open TELEMAC-MASCARET The mathematically superior suite of solvers

Hi,

I have some questions regarding culvert calculations

I run the Telemac-2D example bridge in FV-scheme and plot the culvert discharge over time. The hydrograph has very high oscillation. What is the explanation for this phenomena?
Secondly, I split the culvert equally into three (since there were two culverts now I have six). Other values of culvert files remains same. Then I run the model and plotted the culverts discharges over time (individually and summing also). I noticed the oscillation is reduced to great extend. So my question is does splitting the culvert improve the performance?

Kind regards
Sandip

Hi Sandip,

Hard to say what is the explanation.
It might be :
- error in culvert parameters
- poor quality of the mesh around the culverts ends
- steep slope around culvert ends
- ...

The trick of splitting one culvert in several is just to deal with flow limitation in the subroutines buse.f (limitation to 90% of the available volume in the element in one time step).

If you work with coarse mesh and try to represent a small culvert, then you don't have to split it in three.

Alexis

Hi Alexis,

Thanks for quick reply.

I agree with your point but does this is due to back water effect? As the mesh is small, E-slopes is also small and the outflow BC is not that far from the culvert downstream ends. Could this be resulting in those oscillations?

As I split the hydraulic radius for each culvert decreases and this in turns dampens the oscillations. what do you thinks about this explanation?

and lastly how can we deal this in real life examples where mesh is itself complex except refining the mesh around culvert ends?

Thanks
Sandip

Using culverts, you should check :
- if Qculvert is limited or not by Qmax
- when instabilities appears? when flow type change?

Maybe you can share your case and/or a temporal plot of : Qupstream (boundary), Q culvert, Q overflow (if part of the flow overflow the deck), Free Surface upstream, Free surface downstream and Flow type (you will have to modify buse.f to print FTYPE)...

Is these instabilities sensitive to the nodes selected for culvert ends ? Two sources in the same nodes could be a kind of over constrained?

Alexis

Hi Alexis,

Thanks for the explanation.

Here are the data and figures I used for calculation. I have tried with varying the Culvert flow (CLP variable in culvert file). But the oscillations remains always.

Kind Regards
Sandip

I forget to mention that the original inflow hydrograph from examples is reduced to 6 hours with 120 reaching at 1hr until 5 hr.
This was done only to adjust the listing text, as I plotted the culvert discharge value manually.

As you can see, splitting both culverts in three doesn't just damp oscillations, it also changes the capacity.

I suspect volume limitation because of spatial discretization :
QMAX1=0.9D0*H(I1)*V2DPAR%R(I1)/DT

It means that at each time step, no more than 90% of the available volume can be used by the culvert.

To damp oscillation (but doesn't change the value and the limitation), you can change Relaxation coefficient (0.1 or 0.2 instead of 0.5).


I don't think hydraulic radius is the reason as you consider a Manning coefficient as 0 (FRIC in your culvert file) -> No linear headloss.

Also, in your case, you should not refine your mesh around culvert ends... and will not have this volume limitation with coarce mesh (-> bigger element).

I guess CLP is just the french for "clapet" meaning valve.
Anyway :
0 = no valve
1 = flow authorized only from 1 to 2
2 = flow authorized only from 2 to 1
3 = no flow

Alexis

Thanks for the clear explanations.

I tried as you mentioned to dampen the oscillations but it did not worked.

Sandip