open TELEMAC-MASCARET The mathematically superior suite of solvers

Hello,
I'm trying to test a free-surface flow in which a circular pipe (ascissa 0 to 20 m) exits in a rectangular diffuser.

Conditions are steady state, with negligible friction.

I obtain unexpected results: the flow experience an abrupt change in velocity & free-surface height in correspondence of the transition. It has critical conditions in all the pipe flow, and experiences also a local energy loss (of about 0.18 m), in correspondance of the pipe outlet into the diffuser.

I would expect a completely subcritical flow and no such a strong variations in correspondance of the pipe outlet: what I'm doing wrong?

Here attached:

* configuration file



* hydraulic head


* water and bottom levels


Thanks, yours sincerely,
Michele

Hi
For me, this this the idea to simulate such phenomena with a 1D model which is wrong.
Even you put a cross-section every 10cm, one of the main hypothesis for a 1D model is that the velocity is averaged in the cross-section.
In your case, with this discharge, this is not the case...
So the model try to solve this problem as well as possible but with some limitations.

hope this helps

PS: Think to update your profile

Thank you, I was indeed interested in obtaining the "ideal" solution, i.e. consistently with 1D theory an averaged velocity in the cross section (the other losses could be however be lumped in the model). This test is simplified.

Boundary conditions should be well posed for a subcritical flow (elevation at outlet and flow rate at inlet).

So you are saying that it is not possible to avoid those spikes in flow in correspondance with abrupt section changes. That's a pity.

Hi
I'm not fully sure this will be impossible but at the location of the cross section, there is a variation of velocity (averaged) which should be taken in account by the model...
Less kinetic energy means more potential energy to ensure the conservation...

Hope this helps

I agree, and indeed I find that, apart the solution in the abscissa between 19.9 and 20.2 m, the behaviour is as expected (water table rising after the transition, due to slowdown of the stream).

But the problem is likely to be more subtle: being the flow locally supercritical (in the transition, at abscissa 20.1 m), there is a sort of hydraulic disconnection between the upstream and the downstream of this section (having repercussions on the well-posedness of the 1D problem).
The singular spike is propagating its effects upstream, I suppose.
I have doubts that the spike is "physical": it seems to me that it may originate from the numerical schemes. Maybe could be avoided with a more diffusive discretization scheme? Are there alternative options in the mascaret software?