Thank you. Yes those drops are intentional. The reinforced concrete canals between the culvert outlets and the river will be designed to have maximum flow velocity of around 3.5 m/s. So preliminary, I took the peak flow from each culvert and did a quick calculation in a spreadsheet of the canal slope so that the velocity would be below the limit. The step downs I’ve put in so far are drastic and are located at points where the ground slope changes. Perhaps in the final design we’ll have more step downs of less height.
I will look into running a draw-down model but I admit that I don’t know yet what that means.
In regards to the last comment. I’m embarrassed to say that I don’t really know what the base flows are for these culverts/drainage ways. I’m working in Brazil – the model is georeferenced to SAD69 UTM Zone 24S. I’m told that these streams/creeks/drainage ways what have you, actually have sewage from the urban areas nearby. I would assume this means there is some base flow far above the flows I’ve specified. After saying all that, I’ll be putting in some base flow. My guess at the problem is that the solving routines of HEC-RAS can’t deal with such low flows at the beginning of the simulation with such high slopes and the solutions give these wild energy grades.