Radial Flow Turbines

I wanted to build a small pump to inflate my boat. It turns out that a small radial flow turbine has to spin extremely fast (> 30k RPM) to get any serious pressure from air. I made some anyway, although the motor I had at the time didn't spin fast enough. It was pretty fun to spin them up and feel the suction at the inlet.
I used ANSYS' Vista CCD and BladeGen to design some mini turbines.
Vista CCD lets you layout the turbine analytically.
You can then tweak the design with BladeGen, which outputs 3D geometry as a section of the blade set.
The next step is paterning the blade section to make a full turbine, and adding fillets and a case.
Here are two variants: an 8 blade and a 12 blade mini turbine.
Again from the side. These prints were pretty quick and dirty. PP30GF is usually pretty dimensionally stable, but in this case with a fully-filled bottom layer it tended to pull the packing tape up. Packing tape (which is PP) is the only solution I've found for printing PP, aside from printing on PP sheet, which I've also experimented with.
Some simple FEA (here with TPU).
The TPU is softer than PP30GF, but even so the deformations aren't that big (30k RPM). Even so, maintaining pressure at either end of this turbine would be quite challenging. Maybe a loose labrynth seal.
The motor is from a BRAUN electric toothbrush and can handle 40 W for short periods. But it just doesn't get up to 30k RPM with a turbine on it. I have a rotational speedometer that's laser-based.

This graph shows that at a given RPM, the 8 blade turbine uses more power. I believe this shows that it pumps more air per RPM, and I think the 12 blade turbine is too "choked" by the additional surface area and reduced cross-sectional area of the channels. The rough surface finish of these prints doesn't help!

Here is the 8 bladed mini turbine spinning up with 2 VDC at 6.5k RPM. It's printed with PP30GF which is very tough and reasonably stiff. When these spin up they tend to self-balance.

One funny thing is that if you run them backwards, they still blow in the right direction. Makes sense because they are still slinging the air outward at the outlet. My fluids professor said it was hard to design a radial flow turbine that doesn't work. Point proven!

And here is the 12 bladed design, which was relatively choked as compared to the 8 blade design. This is about 5k RPM.
Here are some earlier failed attempts to print using TPU. It's challenging to print these without any support material, and I ended up switching to PP30GF, which is the only other tough filament I can print. You can see some of my notes for EdgeBander (an edge seaming tool) underneath.
Resume - Lebenslauf