open TELEMAC-MASCARET The mathematically superior suite of solvers

Hi,

I am trying to model the bed load transport throught a weir in river. I need to use both mixed sediment and no erodable bed.

I managed to build my model and to run it for different configuration but it often stop during the computation writing :

" PROBLEM IN LAYER: J,TEST1 63064 8.9307699704120819E-004
IN LAYER 1 SUM OF FRACTIONS NOT 1
ARRET APRES ERREUR DANS LAYER

PLANTE : ARRET DU PROGRAMME APRES ERREUR"

Or

" ERROR ON FRACTIONS
LAYER 1 CLASS 1 POINT 64416
AVAIL= -0.55248695264829550
ZFCL= -6.1840489063233889E-002
EVOL= -6.1840211728283169E-002 ELAY= 4.7381256978155761E-002
ARRET APRES ERREUR DANS LAYER

PLANTE : ARRET DU PROGRAMME APRES ERREUR"

This appears only on the points which are defined as non erodable and should be submit to quite important erosion if not.
I noticed that if I change a bit the highness of the non erodable point, the problem may disappear.

That can happen even after hours of computing and I need each time to recompute from the beginning (after changing highness of non erodable bed).

May be this problem is already known ?

Thanks

Quentin

Hello

I notice that you selected option 3 for the treatment of rigid bed.
This option is obsolete and please try option 1 and let me know.

Best luck, Cath

Thanks for your reply !

I tried today with the 'option for the treatment of non erodable bed' fixed to 1, but the error message still appear after 3h of computation (I mean it is not a problem of beginning of computation).

I join the folder containing all the computation files, maybe it will be easier to understand where the problem comes from.

Thanks in advance

Quentin

The folder is to big, here is a link to download it :

www.yousendit.com/download/WUJaUXVvWlRVVG15VmNUQw

I'm wondering if you had any luck with this problem?

I am getting similar errors - e.g.

PROBLEM IN LAYER: J,TEST1 2532 1.00057642418440
IN LAYER 1 SUM OF FRACTIONS NOT 1
STOP AFTER AN ERROR IN LAYER


Modelling bedload over an erodable bed, but of varying depth down to non-erodable bed. Errors seem to appear after some time when it has worked its way down to the non-erodable elevation.

cheers,
Dougal

Hello,

The problem of keeping fractions between 0 and 1 is very tricky because of truncation errors, so sometimes a fraction slightly over 1 is not a bug. If you take subroutine layer.f in your fortran file, you can change the tolerance ZERO which is 1.D-7, to 1.D-3 and see what happens.

Regards,

JMH

Thanks for the suggestion.

It turns out I had (accidentally copied from old example file)
OPTION FOR THE TREATMENT OF TIDAL FLATS = 2
which was causing all the problems.

Regards,
Dougal

Hello,

I'm now running a model with a rigid bed (but suspended and bedload inputs at inflow boundaries).

Seems to run fine initially, but with a 'computation continued' will inevitably crash with a 'error in layer'.

At the first timestep after restart, at multiple nodes where there is essentially zero sediment depth (i.e. around 1-2.E-005), I get an 'Error on fractions' where the fraction is outside the range 0-1.

If I try changing tolerances, e.g. the lineor the value forit will crash elsewhere within a few timesteps.

If I check for negative layer depths I do in places get very slightly negative depths in layer2 (e.g. -1.1E009, precision errors?).

Any suggestions for workarounds? Perhaps somehow ignore nodes where sediment depth smaller than a min value?

Regards,
Dougal