Culverts issue #277
Comments
|
@mikilterribile our unittests do try to test against expected flows. But maybe you have found an example which we haven't tested! Actually just looking at the unittest code (test_boyd_box_operator) it looks like we have only tested using exchange lines! I will try to find some time to adapt the current tests to cover using end points as well. Another possible problem is that geometry of your input regions (as generated by the end points) might be intersecting with the bank of your river. And finally another problem is that as the operator extracts water from the inlet regions, this can reduce the depth of water in that region, and so affect the flow calculation. We provide the option of setting an enquiry_point (which should be a little upstream of the culvert inlet) to use when calculating the discharge through the culvert. |
|
@mikilterribile I had a closer look at our unittests and we already have tests using So maybe the problem is with how your inlets are being embedded into the topography, or where the enquiry points are situated. |
|
Finally the size of the culvert is width b = 10 m and height h = 1 m and the manning coefficient I have choose is M = 0.01. Solving this implicit equation for Y finds: Y = 0.71. So the water depth inside the culvert in the moment of maximum flux rate should be 71 cm, while the culvert is 1 met |
|
@mikilterribile I see you have a tricky problem. My guess is that there is a positive feedback which is conspiring to reduce the flow in the culvert, ie the flow hits your riverwall and that slows down the flow (and increases the depth) and consequently the flow around the riverwall (out of the stream). By the way, I assume you have set You should be able to improve the flow by increasing the An unsatisfying remedy would be to artificially increase the width of the culvert to match the expected discharge. I'll try to create a simple example which mimics your situation and see if I can track down the problem. I do think it is to do with the way velocity to used in the formulas. |
|
Ok thank you very much.
Yes, I have set losses to 0.0.
Best regards, Michele Zucchelli.
|
|
A thing that maybe can help is that the culvert flow rate calculated by anuga seems to be completely indipendent from the roughness (the manning coefficient). I have tryed to set manning = 0.000001 and manning = 1 with the same results, despite in my code I pass this coefficient: |
|
@mikilterribile Could you test your example using lines. The problem may be due to the boyd box operator spreading out water to the triangles making up the exchange region which are determined by the triangles that intersect the line that is constructed when when using endpoints. The length of the line will be 10m centred on the end point. That may be how the water is escaping the channel. When using exchange lines you can make the lines short enough to ensure the triangles constituting the exchange region are all within the channel. Keep the width of the culvert at 10m for the flow calculations. |
|
Hi Stephen, I have tried with the exchange lines, both with 10 meters length and 4 meters length, centred, but nothing has changed. The flow rate remain about to 50 m^/s when it should be much greater. I copy the log file below, maybe it can help... |
|
@mikilterribile Michele, I notice that the enquiry points are associated with triangles that are also exchange triangles. This might imply a bug in the way we calculate the location of the enquiry points. But for now, could you try manually locating the enquiry points some distance upstream of the riverwall and down stream of the outfall. I'm thinking that the water is removed from the exchange triangles and thus affects the calculation of the flow based on state at the enquiry point. |
|
@mikilterribile Michele, it is also interesting that your exchange lines just intersect with 3 and 1 triangle. That probably means that a slight refinement of the mesh near these exchange lines would be sensible. Could you provide a plot showing the triangles in this region. |
|
|
|
I have tryed to create a high resolution zone around the inlet-outlet of the culvert but it leds to the same results... |
Hi everyone, I am trying to modelling some stuff with the Anuga's culvert operator. I use the box operator with a rectangular section 10m * 1m. The problem is that the flow rate is lower than what expected either if the culvert is gravity led or under pressure.
The amount of water expected, infact, is something like 100 m^3/2 (calculated with the uniform flow formula) while the value calculated by the culvert operator is about 50 m^3/s. I have tryed also to use e pipe culvert operator, instead the box operator, but the flow rate decrease to 40 m^/s.
NB: I use the operator with two points (start and end) and not with two lines.
The text was updated successfully, but these errors were encountered: