TELEMAC-MASCARET Forum: Model large slopes - instabilities and other error msgs (1/1)

Welcome, Guest

TOPIC: Model large slopes - instabilities and other error msgs

Model large slopes - instabilities and other error msgs 4 years 7 months ago #35657

shmulik90

Hi!
I'm new to the mascaret model but have a little experience with other shallowwater-solvers.

I am trying to model a 750m channel with trapezoidal crosssections. I wanted to study the influence of different slopes on the ingoing hydrograph signal (flattening aka. convection).

So boundary condition upstream: Q(t) Hydrograph
boundary condition downstream: free outflow (what does that mean in numerical terms?)

for 1%,2% and 4% the scheme works (super-critical solver == FV scheme?), but for 8% i get some strange results.

questions:

1) It seems that the "mesh" is influencing the solution, i find that there is a minimum size and it changes with slope - why?

2) What are the correct parameters for high slopes? (general and numerical)

3) It seems that the scheme finds no solution for "dry" initial conditions - why? I thought FV-schemes are made to handle dry- wet fronts?

Thanks for any comment!

Ps: model with 8% slope attached;)

The administrator has disabled public write access.

Model large slopes - instabilities and other error msgs 4 years 7 months ago #35659

c.coulet OFFLINE Moderator Posts: 3722 Thank you received: 1031	Hi No model joined (read the forum rules to know how attach files) Super critical solver is a FV scheme (Roe), this is in the documentation. As there is CFL rules, the mesh could have an influence... About the dry initial solution, this has been developped for dam break but it works usually well. Maybe it needs your upstream condition to be Q and Z values... Regards
	Christophe The administrator has disabled public write access. The following user(s) said Thank You: shmulik90

Model large slopes - instabilities and other error msgs 4 years 7 months ago #35661

shmulik90

File Attachment:

File Name: catfl_model_s8.zip
File Size: 350 KB

1) Model hopefully now as zip attached. Sry for that. (As you can see water flows initially in "wrong" direction)

2) CFL criteria has to be met as far as i know for explicit schemes (CFL <=1) and for implicit it can be much larger than 1- right? But as i checked the box "adaptive time-stepping and mac courant 0.8" i thought this will be taken care of automatically by the code? So my choice of dx == cross-sections per length should not make the algorithm fail but only have an influence on the computation time?? Why does it still fail? Is the initial time-step the missing piece?

The administrator has disabled public write access.