open TELEMAC-MASCARET The mathematically superior suite of solvers

I made a transport simulation for a rectangular channel where the channel and the initial conditions are symmetric, however I have got non-symmetric results. is that possible ? Attached is the geometry file, the boundary conditions file, the results file, fortran file used for initial conditions and the steering file

the geometry and boundary conditions files are attached in this compressed file

Hello,

Naturally you can find numerious cases where symmetric case (geometry, boundary conditions, initial contions) gives unsymmetric results. You can see, for instance, the non laminar flow around a cylinder or more generally flows with observable turbulent structure.

Actually for your model, you are not in a turbulent case. However I'm wondering if you omitted to impose a discharge in your liquid boundary number 1. I can see in your steering file that there is no prescribed flowrates and consequently there is alomost no velocity (around 10-5) all over the domain. So the tracer is only diffusing due to the numerical diffusion of the code (since you disabled the disffusion option).

All my best regards

Riadh

The transport of the tracer is not only due to the numerical diffusion of the code but also due to the spatial variation of density which is the main idea behind this simulation case (as you said the water is stagnant).

However, I am still wondering why there is non-symmetry results. Do you know how much could be the numerical diffusion

In this case :
1- you don't need any liquid boundary conditions; change all boundaries to solid wall. The presence of a liquid bounday with (non) prescribed flowrate creates some trouble in the vicinity of the boundary which propagates thoughout the domain.
2- in your case, you have a constant water depth in all the domain. You have also a constant tracer with a discontinuity in the middle. For all the area where the tracer is constant, the situation is not stable (if we are solving the conservative form of the equation)
3- if we resolve the non concervative form (which is preferred in Telemac), the variation of the tracer is due to 3 main reasons:
1- the existing of source or sink in the domain (which is not the case here)
2- the advection due to the presence of a velocity field (which is not the case here)
3- the diffusion of the tracer due to the combined effect of the viscosity (or the dispersion) and the gradient of the tracer.

So in your case the only reason of variation of the tracer concentration is due to the third reason (3-3-). However, you have indicated in your steering file that COEFFICIENT FOR DIFFUSION OF TRACERS =0. Hence, you killed the natural diffusion due to fluid viscosity and the only diffusion that remains is the one caused by numerical viscosity and dispersion.
therefore, you can quantify this numerical diffusion by comparing your results with the analytic solution of the diffusion equation dT/dt +mu*Laplacian(T)= 0

I hope that this helps.
Kind regards

Riadh