open TELEMAC-MASCARET The mathematically superior suite of solvers

Hello,

I've already seen that Strickler coefficients do not have the same signification between 1d and 2d models.
Topic 7938
Topic 11634
Topic 7010

Anyway, my idea was to compare Mascaret and Telemac in different situation starting from uniform flow in a rectangular channel.

Inflow = 29m³/s
Slope is 0.0002 m/m
Width of the channel is 10m
Length of the channel is 50km (but it could be shorter)
and Strickler coefficient (according to 1d hypothesis) is 30m^1/3/s

-> Normal depth should be 4.01m

When I simulate it with Mascaret, it works. I tested three different downstream boundary conditions (Normal depth, Normal depth + 0.5m, and normal depth - 0.5m), and I get the same depth (= normal depth = 4.01m) when I check the depth far enougth from the downstream boundary.

But, the issue is that I can't reproduce it with Telemac2d. I need to reduce strickler coefficient to 17 !!! (that means d50 = 12m !?!)

I need more details to understand the difference between 1d and 2d strickler coefficient in that case.

Any help understanding these concepts would be appreciated.

Alexis

Hello,

two opposite things to distinguish:

1.
In your case, prismatic channel, you have a very low channel width / water depth ratio which is 10/4. The 1D model uses the hydraulic radius = Area / perimeter. In 2D depth-averaged flow modelling such a concept doesn't exist. In 2D the hydraulic radius is basically equal the water depth, which you can assume for channel width / water depth ratios > 5 or 10. With such ratios, you have really two-dimensional flow (x-z firection) with no influence from the side walls.
Hence, the only way to get closer to the 1D model results would be the use of side wall friction by means of a friction law for the lateral boundaries. However, I don't think that you will get exactly to the 1D results, the water level will be lower, still. Besides increasing the roughness of the bottom you could also increase the turbulent viscosity, which however is really not a good approach in especially prismatic channels.
So, if your channel width / water depth ratio is big, you should get almost the same results compared to the 1D model. Maybe the thing turns around and you get even higher water level in 2D since you "see" the numerical diffusion.

2.
In rivers with natural geometries, contractions, expansions, etc. theoretically in 2D one should apply lower roughness for the bottom compared to a 1D appraoch since in 2D you are, more or less, able to calculate head losses due to such geometrical features. In 1D one has to parametrize / summarize these head losses by means of an increased roughness.


Hope this helps.

Clemens