October 2023

Prototype LOX Pump Testing with Water

Test Goals and Summary

My goals for this hardware were mainly that I wanted to exceed the performance (speed, pressure, etc) from my previous pump experiments, and also improve sealing and other internal workings of the pump. In terms of performance, in test 4.5, the pump was able to reach a maximum speed of 18,124 RPM and maximum discharge pressure of 197psia with a pressure rise across the pump of 160psid before stopping due to a mosfet overtemp abort. The sealing was also improved compared to the pump from march although, notably, water and air still perspire through the porous plastic especially at high pressure. The new seal package is somewhat complicated, but essentially is comprised of a series of PTFE o-rings that limit leakage flow while minimizing friction. A compressed air purge and deliberate leak path is then used to prevent the pumped liquid from reaching the motor.

Performance and Efficiency

My original Pump efficiency assumption was a measly 19%. I arrived at this number based on data from my previous pump, and a chart I found in a text book with predicted efficiency vs specific speed and flow rate. I used this number approximately to size the motor and predict current draw etc. Fortunately it appears that this was a very conservative assumption. During test 4.5, the average battery current when the pump was spinning at its highest speed was 31.4A. Using the torque constant of the motor and the shaft speed, I estimate the motor mechanical output to be about 530W. Since the discharge of the pump flows through a characterized orifice, I can calculated the flowrate through it for a given pressure, and then find the fluid power. Fluid power comes out to be 290.7W indicating that the pump efficiency is around 55% which is much higher than I anticipated. I will be looking into this deeper, but I believe I misinterpreted the units of specific speed in the chart I was using. Within metric specific speed, some use cubic meters per hour as opposed to cubic meters per second. Additionally, I had somewhat limited electrical data in my tests from march, which, in combination with the inferior design, probably skewed my efficiency prediction low.

Hardware Inspection

The pump hardware survived with basically no damage. The only odd thing I noticed was one of the PTFE o-rings had signs of wear and each of the o-rings seemed to be more loose in their glands so perhaps the frictional heating of the o-rings affected the plastic housing some small amount.

Next Steps

The main thing limiting me from reaching full power is a thermal issue in the motor controller. Since the motor is un-sensored, the duty cycle must be increased relatively smoothly and slowly unlike the fuel pump which is capable of reaching full power in less than a second. This lengthens the total amount of time that the motor must be running to achieve a given speed during one test. The extended duration of the test allows the controller to heat to unacceptable levels unless it is externally cooled. I will be looking into a cooling solution for the controller. I am also looking for an affordable, trustworthy flow rate measurement method to corroborate the orifice calculation. The fuel pump flow rate could be approximately known based on the shaft speed, but this is not a viable method for centrifugal pumps.

The Bandit engine LOX pump is a single-stage electric centrifugal pump. Developing this pump will probably be the most challenging aspect of the engine project as a whole. Due to my limited budget, my strategy for developing this pump is to initially start with a prototype that uses PLA plastic as a substitute for metal for some specific parts. I will iterate and replace those parts with metal versions as the hardware matures through test. There are 2 main parts of this pump that are designed to be additively manufactured, the impeller and the main housing. Due to its small size, the impeller should be relatively affordable to print in metal. The housing on the other hand will be pretty expensive and I will want to be as confident as possible in its design before ordering one. Once a fully metal assembly is ready, I can begin testing with LN2 and LOX.

I think people underestimate what can be accomplished with plastic. In these tests conducted on 10/21/23 the rotor speed exceeded 18,000 RPM and the pump discharge pressure reached 197psia. I anticipate being able to reach full power using this hybrid metal-plastic assembly.