open TELEMAC-MASCARET The mathematically superior suite of solvers

Hi Mr. Coulet,

With your previous suggestion, I moved the entry/output nodes from boundary to inside , and the result is much better now! see the figures below.

And I tied to use 'ELEMENTS MASKED BY USER' to deactive the elements between boundary and the culvert extremity,but seems not working,the boundary node behind culvert extremity still has velocity. Is there any other method to deactive the elements/nodes bewteen boundary and culvert extremity?

Another thing is that there is still slight difference between folw velocity of culverts although culverts have same discharge, gate area and other setting, Could it be improve?

Best Regards,
Bin

Hi
I'm not sure using elements masked will enhance the solution for culvert as it will react like the boundary...

Maybe you could check/use the parameter about the direction of culvert for the flow at the output. It was supposed to be automatically computed but maybe it's not the case.

Velocity depends on the discharge and the water height. And discharge is the result of a computation using upstream and downstream water level. the computation is done in an explicit way but we smooth the change of value with a relaxation coefficient so it's in fact a semi-implicit or semi-explicit computation of the discharge...
Nevertheless, it's quite sensitive to small changes of water height.

In my opinion, it could be interesting to investigate a solution based on liquid boundaries where the computation of discharge is done for the global system of culverts. This will be probably more stable and give a better velocity field.

Hope this helps

Dear all,

I have done some more reseach on this problem. And I think this problem is caused by the inconsistent V2DPAR(the area around the node.

The discharge(Q) on every turbine nodes are the same but not the area around the node. Because the SMH is equal to Q/V2DPAR. so different V2DPAR cause different water depth in this area.then result in different velocity.

So far, this is my understanding.My solution is to generate a more partially-homogeneous mesh like this.The pink points are the location of turbine so they will have similar V2DPAR and the result looks better.

To generate a mesh like this cost me some time. Is there a better way to generate uniform mesh?

Best Regards,
Bin