open TELEMAC-MASCARET The mathematically superior suite of solvers

Hello,

I'm running a telemac2d simulation with a varying inflow at a boundary (18 nodes). Peak value is 1353 m³/s. At a control section about 900m downstream the results give peak values of 1355 m³/s. There is no tube entering or exiting. Maybe its just an inaccuracy I've to live with, but I'm not sure. Maybe there is a keyword (-set) I could add. I would be grateful for any tipps.

Cheers
Engelbert

Hello,

This is "normal" as the computation of fluxes across cross sections is not exactly equivalent to what is done in the program (basically due to the difference between weak and strong formulations in finite elements). Having exactly the same fluxes is a development to come in the next version, done by Leo at BAW.

With best regards,

Jean-Michel Hervouet

Hi,

as Jean-Michel said, the cross section routine gives not the exact discharge. The new method that Jean-Michel mentioned will work for all schemes that use the "flodel" array. I've allready put my work onto my brunch but I think there will be some minor changes since my Fortran style might not be the best.

At the beginning I was also a little bit disappointed that the flux of the section routine is not 100% the correct. But this was more or less since it is a nice way to check the conservation of mass of numerical schemes with such a cross section (academic world). Being a little bit more realistic, we see that the error is very small.

I think that the error is more relevant for bedload simulations in Sysiphe. Here we have significant differences since the cross section subroutine gives only the potential discharge over all classes.

With best regards,

Leo

Hello Leo and Jean-Michel,

thank you both for your answer. Of course I know that I'm talking about a very small error, but since I'm not deep into the numerics of telemac I was wondering what the reason for the difference is. Anyway, your answers helped me understanding the problem. Thank you again.

Greets
Engelbert

Hi Engelbert,

it is totally correct that you used this method to check your model. I think everyone that does numerical modeling should try to check his model as good as possible. It is very important to ask for such small errors since it improves the understanding.


My experience is that you only get the 100% correct fluxes in numerical models if you compute the discharge within the numerical model since you need the same data and algorithm. Many users do not need such detailed results and compute discharges after their simulations based on the data of the selafin file. For some engineering questions this is sufficient.


Best regards

Leo