FEA actually approximates a structure by filling it with triangular elements and calculating the effect at each corner and in the middle of each straight line, TET-10 is a preferred element, so-called because it uses tetrahedrons and computes at 10 points (6 sides 4 corners).
Engineers like triangles because you can consider the sides to be in tension or compression andsuch effects tend to be an order of magnitude more significant than the bending moments seen at the corners.
Take a look at any tube frame, such as for an aircraft fuselage or a race car, and you will tend to find triangulation. Tubes can be of smaller section when they can transmit loads by tension or compression instead of bending.
None of which really speaks much to the question at hand. The thing that can't reliably or easily be simulated by FEA is the effect of welding in something to close off a C-channel section. Weld tends to be a stress riser, especially when it can't be properly normalized. Welding tends to create an area that has been heated resulting in a higher local hardness, and higher local strength, and at the edge of the heat effected zone is where you are most likely to find fracture first (on good welds).
Which means what? As far as the question at hand, probably nothing again...
In practical terms, though, there is a problem with boxing a c-channel traingularily (if that's a word)... It's an acute corner, and that isn't usually very good for getting the end of a wire to where one piece of metal touches the other. You're more likely to create a weld that is like bridging an air gap, and that means a non-optimal weld... One thing you don't want is transmitting load through weld, you want welding to be the joining of metal from one piece to the other by melting them together.
You could do this by cutting off the unused bottom flange, but now it's more work. And a sharp cormer to hit your head on.