open TELEMAC-MASCARET The mathematically superior suite of solvers

Hello,

I am trying to non-dimensionalized some simulation in telemac2d.
I am first working on a simple flat rectangular channel with an incident wave, with no bottom friction and advection (linearized propagation).
I study the wave before it reach the end of the channel so I don't have to put an open boundary.
After some simplifications of the Telemac equations it seems that I only have to change the gravity term in my simulation (you can see my reasoning in the attached pdf if you have time) to scale in nondimension.



So my g become 1 and I divide all the terms (mesh step ,lenght of the channel) by 10 (except the width=1) to simulate without dimension.
Then I should have the same result for a channel 2000long with a wave h(t,x_0)=10+1sin(wt) (w=2pi),g=10 (dim case)
and one 200m long with a wave h(t,x_0)=1+0.1sin(wt) (w=2pi),g=1(non dim case) if I scaled by 10.

However I always have a result less smooth, even with a change of the accuracy of the solver.
I have tried to add the FREE SURFACE GRADIENT COMPATIBILITY = 0.9 term.

If someone have an idea why it's getting this result.
You can see my graph attached,I take the wave for t=50s way before they reach the boundary.
The steering file is also attached, it's the one for nondim case.The dim steering file is the same but I multiplies all the hight by 10 and the solver accuracy by e-1.
I precise I am still a debutant in Telemac.
Thank you

Best, Alain.

Hello again,

It seems I have fixe the problem by changing the time step of 0,0025 to 0,05 for both of my simulations (nondim and dim).After scaling I have almost exactly the same solution.
I don't understand why, because I think normally a reduction of the time step should reduce the noise and get a smoother curve, so if someone can explain me why I get smoother result with a more important time scale.
Thanks,

Alain