[jabowery]

In the mid 90's I was using TK!Solver to do calculations for an ultracentrifugal rocket engine for which Roger Gregory and I received a patent. The calculations were obviously not going to be nearly as adequate as they would be using CFD simulations today and the patent fees sucked the life out of the project about the time the DotCon bubble burst. That put a stop to our work.

But given:

• the advances in open source CFD
• the increasing enthusiasm in rocketry
• the design focus targeted fabrication in high performance automotive engineers (ie: NASCAR)
• the thrust to weight ratios that looked superior to what SpaceX is now trying to achieve with Raptor 3

I figured maybe an open source engineering project might be worth a shot. Has anyone experience with something along these lines and if so how should I proceed?

Here's a look at the engine -- and before you say it's insane, read to the end. Some things are not obvious (ie: one of the characteristics of patentability).

This is a side view of Fig. 1 in the patent, LOX inlet on left, LPG inlet hidden on opposite side:

sidephoto.jpg
sidediagram.png

Azure is LOX
Green is LPG
Orange is combustion
Gray is co-rotating aluminum structure
26 are the injectors
16 are CuW de Laval nozzle inserts are dotted outline (embedded in the aluminum)

The inlet boundary conditions set on the left of the aluminum structure would be for LOX and LPG.
The outlet boundary conditions set on the right would exit de Laval expansion into vacuum.
The combustion chamber (orange) heat input would, at first, be equivalent to combustion at the outer circumference, or (later) from full-on combustion simulation.

Outlet view equivalent to Fig 3.

bottomphoto.jpg

64 are de Laval nozzles, canted to provide torque to the centrifugal pump.

bottomdiagram.jpg

One non-obvious aspect of this design is the way heat transfers from the combustion chamber to the "unreasonably" low number of cooling channels that double as impellers:

The enormous g-forces and Coriolis forces in the impellers increase the liquid velocities at vastly higher rates than 1D flow thermal transfer would indicate. That plus the fact that aluminum has a very high thermal conductivity yielded surprisingly few cooling channels. However, there is no particular reason the number of cooling channels can't be increased.

Other problematic (or non-obvious) aspects are:

• the degree to which the design tries to avoid an inducer by using the coaxial feeds to pre-swirl
• the pressure required to avoid cavitation on entry to the coaxial feeds (in the absence of an inducer)
• whether the combustion chamber's Coriolis and convection forces are adequate to avert instabilities
• what chamber pressure (hence temperature) causes a phase change in the cooling channels (which dramatically lowers their cooling rate)

We had a few prototypes fabricated in an automotive machine shop by a guy who designed and fabricated custom equipment for silicon fabs. Texas A&M ran dry rotational stability tests on one of them.