open TELEMAC-MASCARET The mathematically superior suite of solvers

Hello everyone,

I carried out some test simulations using T2D, Gaia and a liquid boundary file.
I want to simulate the longterm bed evolution of a reservoir. In my test case there's one cohesive sediment fraction set as inflow concentrations.

The simulations run very smoothly at the beginning and I can see how the cohesive sediments are transported and settled according to theoretical expectations. As a delta forms, water depth decreases closes to the inflow boundaries. Some cells are not wetted anymore.

Then there is always a time step where the cumulative bed evolution at a node is assigned a very large value (e.g. it increases from 1.5 m to 250 m). As a result there's a very high free water surface, bed shear stresses and the solution diverges ( DEBIMP: PROBLEM ON BOUNDARY NUMBER 1)

Until now I was not able to solve this problem and wanted to know if you experienced sth. similar or if you can give me recommendations to solve this issue? There may be a problem with the erosion of the cohesive material (or mixed; due to the initial sand fraction).

I'm using a variable time step and set the Courant number to 0.9 but tested 0.5 as well. Concerning the bed model I tested BED MODEL=1 with one layer (see attached) as well as a multilayer model (=2) with consolidation.

I appreciate any advice on preventing this large sedimentation in a single node. Please find enclosed the steering files for one of the tests.

Kind regards,

Kilian

Hi Kilian,
There are some things that you can try:
* perhaps your CFL condition is still too high for the physical problem that you want to solve (delta formation is in the end a sediment wave propagation) => try a smaller CFL value, or a small (fixed) time step for testing.
* you can have some hints also by muting the erosion or by controlling the deposition rate of the cohesive fluxes at the interface water-bottom. Try:
- Partheniades = 0.0 (erosion flux = zero)
- Play around with the critical shear stress for deposition, from very high values => always deposition to smaller values => controlled deposition.
Hope it helps,
Best wishes,
Pablo

Hi Pablo,

thank you very much for your help and ideas!

I set up two more simulations with different critical shear stresses for mud deposition and was able to run the model for the whole 3 year period.
The keyword works very well and is useful to avoid sedimentation in shallow areas close to the inflow boundary.
You're also right concerning the time step, I have to find a good compromise between stability and calculation time. Otherwise that will lead to instabilities especially in the tidal flats.

I have not played yet with the Partheniades constant but I will definitely keep in mind.

I had a another test case with more realistic and lower longterm sediment input into the reservoir but fluctuating water levels (+-10 m). In this case, I experienced a stable model by changing the inflow velocity profiles from 5 to 1. A lowering of the critical shear stress for mud deposition was not necessary. That might be also of interest for some modellers.

Best regards,

Kilian