You're right zpiloto - testing can hurt.

More than testing has hurt... what with the vacuum leak, transmission swap & clutch cable adjustment, and now this boat tail testing, it all adds up to not a record tank. I'm struggling to just keep it above the current 90 day average.

But - the results are good news for you. They support the idea that your mod may have been successful, despite the non-test. You have pics yet?

I'm thinking this is worth tuft testing, refining and fabricating properly. I've also asked for feedback at MaxMPG.

Consider ...

Today I should have tested at 85 km./h, not 55 mph.

Wonder what a boattail on my xB would look like....I think the rear belly pan will be first then maybe the vortex generators.

I was able to test late today without positive results. There was no significant change in FE. I was surprised because of the good FE on the commute without the EOC. (Need to start that thread). Proof positive for me that you have to do at least A-B testing for the butt dyno just don't cut it. The shape of the Mazda doesn't cry out for the boat tail like the fire flea though. I think you'll see even better results with version 3.
Since it really didn't hurt the FE I might try again with a longer extension, and see what effect that has, but I'm out of good peices of colorplast. You can come a long way down without obstructing the view through the rear window.

Here's a pattern for easy to make vortex generators:

Use doublesided tape or silicone caulk to mount...can bend by hand the surface that mounts...to conform it to the car's surface to some extent.

If you have a roll of flashing material 2 1/2" wide....you can make a bunch of these by making one 30* cut...alternating the direction.

Narrow end faces forward.

Sharp point on vanes needs to be rounded.

Vanes can be made higher by making the mounting surface narrower...Mitsubishi says up to 1" high OK.

That rear partial boat-tail that MetroMPG built is pretty much what I was trying to recommend to krousedb for his Del Sol.

I considered such a mod for my GT6, but it would make the rear hatch area inaccessable.

I'd bet a nickel that some vgs 4" apart and 4" back from the break above the rear window...and some along the sides spaced 4" & 4" would do just as well. But only a nickel.

I don't think VG's will work in my hatchback application.

My understanding of the Mitsubishi document (PDF) is that the VG's work in the scenario tested because they delay flow separation on an otherwise sub-optimally angled surface (in that case, the rear window of the EVO sedan, which slopes downward too sharply toward the trunk to retain attached flow from the roof).

Without that downstream surface for their vortices to interact with, they have no boundary layer to affect. Their overall result in that case may actually be to slightly increase drag (as is also acknowledged in the document).

Also note the Mitsu document estimates a Cd reduction of -0.006 through properly placed delta-wing VG's on their sedan. (-0.004 for the bump style VG's, also tested.)

I plugged in my car's numbers into the Aero & Rolling resistance calculator https://metrompg.com/tool-aero-rr.htm, and, assuming a Cd change of -0.006 could be achieved, it only improves fuel consumption @ 55 mph by 0.83 mpg (US), or 1.34%. So the cardboard hatch tail still comes out ahead, at 2.3%.

I'm not saying that VG's won't work in ALL hatchback applications, but the hatch angle would have to be much less steep than mine, where some flow is retained part-way down the hatch - closer instead to the angle of the Evolution's back window.

So, I'd say: save your nickel! :)

I would say to try drawing the curves in CAD and let the computer figure out the intersections. If you use a sheet metal tool, you can then "flatten" the part and print full size to make a stencil. This is how I designed the pulse jet engine I plan on building someday.