Can a segmented secondary coil work?

[Hafting]

I am planning a coil. So far, I have decided on a QCW that utilize the AC supply as a natural voltage ramp, as this limits the component count and keeps it simpler. I want to build a bipolar coil, in order to have sparks from two ends. But I want both ends to point upwards to some extent.  The first idea was to make a U-shaped secondary.  (Well, less than a complete 180 degree U, so the two tops don't get too close.) 

But making a U-shaped coil seems tricky. So the new idea is to get a straight pipe, and saw almost through it at regular intervals. This so it may be bent later.  Then, wind a coil while the pipe is straight, skipping past those sawed-through places. This should yield a coil with 3 or 4 close segments. After this, the pipe can be bent perhaps 30 degrees in each of the sawed places. The coil segments will still be close, and hopefully with enough magnetic coupling that they work like one big coil.

So the question is, if this might work?  Or is there some reason why it cannot? 

[davekni]

So the question is, if this might work?  Or is there some reason why it cannot?

I expect it can work at least to some degree.  Will be interesting to see how it behaves.  Arcs may bend towards each other or even join depending on top spacing.

Two areas may need attention.  Perhaps more as others respond with different thoughts.
1) Coupling may be lower depending on coil shape, though it is also possible that coupling would be higher.  Might be able to estimate with FEMM simulation, even though FEMM is only 2.5D.  Simulate only the primary coil, then manually check field strength at points along your secondary shape and along a comparative straight shape to see which is higher total.  QCW coils generally work best with high coupling.  A reasonably-long straight section of secondary inside primary should help (or outside if you chose an internal primary).
2) Electric field along secondary will be more nonuniform, so increase chance of arcs part way up secondary.  This is less of a problem for QCW since peak voltage is lower relative to normal DRSSTC.  Still, you may want to add thicker coats of urethane epoxy for insurance.  Perhaps a thin coat to keep wire in place before bending, then remaining coats after bending.