A methane-GOx rotating detonation rocket engine, built around a variable impinging injector and an aerospike contour sized to its detonation cell width.
DEIMOS is PURPL's methane-GOx rotating detonation rocket engine — a continuously-rotating detonation wave replaces conventional deflagrative combustion, which in principle recovers pressure gain that a standard rocket chamber simply throws away. The catch is that almost nothing about a detonation chamber behaves like a standard one: injector response time, wall heat flux, and nozzle expansion all have to be re-derived around a wave that's circling the annulus at several kilohertz.
My work on the sub-team covered two of those pieces directly: the injector layout that has to survive and re-fill behind that wave, and the aerospike nozzle that has to expand a flow with a circumferentially-varying pressure field instead of a clean, uniform one.
The injector has to do two things that are somewhat in tension: meter propellant precisely enough to hold a stable mixture ratio, and re-fill the annulus fast enough to keep up with a detonation wave that returns to the same circumferential location dozens of times a millisecond. We used a variable impinging element layout — element spacing and impingement angle tuned around the annulus rather than held uniform — to manage local fill fraction and mixing quality without over- or under-feeding any one region of the chamber.
A conventional bell nozzle assumes a roughly uniform inlet condition. An RDE doesn't give you one — pressure and temperature vary around the annulus depending on where the detonation wave currently is. An aerospike geometry is far more tolerant of that nonuniformity and continues to expand reasonably well off-design, which made it the natural fit downstream of the chamber.
I led the contour design, sizing the spike against the chamber's predicted exit conditions and detonation cell width so the nozzle's expansion characteristics stayed matched to the engine actually feeding it, rather than to an idealized uniform inlet.
Injector and nozzle decisions were both made against predicted performance rather than intuition — fill-region length, expected wave speed, and exit conditions all came from the quasi-1D performance model I built in parallel (see the RDE performance toolbox case study). That let geometry choices on DEIMOS get checked against a model before committing to hardware, instead of the other way around.
DEIMOS and the design work behind it were published through AIAA Region III. The engine continues to move through PURPL's test campaign, with injector and nozzle iterations feeding back into the performance model as hotfire data comes in.
Hardware design complete; integration and test campaign ongoing with the PURPL DEIMOS sub-team.