open TELEMAC-MASCARET The mathematically superior suite of solvers

Dear Telemac Community,

I have a question regarding some of my results obtained with different bed slope formulas for a 2D planar embankment breach induced by overtopping flow. The experimental set-up im trying to reproduce is based on the following:
- Finite reservoir volume
- Fine sediment of d50=0.245mm
-Homogenous embankment
- Overtopping flow induced by: instant release of a plexiglass release gate ( water level was raised 0.1 m over the crest)

I attached the files (pngs.rar) containing experimental set-up and my results.
I did my simulations based on the MPM formula with a bed slope correction based on Soulsby which gave me the best fitting results. When i try the Formula by Van rijn or the total load formula by Soulsby-Vanrijn the discharge increases considerably compared to the original experimental results. I ask myself if this result is physically logic as in the sisyphe manual v6p3 it says: "
Most sediment transport formulae are based on experiments performed under fluvial, unidirectional
flows. These formulae shown a rapid variation of the bedload transport prediction, as a function of the mean flow intensity. Therefore, an increasing of the current velocity by 10% will
result, depending on the formula being used, in an increase of the transport rate of over 30%
(Meyer-Peter), 60% (Engelund-Hansen) or almost 80% (Einstein-Brown). Therefore, any error made
when calculating the hydrodynamics will be significantly amplified by the sediment transport rates
estimates ".

Is it possible that my hydrodynamics are computed wrongly? How could i gain insights on this issue?

Secondly, my hydrographs depict high fluctuations in the domain between t= 0-50s. Is it possible that the simulation is unstable (my time step is already really small... t=0.001)? Or does the model has difficulties predicting such complex hydraulic boundary condition ( instant high level water release)...

I hope you can give me some thoughts about this issue.

Thanks a lot

PS: I attached the files containing the results and also my cas files of the MPM and Van Rijn simulation.

Hello

You have answered to your own questions:
1- sediment flux is very sensitive to hydrodynamics fluctuations
2- thus the first step is to be sure that hydrodynamics is well reproduced. To do that you have to calibrate your model using some elevation and velocity measurments. You can start by running Telemac2d alone and than couple it with sisyphe.
3- This kind of flows is highly transient and it could be better to use finite volumes
4-results with MPM are not bad, why you will not go on with that option?
5- the steering file is very poor, use rather a steering file from one of the validation cases of Telemac2d. Choose the one which is the closest to the physics of your model.
6- time step is very small, you did not attach the mesh to see if the time step is well chosen of not, however, you can see that when reading the listing and the result file.

I hope that this helps

with my kind regards

Riadh

Hello,
I'll complete Riadh's comments on this interesting topic.

* Sediment transport formulas are usually designed to be applied under particular flow and sediment conditions. The use of MPM with the slope correction probably fix better the range of validity of the other sediment transport formulas that you also considered.

* Please take also a look of your Froude number. Results can be strongly affected as you move towards the supercritical range. Please also consider to take a look to this article :
Soares-Frazao, S., et al. (2012). Dam-break flows over mobile beds: Experiments and benchmark tests for numerical models. Journal of Hydraulic Research, 50(4), 364-375.

All the best,

Pablo

Thanks again for both of your comments and the paper! I will have a deeper look at it.

@riadh: the reason I am trying Van Rijn as well for the bedload is that my sediment is very fine d50= 0.245mm and the VR formula was calibrated with that range of sediment diameters. But still, I am having troubles to physically explain the divergence between both of the formulas ( MPM vs Van Rijn) since the difference is very big when I look at my hydrographs. That's why I thought that there might be a mistake in the hydrodynamics... But double checking they seem to be fine...

Hi Galopito,

I’d like to add a small observation to Riadh’s and Pablo’s comments.

The breach discharge is a lumped consequence of both the hydrodynamics and computed shape of embankment. A match between computed discharges and measured does not necessarily reflect the adequacy of your sediment transport formula nor its parameters. Thus, I join Riadh on suggesting validating your model with elevation and velocity measurements. Maybe evolution of embankment profiles can also provide valuable insight. I also recognize in the joined figures the experimental setup of Mahmoud Al-Riffai. I guess this kind of information is available, water surface and embankment profiles at least.

On interesting case to look at for flows over embankment like geometries is: Lukas Schmocker, Berglind R. Halldórsdóttir, and Willi H. Hager (2011). Effect of Weir Face Angles on Circular-Crested Weir Flow. Journal of Hydraulic Engineering 137(6). It can provide interesting calibration data for the hydrodynamics parameterization.

Moreover, considering two sediment transport formulae, a small difference in sediment transport capacity at a certain time step will result in a small difference in the updated embankment profiles. Then, at the next time step the difference in the computed sediment transport capacity (inherent to the sediment transport formulae) will be exacerbated by the by higher overtopping heights… and so on for the next time steps. Thus the “runaway” of the process and the intensified hydrograph. Values of bottom sheer stresses can also provide interesting insight in this matter.

Regarding the choice of the formulae, I join Pablo on checking the range of validly of the formulae you use not only regarding sediment size but also flow conditions and slopes.

Ismail Rifai
Phd Student at EDF R&D (LNHE Department)

Hello everyone and thanks again for all the answers!

The available data I have are the bed profiles and water surfaces at different time steps ( from 0 to 30 s but not from later steps) and the breach hydrograph. this question might seem obvious but how should i further validate my model ( using velocity data) if i dont have that kind of information from the experimental work?... Also regarding the Froude numbers, i can easily plot them but i dont have original data to compare it with... For instance, at some time steps the Froude number were around 5-6... So how should i use that kind of information?

Thanks again