open TELEMAC-MASCARET The mathematically superior suite of solvers

Hi Tom,

Thanks a lot for the help!
Feel free to let me know if you need more information from my side. I will also keep you updated if I could find anything.

Thanks again and best regards,
Qilong

Hello Qilong,

here some suggestions about your steering files:
- there are some recommendations about the options for telemac3d that you can find in the user guide (Appendix E). They can improve your simulations and they can have an important impact on your results;
- if you really don't want change all your parameters according to recommendations, at least you should change the advection scheme for tracer: characteristics are not suitable. I suggest you to use SCHEME FOR ADVECTION OF TRACERS = 5, then you can test SCHEME OPTION FOR ADVECTION OF TRACERS = 2 or 3 (if there are not tidal flats) or SCHEME OPTION FOR ADVECTION OF TRACERS = 4 (if there are tidal flats);
- it seems that there is a problem in the mass balance for sediment when using ADVECTION-DIFFUSION SCHEME WITH SETTLING VELOCITY = 0. I suggest you to change it into 1.
- I would clean up your gaia steering file (why do you have BED LOAD FOR ALL SANDS = YES if you have only 1 cohesive sediment ? ..)

Best regards,

Sara

Hi Sara,

Thanks for your suggestions!

I noticed that the SCHEME FOR ADVECTION OF TRACERS = 5 has better mass conservation. I will test it for sure. and also the different options of SCHEME OPTION FOR ADVECTION OF TRACERS.

Actually I have a question about the ADVECTION-DIFFUSION SCHEME WITH SETTLING VELOCITY. I set it to zero because I found in the subroutine compute_bc_sedi.f:

The option of SETDEP affects the treatment of the bottom boundary condition when solving the transport of sediment.

I noticed if I set SETDEP=1, I have to deal with erosion/deposition flux in set_diff.f (By the way, I'm now developing a multiclass flocculation model for Telemac-3D and GAIA and I need to add additional treatment for the erosion/deposition flux for different classes of sediment). For now I add it in compute_bc_sedi.f. Do you suggest that using SETDEP=1 is a better way?


The BED LOAD FOR ALL SANDS = YES is from previous tests, I think if I don't have sand in the model, it is ignored?

Kind regards,

Qilong

Hi Qilong

First, using SETDEP=1 can improve mass conservation, but I think only if you have tidal flats in your model (which you don't).

As you say, in COMPUTE_BC_SEDI.F the erosion and deposition flux exchanges with the bed (ATABOF and BTABOF) are not calculated if you have selected SETDEP=1. This is because they are instead calculated in SET_DIF.F at the same time as the vertical diffusion and settling are solved. Note that SET_DIF.F is called for each tracer separately (it is called within CVDF3D.F which is called for each sediment class/tracer)

Without knowing a bit more about what you want to do with the erosion/deposition fluxes, I can't really help (and it is a bit off topic for this thread!). What I do know is that the overall floc settling velocity will need to account for the interaction of the multiclass sediments. That will probably need to be coded into COMPUTE_SETTLING_VEL.F (old VITCHU.F) somehow. Modification to the erosion flux will probably need to be dealt with within GAIA bed erosion routines.

I hope that helps.
Tom

Hi Tom,

Thanks for explaining in detail! and good to know SETDEP=1 can improve mass conservation. I will think about how to adapt my code to this option.

In general, in the flocculation model we deal with up to 3 classes (5 tracers in total, including indexes account for the interactions between classes). When it comes to the bottom, we assume the larger sediment classes are destroyed by large shear. So all the classes will be converted into the primary particles. In this case, bed only contains the primary particles. At bottom, I deal with deposition fluxes from different classes, as you said, the settling velocities are also different and changing dynamically (this is done in COMPUTE_SETTLING_VEL.F). I also converted mass of other classes into the basic class (in COMPUTE_BC_SEDI.F for now) and make sure the mass is conserved. The erosion and deposition processes indeed are taken care in GAIA.

I shared a preliminary report with EDF developers regarding the development I did last year, and got some feedbacks. Now I'm trying to revise and finalize the code. I definitely will share again with you and other developers. And hope to discuss more about it!

Best regards,
Qilong

Hi Tom,

I tested the model with the option SETDEP=1. I noticed that, in the mass balance info printed out by Telemac, the diffusive flux at the seaward boundary becomes zero.

- Does it mean that the option SETDEP=1 solves the 1DV transport with eroson, deposition and diffusion at each node, and then advection in horizontal?

- Is it designed in such way that the horizontal diffusion is not considered?

- Is there any numerical diffusion left in the results in this case?

Kind regards,
Qilong

Hello Qilong

The SETDEP option treats the vertical diffusion (along with settling, deposition and erosion fluxes) only. The horizontal diffusion is treated separately by the usual TELEMAC diffusion schemes (see cvdf3d.f and diff3d.f).

There will be some numerical diffusion (it is only a first order scheme). Since it is a 1D solver, there is potential to make it second order in the vertical on day (something on the "to do" list!).

Does that answer your question?

Best Regards
Tom

Hi Tom!

Thanks for the explanation! I was asking about this because I found something interesting in my results in terms of mass balance.

First of all, even with the settings

SCHEME FOR ADVECTION OF TRACERS = 5
SCHEME OPTION FOR ADVECTION OF TRACERS = 2
SETDEP = 1[/li][/li]

I still have the MASS LOSS = -0.1195602E+10, as I had before. I think it means the mass is created in the domain.

Then, I computed the mass of bed and mass of suspension, and advective flux (I use 3D results and do the double integral over each transect). All these three quantities match the info given by Telemac printout on the screen. The only thing I'm not sure is the diffusive flux.

In my model, I reconstructed the total sediment flux at each transect by assuming the mass conservation for each part of the domain, since I know the mass in the bed and mass in suspension from the results. I also know the advective flux, either by calculation or from the Telemac screen printout. I subtract the advective flux from the total "reconstructed flux", the leftover I assume it's from diffusive flux.



The above figure shows the flux at the downstream boundary, at which I imposed the time dependent sediment concentration. It shows that the advective flux matches very well (they are overlapped). But there are big differences in the diffusive flux.

I still don't understand about the mismatch in diffusive flux. I suspect that it has something to do with the numerical diffusion, and the mass creation in the domain...

Best regards,
Qilong

hey can i know, how do you get the result in the graph you shown ?

Hi Sara,

I am using sisyphe, not GAIA. Do you know what I need to use as a key work for BED LOAD FOR ALL SANDS = YES in sisyphe? I have two grain size, one non-cohesive and one cohesive, I don't see this keyword works in sisyphe.

Best,
Sumaiya