open TELEMAC-MASCARET The mathematically superior suite of solvers

Hi,
I'm attempting to model bedload transport and deposition over an initially rigid bed. I can't successfully define bedload inputs.
I’m using multiple (3) sediment size-classes, SISYPHE coupled with Telemac2d.

1. At the very first iteration, there is bedload transport throughout the model domain. I.e. doesn't seem to respect the rigid bed at this initial timestep.

2. At this very first timestep, the boundary Bedload Flux as shown in the Sortie file is 1.25x what I define with PRESCRIBED SOLID DISCHARGES (I guess the 1.25 is the difference between 'without voids' and 'with voids' (?)

3. If I look at the output QSBL for this timestep, it gives a quite different flux. Also it is variable across the width of the (variable depth and velocity) boundary, not what I was expecting when I set the same values for 'Q2BOR' across the boundary in the boundary conditions file.

4. Looking at the QS1, QS2, and QS3 outputs along the boundary, there is a ratio between the different classes, but not sure what it is (i.e. not set by INITIAL FRACTION FOR PARTICULAR SIZE CLASS, as I thought it might be).

5. By the second timestep, bedload transport has stopped across most of the model domain, and the inputs at the boundary (as observed with either the sortie file or QBSL/QS1,QS2,QS3 outputs) have dramatically changed. At some boundaries increased, at some decreased..

6. At these non-initial timesteps, the split between size classes across the boundary is odd, e.g. in the deeper channel section there is only input from Class1 (finest), at other points only class 2, at other points only class3..

7. Setting inputs directly in the boundary conditions file gives the same result – dramatic change in input after initial timestep, and the output in QSBL doesn’t relate to what I entered in the BC file.

8. Similar results for different bedload formula (Meyer-Peter, Einstein-Brown, van Rijn).


Sorry for the long post – would really appreciate any explanation or advice.

(My next thought is to try defining inputs in conlit.f, but I hoped to get away without dabbling in fortran – any advice on this also appreciated).

Thanks,
Dougal

Dear Dougal,
the number of superimposed processes of your application makes difficult to give a prompt response. In order to « isolate » the cause of the problem(s), I'd recommend to do the following :
(1) try to run a simulation without rigid lid and with only one sediment-size class
(2) run a simulation now with rigid lid and with one sediment-size class
(3) go to the a next simulation, without rigid lid and with the 3 sediment size-classes
This could give us some hints to identify from where the problem comes on.

Let us informed, 

all the best,
pablo

Hi Pablo,

Thanks for your suggestions, there were a few things going on!

The issue regarding defining a sediment input rate still show up in your run (1):

-The bedload flux as shown in the sortie file doesn’t resemble my PRESCRIBED SOLID DISCHARGES, or the QBSL results. In fact it shows a negative flux on one of the outflow boundaries, and 0 on all other boundaries.

-The QSBL results show sediment inflows at the inflow boundaries, but not the quantities I had prescribed, and spatially variable across the boundary, as opposed to my constant values for Q2BOR in the BC file.

-Setting desired solid discharge in the BC file without using PRESCRIBED SOLID DISCHARGES gives the same results.

-Prescribing 10x less or 100x greater sediment inflows in PRESCRIBED SOLID DISCHARGES produces the exact same results.

-Changing the sediment diameter changes the quantities in sortie file and in QBSL, but otherwise the above observations all the same.

Attached are my steering files, BC file and sortie file, if you wish scan through for anything obviously wrong.

Are you please able to send me a model that successfully uses PRESCRIBED SOLID DISCHARGES so that I can see that it works as I imagine it should?

With thanks,
Dougal

Hi Pablo,

I have found the validation case ‘002_canal_solid_discharge_inflow’, which runs successfully with an imposed bedload input.

The Bedload Flux for Boundary 2 shown in the resultant Sortie file is constant and equal to the PRESCRIBED SOLID DISCHARGES / (1- NON COHESIVE BED POROSITY). As expected.

I’m not sure how to interpret the results file, however. The Bedload discharge (i.e. DEBIT SOLIDE or QS CHARRIAGE) value at the inflow boundary is changing in time, in contrast to the constant prescribed input. What is this parameter? It has units of m²/s - I had assumed it was bedload flux per m width, but evidently not.

Best Regards,
Dougal

Hello,
maybe you can attach all necessary files for the run,
i.e. the selafin file C2_GEO.slf

cheers
Annalena

Hi,

I've attached complete running version of the model (with coarser mesh, to keep the file-size reasonable). I do appreciate you taking the time to have a look through.

The model has 4 inflow boundaries with prescribed flowrates for which I am attempting to prescribe a bedload input, 2 outflow boundaries with prescribed flowrates, and 1 outflow boundary with prescribed elevation.

I have a feeling that I haven't got the Sisyphe Boundary Conditions file right.

Among other questions..

1. What values should the Sisyphe BC file have for each of the three boundary types described above?

2. Can someone please explain what the parameter 'QS BEDLOAD' in the results file represents, and how this relates to BEDLOAD FLUX in the sortie file? If I successfully prescribe bedload input using Q2BOR in the BC file, should I see the same values in 'QS BEDLOAD' in the results file?


Thanks,
Dougal

Sorry, attachment too big at 2.5MB.
Can be downloaded: docs.google.com/open?id=0B6vTQb1BDNoAMXo5eGNPejNkZGc

Hi Dougal,

I am not sure my comments will solve your problem, but there are a couple of things I see in your questionsthat seem incorrect:

The prescribed solid discharge is the total flux through the whole boundary (per liquid boaundary, not per elemant/node)

In your steering file the line
PRESCRIBED SOLID DISCHARGES =0;0;0.0033;0.001155;0.00559;0.00431;0 / m³/s without voids
is very long (If I would run this, the part =0;0;0.0033;0.001155;0.00559;0.00431;0 / m³/s without voids
would not be used. THis might explain that changing the numbes doesn't change anything.
Try splitting it over two lines:
PRESCRIBED SOLID DISCHARGES
=0;0;0.0033;0.001155;0.00559;0.00431;0 / m³/s without voids

QS BEDLOAD is the sediment transport rate per meter for each node

You seem to use a seperate BC (cli) file for sediment and flow. You can use the same for both as Sisyphe does only use
X1, X2, X3, X4, X5, X6, X7, LIEBOR, EBOR, X10, X11, N, K.

In your cli file, you set LIEBOR to 4, which makes it a free bed level and will allow erosion on the boundary. This contradicts your comment that you are running with a rigid bed at the boundaries. Change LIEBOR to 5 (fixed erosion rate) with EBOR set to 0.
You cannot prescibbe trasnport rates in the cli-file, only erosion/sedimentation rates.

As you are using
OPTION FOR THE TREATMENT OF NON ERODABLE BEDS=0
which makes SISYPHE ignore the rigid bed
use
OPTION FOR THE TREATMENT OF NON ERODABLE BEDS=3 instead.

I hope all this helps in finding the solution to your problem

Michiel

Hello Michiel,

Please note that the keyword:

OPTION FOR THE TREATMENT OF NON ERODABLE BEDS

is not used if you use the finite element option, it is used only with the finite volumes option. Actually in finite elements we solve the Exner equation in a way that preserves the monotonicity of sediment height (same advection schemes that preserve depth positivity), so you naturally cannot dig into the rigid bed.

With best regards,

Jean-Michel

Hello,
to complete the previous comments, please take a look to the attached document.

all the best,



Pablo