left one is with FINITE VOLUMES = YES. That and the treatment of non erodable beds is the only difference between the 2 figures.

Also we tried adding much more diffusion by taking a BETA = 10 for the slope effect formula 1. This only resulted in more erosion/deposition, but the pattern did not change. Now I am running a simulation with a friction angle set at 10° to try smoothing the bed. Where can I find the option for smoothing the bottom? Is that mass lumping, because that is on.

Thank you very much for your help

Kind regards

Tim

Hello,

The option for smoothing the bottom is just an option for starting a computation with Telemac-2D (see the call filter in corfon.f, library telemac-2D), it is not implemented in Sisyphe for smoothing the bottom (which would be a very rough solution). I just suggest that smoothing in that way the bedload transport (Q in the Exner equation) would bring an improvement, but it is not implemented either. So for now we just have the option SEDIMENT SLIDE with a given FRICTION ANGLE OF THE SEDIMENT, and obviously the option with finite elements is better.

Regards,

Jean-Michel Hervouet

Hi

I did a run with a friction angle of 10°. If I take a lower angle I am going to lose my sanddunes so I cannot go lower:





You can see the wiggles are a bit smaller and indeed their slope does not exceed 10°.

Hello,

Yes, it is better, except near the island but this is normal, there is no sand to slip and fill the gap, so the slope remains, and this is the scour.

JMH