Liquid Oxidizer Impeller

At my school there is a model rocket group, Launch Initiative. They design, build and fly mid size model rockets. Typically they use purchased solid rocket motors, or hybrid (solid fuel, liquid oxidizer) motors they design and build. While working on some other projects, they are planning for the future and trying to push their engineering skills by designing a hybrid rocket motor with a oxidizer turbo pump.

This project was more of an experiment and test of skills, and this version wasn't intended to fly. Twofold to this project was the design and mathematical modeling of a pump impeller and turbine, and second to manufacture the pumps and impellers. The design was done by one of the Launch Initiative team members; I was only on board for the machining.

This is the second generation of pump impeller, made in aluminum. This machined one is scaled down from its original size to result in a roughly 4 inch (100mm) diameter. This scaling was done primarily for cost; I already had material for this smaller size. It was also done to replicate the size of what a final impeller would be. A flight ready version of this impeller would be out of stainless, and the turbine would be inconel. When I finished machining this impeller, the team had already progressed to the fourth generation of impeller which added splitter fins. 

All programming was done in Fusion360 with a custom post and run on an Okuma Genos M460V-5AX. Absolutely fantastic machine, no complaints with it. There is a real joy in using a machine that does what you tell it to first time, every time. I cannot say the same about Fusion for this type of work; not at all suited to it. I first tried to program the part with simultaneous 5 axis moves, but Fusion couldn't handle it. In the end I had to split all the faces up and do them as positional 5 axis moves. Visually not as good but these impellers would have to be polished before service anyway so this would be functional.

Cycle time on this was around 4 hours a part. This could definitely be cut down by pushing roughing and semi-finishing federates. Using a better CAM package to optimize the tool paths would also help. I have an Esprit license which is the preferred CAM for Okuma machines (they have a partnership and sponsor my school), and the machine simulation models are very accurate. 

Checking clearance inside the machine when testing

Fusion doesn't have machine simulation, so I was a little weary for some of the 5 axis moves. I had improvised a basic simulation that would prevent major crashes, but it was inaccurate for close clearances. There was a risk of crashing the back of the spindle into the table when rotated up at steep angles. This was solved by propping my camera on the trunnion inside the machine. From there I could stream to my smart phone and zoom and watch the clearance in critical areas. I would run the part dry with no stock at a slow speed to I had a hope of being able to react if something goes wrong.

The machining process was quite straightforward. Most of the tooling was Sandvik inserted, solid and drills. I used to be skeptical of the premium price of Sandvik solid carbide tooling. After using them for an extended period of time, I think they are worth the price. The carbide quality is just phenomenal and the cutters just last and last and last. Maybe if were paying for all the cutters out of pocket, I would change my tune but as long as work is buying them (or Sandvik is donating them).

Various stages of machining

In the end, this part was also used as a demo piece for an on-campus creativity festival. I put together a little display with the different stages of of machining seen above, along with the tools used and the chips they produce. This was to show to people who don't know about machining how a part is made. A lot of people thought these impellers were 3D printed and were surprised to find it was cut out of solid.

The project was fun. I think I made 6 of these in the end. I would like to revisit it eventually with a better CAM package. This was only my second 5 axis project and I made it with only 100h of 5 axis machine time under my belt.

Pretty reflections after roughing