open TELEMAC-MASCARET The mathematically superior suite of solvers

Hello,

Yes, exactly, ZERO given in the DATA; I realise that for some reason "epsilon" is only a French expression among engineers and scientists to mean something very small...

JMH

Ok, nevertheless thank you for trying the compilation with optimizations off!
This possible cause sometimes swirled in my mind.
Best regards
Clemens

Hello,

For the sake of completion I give my feedback regarding the problem I encountered. I have found 3 steps that seem to help tackle the error in layer and let me complete the simulation:

1. Reducing the morphological factor
2. Increasing the ZERO value from e-7 to e-5
3. Reducing the number of size classes

The problem is still there, but at least I got the simulation to the end.

Regards,
Costas

Hello,

After discovering that I can output the sediment fractions for each class in the results files of Sisyphe (didn't know that 'till now ), I did some further runs, now also checking the sediment fractions.

I have found some spurious fluctuations of sediment fractions in every part of my domain where I had defined an erodible layer thickness of 0 (e.g. rock outcrops). It certainly seems that erodible thickness=0 is causing trouble in LAYER when init_compo.f is used. I changed my erodible bed definition so as to allow a minimum erodible thickness and so far the computation runs smoothly.

So I guess my previous remarks are not valid. I will get back when I conclude my tests.

Regards,
Costas

Dear Costas,

did you solve your problem with this minimum erodable layer thickness or does it start again after the erosion of this minimum layer?
I also had this problem with erodable layer thickness of 0 and init_compo two years ago. I solved the problem using larger grain sizes in those areas, which will not be eroded. But there are some side effects - in some cases the erosion in the next erodable node tend to be to high....

Regards,
Gabi

Dear Gabi,

No, it did not eliminate the problem completely, but corrected the spurious results in sediment class fraction and helped the computation complete a lot more timesteps. Erodible layer thickness=0 is definitely causing trouble, that is for sure.

However, as my computation advances, eventually I do have areas in my domain that the erodible layer thickness is a round 0 and the computation continues normally. So maybe is it solved and the error lies somewhere else? I have yet to conclude my investigation, I will post any findings.

Regards,
Costas