Inside Out?
Compared to the way most of the axial flux motors are built you are looking to create an inside out version. Usually the rotor holds the magnet wires and they use dual discs of magnets on the outside of the rotor which amplifies the magnetic flux inside.
I can see the value of doing it the other way.
You could do an axial flux design with the poles facing "parallel" to the motors axle alternating N - S - N - S... and then surround it with coils on either side. Or radial. I see your question, but I don't know what the formulas might be to know which is better. I'm using radial because I want any magnetic junk that might get onto the disc while riding more likely to get thrown off. (so my choice was purely practical)
The advantage of the Halbach configuration is that for the same amount of magnet material you get the square root of two (about 1.4 or 40%) more magnetic strength. Also with the Halbach you need no iron at all as opposed to being able to just live with a weak backside field without it.
The advantage of the standard configurations is that it's going to be easier to assemble the motor because there is no need to install magnets that want to rotate themselves out of alignment. It's actually the difficulty of assembly of the Halbach that is making it tricky. (from what I've read)
---------------------------------
This standard axial flux motor design could be used as a "Disc Motor" and attached directly to the wheel making effectively an ebike "hub motor" for a small fraction of the weight.
No matter how the idea proceeds (Halbach or otherwise) it's good to see that there is interest in this new direction for ebikes.
You also might search the net and download every pdf file you can find about axial flux motors. I've spent months doing the research and if you are building a motor you owe it to yourself to "do your homework". The odds are that if you search long enough that you will find specific formulas for how to evaluate different designs.
One last thing... there are software programs that are sure to exist that could model the motors visually. I'd love to get my hands on something like that as I found that when I was learning about circuit designs and discovered SPICE software that what was difficult all of a sudden became easy.
If you find some freeware motor simulation software be sure to let us know !!!
It would be great to get a hold of something like this:
http://www.infolytica.com/en/products/motorsolve/bdc/
They appear to have a trial version... maybe you could try it out?
Let us know what you think...