Many water surface profile software programs prior to HEC-RAS had a quirky rule against having perfect vertical sides.  In other words, you could not have two or more elevations at the same stationing.  A perfect example of this is a rectangular flume.  The workaround for the older software was to define the tops of the flume to be 0.01 ft or meters just outside of the bottom edge of the flume, like this:

In HEC-RAS, you don’t need to do this.  In fact, you shouldn’t do this as you’ll get the wrong answer.  Let me explain why…

In a rectangular flume, calculating normal depth is rather simple.  A little trial and error with Manning’s equation and you have it:

  Let’s say the discharge is 12 cms, Manning’s n value is 0.014, the width is 4 meters and the slope is 0.001 m/m.  For a rectangular flume, Area, A = depth, D times width, B, and the hydraulic radius, R is Area divided by wetted perimeter, or BD/(B+2D).  Using a little trial and error on a spreadsheet, and we get Depth, D = 1.48 meters.
If you ran the example above in HEC-RAS as a long prismatic flume, you would get D = 1.19 meters.  The reason for this is that in the above example, RAS actually breaks out the very small 0.01 meter wedge on each side and treats them as the left and right overbank, which in HEC-RAS are always computed as individual conveyance elements from the main channel.  So RAS is actually solving Manning’s equation for each sub section independently and then summing together for total conveyance.  You are effectively taking the wetted perimeter (or roughness) of the side walls out of the equation.  You are making the flume much less rough, and much more efficient, so it can push more water at a lower depth.
The wedges on the sides are so small that the Area and hydraulic radii are effectively 0 and the computed discharge is also 0.  If we now compute the main channel without the side wall wetter perimeter, the hydraulic radius for the main channel becomes R = BD/B and the hydraulic radius is the same as depth, 1.48 m.  Since R is the same as D, the Manning’s equation can be solved explicitly without trial and error,
where Q = 12 cms ,B = 4 meters, n = 0.014, and S = 0.001.  The resulting depth is 1.19 meters.

Fortunately, in HEC-RAS, we can just code in the flume as it really is, sharing two elevations for one stationing value as follows:

If, for whatever reason, you wanted to keep the same station elevation points as the initial try, you could still get the correct answer by moving the left and right bank stations to the outside points as follows:

The bottom line is the first example above will give you an incorrect normal depth of 1.19 meters.  The second two examples give the correct answer of 1.48 meters.  Give it a try and see if you can replicate this.  More information on conveyance distribution in HEC-RAS and it’s implications can be found in the hydraulic reference manual on page 2-4.