How accurately does TUFLOW FV simulate weir flow when not applying a weir structure?

[Guest Chris Nielsen]

As a precursor to this discussion, TUFLOW FV allows specification of the weir equation as described in this Specifying a weir using this method provides an exact solution to the weir equation.

 

But, TUFLOW FV will simulate weir flow using the standard shallow water equations (SWE). How accurate is this approach?

 

Consider the test with a broad crested weir which is 5 m high and 250 m across with a 1000 m³/s discharge applied. Using the standard weir equation, the water level upstream of the weir is 6.765 m. Running a simulation without applying the weir equation (and using a Manning’s n = 0.018), upstream water levels can differ.

 

 

A series of tests were applied with various cell resolutions across the weir and compared to the “exact” solution from the weir equation. The tests and the resulting elevation upstream are shown in the following table.

 

 

As shown, the nodestring structure (test 4) perfectly matches the weir equation (test 6). The solution using the shallow water equations (SWE) tends to underpredict head loss across the structure. If the number of cells defined across the weir crest is increased, a more accurate solution is obtained. Note also that the solution is dependent upon friction (which is one of the dominant physical processes being simulated by the SWE, as shown in test 5).

 

Concluding, it is recommended that in situations where an accurate representation of weir flow is required a nodestring structure that inserts the weir equation into the solution scheme is used. In other situations, using the SWE (in other words, just letting TUFLOW FV simulate weir flow without any direction specification of weir structures) may be entirely acceptable. Importantly, the modeller should be aware that differences do exist between these two approaches.