open TELEMAC-MASCARET The mathematically superior suite of solvers

Hi,

Has anybody managed to run a SISYPHE/GAIA model using a time variable morphological acceleration factor (MOFAC). As I understand, this is not possible in the vanilla version of SISYPHE and would require some modification of the source.

I would like to attempt a medium-term morphodynamic model, whereby the acceleration factor varies based on the wave conditions being modelled, similar to the reduced input methodology of Lesser (2009), "An approach to medium-term coastal morphological modelling"


Regards,
Toby

Hello Toby.

What I do to tackle the problem you mentioned, in principle, is to discretise the wave climate in equivalent wave conditions and for each condition define a MOFAC and computation duration with respect to the wave condition's frequency of occurrence. The schedule each computation the one after the other using TELEMAC's 'COMPUTATION CONTINUED' functionality.
Else, you can check whether you can change MOFAC in realtime using the API.

Best Regards,
Costas

Hello Toby,

we are working with the morphological factor for river applications at BAW. So far we didn't try a variable morphological factor. But we tried a variable time step:
We increase the time step (dt=dt*1.1) if the number of iterations of the solver is lower 25 and decrease the time step (dt=dt*0.9) if the number of iterations of the solver is bigger than 40. Additionally we limit the time step between a lower and an upper boundary.
You could try a similar procedure for the morphological factor.
In our river application using the variable time step did not give the effect we hoped for. The computation time was nearly the same than with a fixed time step. The results were also very similar. But the post-processing becomes much more complicated with variable time steps. For this reasons we decided to stick to constant time steps.
But this doesn't mean that you will also not have an effect of using a variable morphological factor. It depends very much on your hydrograph and on the chosen procedure (relaxation factor, bounds).
Best regards,
Rebekka

Hi
Just a small remark for Rebekka and others who are interested to try the proposed methodology:
Multiply by 1.1 to increase and multiply by 0.9 to decrease is not a neutral operation as the result became the initial value multiplied by 0.99!
In my opinion, it's better to multiply by 1.1 and divide by 1.1

Regards

Appreciate the suggestions - likely I will attempt it through a series of continued computations.

On another note, I am having difficulties getting computations to run stabe with a morphological factor applied. Initially I was using T2D-SISYPHE. There are a lot of intertidal areas and some places for which high erosion occurs early in the model however the simulation runs without issues for the entire sim period without acceleration.

However, including the morphological factor (tried 10 and 20), the model randomly crashes on one of the parallel cores about half way through the simulation period. Thinking that it may have been bug related I have recently ported the model to GAIA, however it also seems to be crashing.

Given that the MORFAC simply accelerates the bed evolution I wouldn't expect it cause too many issues stability wise, especially considering that the results seem to look ok at the time of the crash (hydro and evolution). From what I've seen people have managed to run much larger morphological acceleration factors with suites such as DELF3D-FLOW.

What have your experiences been like?