open TELEMAC-MASCARET The mathematically superior suite of solvers

Hello,

I have 2 questions about tracer diffusion in Telemac3D (v6p3 and v7p2):

1/ All TELEMAC-3D user manuals (up to v7p2) say that for keyword "SCHEME FOR DIFFUSION OF TRACERS"("SCHEMA POUR LA DIFFUSION DES TRACEURS") one value per tracer may be entered. However, according to the file 'declarations_telemac3d.f', SCHDTA is a single integer (while, in comparison, SCHCTA for advection is an integer array with size MAXTRA). Could you confirm that this is correct and, if so, update the user manual accordingly?

2/ I am considering several tracers, all of them with source terms hard-coded in 'source_trac.f'. Apparently, diffusion and source terms are being treated together in 'cvdf3d.f' and 'diff3d.f'. So, if diffusion is switched off (for all of some tracers), the source terms may not be taken into account, which is not what I want.
I have tried to set to zero the variable VISCTA%ADR(...) just before the advection-diffusion step (CVDF3D call) in 'telemac3d.f', but I am unsure this is totally harmless for the computation...
What would you recommend in order to switch on/off the tracers independently but keeping explicit/implicit source terms activated?

Thank you very much for your help.
JBS.

Hello

1- Yes this is right, there is only one option for diffusion of all tracers. This should be modified in the user manual
2-Source terms are handled similtanously with diffusion only when you use advection scheme = Characteristics or SUPG. In the remaining cases (distributive scheme or finite volume upwind) source terms are handled with advection part. That's why if you switch off diffusion (by using option 0) your source terms are well handled if you do not use characteristics or SUPG. Otherwise, your solution to prescribe visc=0 will do the job. I recommand to use distributive scheme and to switch off diffusion.

I hope that this helps
kind regards

Riadh ATA

Thank you Riadh for these clarifications.

Then I understand better my problem.
I always use distributive schemes (14) for tracer advection. But one of my tracers should be neither advected nor diffused. So I also had to switch off advection for this tracer, which now explains why the source terms have been disregarded.

Is there a "clean" way to take source terms into account when both advection AND diffusion are to be deactivated?

(activate advection for this tracer and set advection field to 0?...etc.)

Thank you again for your help.
JBS.

I have done this for several tracer generated by the new water quality module WAQTEL. For diffusion, I have done exactly what you have done by setting (see subroutine nametrace_waq.f in sources/waqtel)
CALL OS( 'X=0 ' , X=VISCT%ADR(ITRAC)%P)

for advection you can set ICONVFT(Itrac)= 0. But I'm not sure if it will work efficietly. There are other solutions you can test later, but try these ones and let's cross fingers :)

kind regards

Riadh

Thank you Riadh.

It seems subroutine 'nametrac_waq.f' is called by TELEMAC2D routines only, not TELEMAC3D... (I have never used WAQTEL so far, I don't know how it works)

In my particular case in TELEMAC3D, when:

- DIFFUSION is ON for all tracers (SCHEMA POUR LA DIFFUSION DES TRACEURS = 1),
- manually setting viscosity to 0 (as you shown) before tracer advection-diffusion step in 'telemac3d.f' (in fortran file) for all tracers, and
- ADVECTION is OFF only for that 'static' tracer #2 (SCHEMA POUR LA CONVECTION DES TRACEURS = 14;0),

implicit and/or explicit source terms are taken into account somehow (I mean, exchanges between tracers are definitely occurring). The only question is: are we sure that both explicit and implicit source terms are correctly taken into account in the calculation?...

(I recall I am still working with version v6p3 - but not sure this is very different with v7pX)

Thanks again for your help,
JBS.

Hello

nametrac_waq is called also by telemac-3d for last release v7p2.
Waqtel is simply used by coupling it to telemac2d or telemac3d as we use to do with sisiphe or tomawac. However, Waqtel is not designed to work in a standalone mode. its job is limited to the computation of source terms for WAQ processes.

Implicit and explict source terms are treated mainly when we build matrices (for implicit) and right handside term (explicit) which will be given to subroutine solve to apply the resolution of the linear system.

So far, the resolution of transport equation is achieved separately for each tracer. Therefore, there are no implicit coupled source terms between tracers.

I hope that this helps

with my best regards

Riadh

Ok for WAQTEL.

About my present configuration, should I conclude that my current way of doing (exposed in previous post) is ok?...

JBS.

For diffusion it is OK
For advection I'm not sure that this holds. with option 0, Telemac will keep values of tracer unchanged from time tn to time tn+deltaT. Thus sources contributions will not be considered.

Kind regards
Riadh

Ok.

I was just looking at source files 'cvdf3d.f' (and 'diff3d.f') which is called for this operation.
If advection is off (SCHDF=0) this is true, there is no change (FC=FN).
As diffusion is on (SCHDF=1), it seems diff3d is necessarily called with S0F and S1F as inputs (YAS0F and YAS1F being true). Then, in 'diff3d.f', I see that both explicit and implicit source terms are taken into account, which is consistent with my numerical observations.

Therefore, I might be wrong (I have not investigated more in depth) but it seems source terms are being included in the calculation when diffusion is activated (SCHDF=1) while advection is deactivated (SCHCF=0).

Is that correct or is it more complicated?

Thank you,
JBS.

Hello,

It is a bit more complicated. The treatment of source terms in diffusion depends on the advection scheme. Some advection schemes (distributive schemes 4, 5 and NERD : 13) will treat the source terms, while others (characteristics) will not.

With best regards,

Jean-Michel Hervouet