open TELEMAC-MASCARET The mathematically superior suite of solvers

Hello,

in a 2-D hydrodynamic simulation I encountered water surface striae at steady state in various locations of my model, e.g. see attached image.





With reference to the forum post - www.opentelemac.org/index.php/kunena/17-...n-the-evolution#5140 - it seems to be the same phenomenon and indeed the striae appear where the unstructured grid has some structured alignment. The water level differences are low compared to the water depth but for longer sediment transport simulations they play a large role. And it is not nice to show plots of water surface elevations with those striae.

I used the following numerical parameters:

---

TYPE OF ADVECTION =1;5;1;1
TREATMENT OF NEGATIVE DEPTHS =2
MASS-LUMPING ON H =1.0 / for treatment of neg. depths = 2
CONTINUITY CORRECTION =true
NUMBER OF TIME STEPS =720000
FREE SURFACE GRADIENT COMPATIBILITY =0.9
TREATMENT OF THE LINEAR SYSTEM =2
SUPG OPTION =0;0;2;2
TIME STEP =0.5

---

I would be very glad for any hints how to avoid these striae and also for explanations why they appear.

Best regards,
Clemens

Hello,

You should try to see if a lower value of the key-word:

FREE SURFACE GRADIENT COMPATIBILITY

has an effect (you can go down to 0.), in which case you see the effect of the inf-sup condition that says that if depth and velocity are discretised in the same way wiggles may appear.

Regards,

Jean-Michel Hervouet

Dear Jean-Michel,

thank you very much for the prompt answer!
Lowering the value to 0.5 helped a little bit and by lowering the value to 0.1 the striae almost disappear and the water surface elevation in my upstream zone of interest is almost the same. So this works very fine!

The mesh has 300000 elements and the physical duration is 100 hours, so the computation time even on the cluster is quite expensive.
So, the general question: how to define a priori the value for the free surface gradient compatibility? What are the drawbacks when lowering this value and if there are any, how to assess them? Otherwise I would set a low value for all my projects and save computation time.

Best regards,
Clemens

Dear Clemens

This keyword gives the compatibility bethween water depths and velocities in the continuity equation. That means that for a value of 1, there is a full comatibility and hence the equation is fully verified. When we introduce a value lower than 1, we relax the relationship between velocity and water depth (continuity not strictly verified) in order to prevent spurious oscillations (like the striae you've observed in your model). These oscillations are mainly observed in regions with steep bathymetry gradients, but not only, and they could have their origin from the inf-sup condition (known also as LBB condition) in cases where we have the same discretisation in space for velocity and water depth.
Consequently, an other way to prevent the inf-sup oscillations, is the use of different space discretization (DISCRETIZATIONS IN SPACE = 13,11). However, this solution is more expensive and could be prohibitive for long time simulations (like yours)
Jean Michel, please not hesitate to complement this very succint reply.

With my best regards,

Riadh ATA