I am modelling flood flow – as steady state flow in a semi-natural vegetated creek with a high longitudinal gradient (average around 0.5%) using HEC-RAS 4.1.
In a number of locations – the model computes (from downstream to upstream) sub-critical to hydraulic jump to super-critical to sub-critical. The result is large variations in depth and large (in the model) energy losses from cross-section to cross-section.
I have observed actual flood conditions in this creek – the reality is actually the formation of undular flow (or undular jumps) along the channel and no “real” hydraulic jumps. My visual “judgement” is that the real world energy losses are lower than that computed by the model, and that the actual flow profile is “smoother”. My judgement is that the rather large energy losses computed by HEC-RAS are unrealistic – so that the computed flow profile at the upstream location is unrealistically high.
I have “tweaked” the model by removing the contraction and expansion loss coefficients in a reach of interest where the model computed the above conditions and obtained an overall “smoother” and lower profile…. that I “feel” is more realistic.
I think this is art rather than engineering, but I think the resulting profile is more realistic – even if still too high at the upstream end of the reach.
Appreciate others opinions – is there a better method of getting a realistic flow profile where the above flow conditions occur?