open TELEMAC-MASCARET The mathematically superior suite of solvers

Hello,

Regarding the Nikuradse friction law you're right. It appears that depending on the value of the Reynolds number the first cell can be around y+=100 or more. On this topic I recommend the paper by Guo in Journal of Hydraulic Research (2017): Eddy viscosity and complete log-law for turbulent
pipe flow at high Reynolds numbers.
However, regarding the Chezy model, even though u* is calculated based on the bulk velocity there is an assumption that the velocity profile follows a logarithmic profile (the head loss coefficient is calculated through the Colebrook formula). So I am not too sure what should be done regarding the first cell... if you place it too high you will miss the detail of the strong velocity variation close to the bed and the Colebrook formula won't really be valid in the numerical model.

Best regards
Agnès

Hello Agnès,
Thank you for your reply.
I will have a look at the paper and investigate in more detail the sensitivity of the model to the number of horizontal planes.
With best regards,
Jérôme

Hello,
I am working on the same kind of cases and I investigated the code. I think I found something that might explain this dependency to the number of planes (I observed the same small dependency watching vertical velocity profiles).

With K-Epsilon model, the logaritmic law enables to compute frition velocity taking the reference velocity U at the top of the first cell. With that velocity, boundary conditions are applied to the bottom with those formulas:
K =U*^2/sqrt(Cmu)
Epsilon=U*^3/(Kappa delta)

And when we explore the code we note that delta=dz/FICT with FICT=2 by default (that can be change in the .cas file). Epsilon is then dependent to the size of the mesh dz and it changes a lot on both velocity and turbulence (here are results with FICT=1 and FICT=2 as the only difference).

But maybe I made mistakes while exploring the code..
Best regards,

Vincent

Hello Vincent,
Thank you for your message.
Your affirmation on the treatment of the friction is right.
I understand the model dependency to the vertical mesh and we had interesting discussion on the choice of the law of the wall with Anges and Clemens (in the present topic).
However, my initial problem concerns the model dependency to the horizontal mesh.
When I use cell size smaller than 1m, the level of turbulence decreases anormally (in the zone with the fine mesh).
After several unsuccessful tests (in removing the mesh dependency), I decide to use 1m cell size because this value is very close to the mesh convergence.
Thank you.
Best regards.
Jérôme