open TELEMAC-MASCARET The mathematically superior suite of solvers

Hello everyone,

I am trying to model a weir inside a channel in an agricultural field. My model outline is rectangular, with the right side as the inflow boundary condition (using a liquid boundaries text file) and the left side as the outflow boundary condition (free outflow boundary condition 444). Inside my domain there is a rectangular channel and inside the channel there is a weir. I have created the weir as a rectangular hole using the hardlines option in BlueKenue. The weir example case in the verification manual is not suitable for me since I cannot cut the mesh into two parts. This is because I also want to see the flow in the floodplains (the banks of the channel). Also in the future I want to implement nodal rainfall as input water so I need a coherent mesh. Hence I model the weir as a hole in the mesh.

I am using Telemac2D V8P4

The problem I face when running the model with the weir formula is:

1. The simulation takes much more time to complete than a simple case without weir (22 hours vs 2 hours on 3 cores)
2. There is some 1700 meters of water depth just upstream of the weir, which definitely means something is going wrong. It seems Telemac is not transferring the water form the upstream to the downstream of the weir.

Since I am not very experienced with Telemac, maybe I am doing something wrong? Can you please check my input files or maybe run the same model and let me know what is going wrong. I will be grateful for any advice.

Please find attached:
1. Geometry slf (model10_geom)
2. A warm up slf: This is used to specify an intial water depth for nodes at the inflow boundary condition (model10_warm)
3. Results slf (model10_results)
4. weir definition text file (model10_weirs)
5. liquid boundaries text file (liquid_boundaries_lowQ)
6. boundary conditions file (model10_BC)

Here is the image of my domain https://drive.google.com/file/d/1LAfDVIQ1RYzGJc4ewLadjCgXkDK--4v2/view?usp=sharing

Here is the zoom in on the weir:
https://drive.google.com/file/d/1Z3edPVBzxonxnVPjCcPFCETqdKwW1oVa/view?usp=sharing

I just checked that the results files was not uploaded in my previous post because of the file size. So here it is in a google drive link: drive.google.com/file/d/135lqG5WU6fluBnQ...yir/view?usp=sharing

Hi,

I think the mistake lies within the definition of your weir crest heights. I think you gave the elevation above the bottom, while it should be absolute elevation (m.a.s.l.).

Kind regards
Frederik

Hi Frederik,

Thank you for your answer and for checking my files. I have changed the weir crest height to absolute elevation rather than elevation above the bottom, as you mentioned. However, my model is still not converging. I have a time step of 0.001 seconds. But I still get the error :
"GRACJG (BIEF) : EXCEEDING MAXIMUM ITERATIONS: 1000 RELATIVE PRECISION: NaN"

I can't seem to figure out what might be the problem? Is it how I am still defining the weir in my mesh? Is it the numerical parameters in the cas file (I use the NERD advection scheme for Tidal flats)? Is it something to do with the free outflow boundary condition at the outlet? (number 444)

Any suggestions will be appreciated.

With kind regards,
Karan

Use a debug build (see hint in www.opentelemac.org/index.php/kunena/16-...nity-value-nan#45444)
It will indicate you the faulty line generating a NaN value