open TELEMAC-MASCARET The mathematically superior suite of solvers

I had a model that ran fine but under certain circumstances the advection of tracers was too low. The problem arose when the total difference in tracer (i.e. max minus min) was about less than 5%. This did not make any sense to me so I changed the model to be convservative with respect to tracers. The model now exceeds maximum iterations and does not run. I've attached the case file and noted the changes below. Does anyone know what could be wrong with the new changes? Thanks!!

New changes:
TREATMENT OF NEGATIVE DEPTHS =2
MASS-LUMPING ON TRACERS =1
TYPE OF ADVECTION =1;5;14;1 (switched to #14)

Hello

Why this does not make any sense? the advection of the tracer is triggered by a large hydrodynamic current combined with a high gradient of tracer. In your case, we do not have any idea about the velocity field, however, the gradient of the tracer is small, which can explain your observations.

Concerning your issue, this could only be a problem of large time step. Try to decrease this latter.

It is recommanded to use OPTION FOR THE TREATMENT OF TIDAL FLATS =1. Moreover, mass-lumping of tracer=1 will increase the numerical diffusion.

I hope that this helps

with my best regards

Riadh

Thank you Riadh.

Reducing the timestep solved the problem. The tracers are now behaving properly.

I had to reduce the timestep from 0.25s to 0.03s, which is a major reduction! Are there any suggestions to increasing the timestep? I assume the timestep restriction is because I changed the advection solver to '14'

Hello

Actually the time step for each scheme depends mainly on the mesh size and to transient effects. I guess you have a very refined mesh. Thus if you switch from scheme 1 (Characterics, unconditionally stable) to scheme 14 (explicit conditionnally stable), you will be obliged to decrease significantly the time step.
you can try these hints:
- achieve a sensitivity study on time step (increase it until divergence)
- if your study is for a long time, with a highly transient effects at the beginning anf then a more smooth flow, you can increase the time step during the computation (by editing subroutine telemac2d.f and just include IF(AT.GT.10000.d0)DT=1.d0)

I hope that this helps

with my best regards

Riadh ATA

I learned the following things about maintaining conservation with respect to tracers.

1. To get the correct fluxes into the model, I must set TREATMENT OF FLUXES AT THE BOUNDARIES =2;2;2

2. Timestep was primarily dependent on COEFFICIENT FOR DIFFUSION OF TRACERS

3. The following parameters correctly advected the tracer without excessive numeircal diffusion:
TYPE OF ADVECTION =1;5;14;1
TREATMENT OF NEGATIVE DEPTHS =2
OPTION FOR THE TREATMENT OF TIDAL FLATS =2