open TELEMAC-MASCARET The mathematically superior suite of solvers

Hi all,

I have a few questions and comments on bedload sediment transport models and hiding functions in GAIA.

First, with regard to the Hunziker modifications to the Meyer-Peter and Müller (MPM) equations: is there a reason this is not present in GAIA documentation, such as it being prototypical, or simply not enough time to write about it?
For those who are curious about it, it is a bedload equation from Hunziker and Jaggi (2002) and can be activated by setting BED-LOAD TRANSPORT FORMULA FOR ALL SANDS = 6 (according to bedload_formula_gaia.f). Despite being derived from the MPM formula, there is no need to specify a MPM COEFFICIENT in the steering file (set to 5.). However, it is necessary to specify the CLASSES SHIELDS PARAMETERS = 0.05 to be consistent with the formula. Also, the hiding factor formula is built-in, so there is no need to specify one. Results are consistent with classic MPM, possibly slightly better in my test case, but I haven't looked in detail.

Second, speaking of hiding factors, it seems there is no documentation at all in the GAIA user manual on these factors, though they are obliquely mentioned (without sources). I had to find the references in the SISYPHE user manual. Again, is this because there hasn't been time to transpose these sections from SISYPHE to GAIA?

Third, with regard to the hiding function of Egiazaroff (HIDING FACTOR FORMULA = 1), I wanted to point out to anyone that uses it that if the ratio Di/Dm (class / mean) is 1/19, the equation blows up to infinity due to a division by zero. A verification of the equation provided in SISYPHE's user manual or in bedload_hiding_gaia.f will demonstrate this.
The Ashida and Michiue (HIDING FACTOR FORMULA = 2) hiding function solves this by changing the rules such that when Di/Dm < 0.4, the hiding factor = 0.85 * Dm/Di.
As an aside, it seems in my preliminary research that these hiding factors essentially stop all movement; at least this is the case with a fairly broad and coarse distribution (0-112 mm).

Fourth, with regard to the hiding function of Karim, Holly & Yang (HIDING FACTOR FORMULA = 4 – where is 3?), it seems to me in perusing the relevant code (`grep 'K_H_Y'` and `grep 'KARIM'`) that the value for K_H_Y (exponent to the hiding function) is never actually defined anywhere. I assume the default value for an undefined double is 0.D0 and therefore, the exponent 0 ends up giving a hiding factor of 1.D0 across all classes. Is my analysis incorrect, and if so, where is K_H_Y set to anything?
Additionally, there is some confusion in the naming of this function. In SISYPHE, it is named Karim and Kennedy, but in GAIA, it is named Karim, Holly & Yang. I have been unable to find papers with this exact ordering of the names, which makes me wonder: where on earth did this name for the function come from? And why isn't it consistent between SISYPHE and GAIA?

Fifth, with regard to the critical shear stress in van Rijn's bedload equation, I was wondering if there was a way to include the parametric curve given in van Rijn (1984). The Shields curve is given parametrically in Fig. 1 of van Rijn's work on bedload transport, for different values of D_*. I'm going to try implementing these equations in bedload_vanrijn_gaia.f for "fun", but I have already calculated the values outside GAIA.
More broadly, a list of explicit empirical equations is available in Fluvial Hydrodynamics by Dey (2014). I'd be interested in seeing these equations implemented in GAIA someday, instead of having to calculate these curves by hand and then plug them into CLASSES SHIELDS PARAMETERS, but that is a bit of a dream...

I understand a lot of these points are fairly fine and come closer to suggestions than questions, but I want to help in advancing the OTM knowledge base for future users. Hopefully some of the comments on my observations can be useful, if nothing else.

Thanks for reading the above!
André Renault

Update with point 5: I found a subroutine shields.f that uses the van Rijn parametric curve (with some corrections that are not explained). It seems to be called by gaia_init.F with no conditions. Does this override the provided CLASSES SHIELDS PARAMETERS? If so, what is the use of providing them and why are they mandatory? If not, how can one use the shields.f subroutine to calculate critical shear stress automatically without having to provide CLASSES SHIELDS PARAMETERS?

Hello again André,

For sure, writing or updating manuals is a quite long and tough task in particular for new modules (I did a lot for TELEMAC-2D, TELEMAC-3D and WAQTEL in the past). I am pretty sure my colleagues who develop GAIA are quite busy and do not always have too much time to spend for documentation.

Anyway, any help is welcome. You can suggest modifications/additional parts in the LaTeX files and they can be pushed to main (and future releases) after review from GAIA experts.
You can proceed by Merge Requests on the gitlab e.g. or gitlab issues with your suggestions I think.

Thanks for your feedback,

Chi-Tuan

Hello Chi-Tuan,

Thanks for this reply. I'm currently focusing on finishing my degree, but I would love to contribute to documentation when I can… the bigger problem for me is that I still haven't figured out how git works.

As an aside, I'm impressed with the number of replies you made yesterday

Hello André,

You can also upload your modifications in this topic and perhaps someone can merge to the main.

Chi-Tuan