open TELEMAC-MASCARET The mathematically superior suite of solvers

Hello,

I've launched two steady simulations of a river with a couple of bridges where the flow is expected to pass from free surface to closed-conduit. The simulations are as the following :

1. Steady state simulation without upper bridge deck. (The flow is never in closed-conduit here)

2. Same simulation with the upper bridge decks using a TUBE FILE. (The flow can pass from free to closed conduit here).

The results of the elevation of free surface are shown on the attached figure. The different bridges are modelled with a TUBE FILE containing in total 77 tubes to describe all the bridges (11 tubes per bridges approx). I've taken care to calibrate CE and CS on time period where the flow does not go in closed conduit (only dataset available). Their value is CE = 0.1 and CS = 0.3.

However, when I launch the two simulations described above and shown on attached figure, i get weird results. The blue curve is without bridge decks and gold curve is modelled with TUBES.

I was expecting that the model in closed conduit (gold curve) gives a higher free surface near the upstream bridges where the flow is pressurized. However, i'm getting the contrary.

My questions :

Has anyone experienced something like this with TUBES ?

Is my interpretation of the results is OK ???

Are my CE and CS too low for when the flow becomes pressurized ?? As mentionned before, they were only calibrated on time period where the flow is not pressurized under the bridges.

If my CE and CS too low, what are standard values for pressurized flow ??


Thanks


Mathieu

Hi Mathieu

One question about your first simulation (blue curve). How did you represent the bridges in this case (tubes with a high value for height?) or directly in the mesh?

How is your mesh around the bridges?
You indicate 11 tubes per bridges, it's a large value! How large the river is?

I'm OK with you, you should have a water level higher with the pressurized flow but without any other informations it's hard to conclude

Best regards

PS: Please update your profile!

Hi Christophe,

Thank for your quick reply

Here i'm posting 2 .jpg of the mesh around a particular bridge that is supposed to go from free to pressurized.

The mesh in the river bed is between 1-2 meters and 5 to 25 in the flood plain. We have lots of data in the river (ranging from 1 to 5 m) so that is why we chose to go with a very small spatial discretization.

The river is approx 15 to 20 meters large. I've put a lot of tubes so that "simulation 2" is as close to "simulation 1" as possible.

I'm satisfied with the vector field in both simulations. However, it's the free surface that i'm not satisfied with the simulation with tubes (gold curve).

I fear that my loss coeff CE and CS are too low, but even how, i would have expected that gold curve is higher than blue curve.

CE and CS were calibrated on data set where its exclusively free surface flow.

Do you think its the cause ???

Should I increase the coefficient when the bridge becomes in pressurized flow ???

Are the values of CE and CS in the TELEMAC bridge example are standard values ??? Maybe I should CE = 0.5 and CS = 1.0 when the flow because pressurized ???

What do you think ???


Mathieu

Hi

Ok for the mesh part. I suppose you checked the cross section in the mesh and describe in the tubes are similar.

After thinking about your coefficient, I think this is the main problem.
I don't know what is the value you gave to L but I suspect it's 0.

If you look into the program for the different computation of discharge in the tubes, you will see that the parameters CE, CS and L are not always use (it depends on the situation).
Probably the rules I defined in buse.f are not well define for some cases...
In buse, pressure flow means Upstream water level or Downstream water level is higher than the height of tube.
If this is the case upstream, we consider only CE and L.
If not, then this is the case downstream and we consider CE, L and CS
In both case, if the sum (CE+L or CE+L+CS) is lower than 1 the computation will give a lower head loss for the same discharge.

So I think you should modify your coefficient to have a sum greater than 1.

About the calibration of those coefficient with a free surface flow, if you look also in the code, you will see that the computation discharge follow a weir law and the coefficients are only used for flooded situation.

If you think this programs is not well suited to your case, you could also adapt it.

Hope this helps

Hello Christophe,

Thank you for your explanation, it's very interesting !

Based on your description, I think the program is suited for my case.

However, I'll certainly look into the code to improve my understanding.

For L, I have used 0.2 as in the TELEMAC example. I haven't tried to modify it for now as I only wanted to play with CE and CS. I'll try to increase L so that the sum of the coefficient is higher than 1 and see what it does.

About the part where you said that the coefficients are only used for flooded situation, I see that if I make them vary with a low flow it does makes a difference. At this low inflow, the simulation remains free surface everywhere, but the coefficents CE and CS does have an impact. (see attached figure about my calibration). L = 0.2 for each curves...

If I understand what you said, CE and CS should not have any influence if flow is free surface ??? Is that correct ? Maybe then I have an error in my Tube file...


Thanks

Mathieu

Ok I've looked into the code and CE, CS and L are used no matter what.

What would be interesting to program but maybe too advanced for my Fortran knowledge would be to keep river mesh below the bridge when flow is free surface and then switch to tube file when one of the conditions in BUSE.f gets met. In that way free surface flow would be best described by 2d model as longest as possible. I'm not sure if it would require two different mesh or not.... Or if it could be done with only one...

Anyway thanks


Mathieu