Regenerative Engine
Once I understood the breadth of the resources necessary to build an orbital rocket (an entire team, millions of dollars, etc.), I decided to recalibrate my goal to develop and fire a liquid rocket engine by the end of high school. After diving headfirst into liquid rocket engines, I made a number of mistakes, but learned so much from each one.
My first liquid injector prototype was intended to be a showerhead-type spray injector. I crudely drilled many 2.5mm holes into a polished steel pipe end cap. In hindsight, this injector (pictured below) was deeply flawed. As it was my first attempt, I used very little math and approached it as an experiment.
At the time, I was going to use this simple injector with the hybrid nozzle (previously shown). I never intended to fire it in that configuration. Although I knew very little about liquid rocket engines, I did know it wouldn’t work with that injector. I started researching injector geometries and educating myself about orifice diameter, spray type, pressure drops, pintles, swirlers, like-impinging, impinging triplet pattern, etc. The volume of information available in this area of liquid rockets was daunting. I worked my way through a significant amount of data and decided to make a swirl injector, specifically a coaxial swirl injector.
After designing the first test coaxial swirl injector in Fusion 360, I printed it, and performed a water flow test. Unsurprisingly, it didn’t work. It atomized minimally and fanned radially too much. The pressure-drop calculations proved to be too low. The design, test, and iterate process repeated until the successful 6th iteration of the swirler. After a successful flow test, I could fire it with isopropyl alcohol to test combustion.
Images: Coaxial Swirl injector V0.1, July 7, 2020, swirl injector V0.1 Flow test July 11, 2020, Coaxial Swirl Injector V0.6, September 11, 2020
Image: Coaxial Swirl Injector combustion test V0.6, September 28, 2020, Improved Plumbing, Oct 4, 2020, and Assembled liquid engine V1, Oct 6, 2020
After improving my test stand infrastructure, mainly the fuel tank plumbing, I mounted the coaxial swirl injector v0.6 to the hybrid nozzle to act as a combustion chamber to see if it could work. I discovered there was no way this could work due to a lack of oxygen flow; however, it was a good learning experience.
After four attempts at firing, the engine didn’t ignite. There were too many issues with the setup. I was using compressed air, not pure oxygen; the combustion chamber was too small; tank pressures were too low; solenoid valves leaked; the control system was manual, including ignition; and the igniter was inconsistent. I eventually addressed all of these problems and successfully scaled up.
Image: Liquid engine V1 on a crude test stand, Oct 8, 2020, Crude test setup V1, Oct 8, 2020, New swirl CAD design, Oct 26, 2020
The engine I began designing, specifically this with this injector, was regeneratively cooled and will be 3D printed out of 304L stainless steel. I also switched from barb fittings to compression tube fittings to deal with the high pressures this engine would experience.
I continued to read and apply what I learned to many iterations, ending up with R7 v1.1. I was confident with this design but incessantly reviewed and modified parts for a number of months.
Image: Fully assembled R7 v1.1 render, and R7 v1.1 chamber render Jan 21, 2021
During the down time between design and printing, I honed my CAD and design skills enough to do a rough design from scratch of a 15kN (kilonewton) or 3500-pound force engine called Asteris 1B. It is a kerolox (kerosene and liquid oxygen) rocket engine driven by electric turbopumps. This was an extremely rewarding process as I began with novice CAD abilities and through this design process excelled more than anticipated.
Image: Asteris 1B engine render April 1, 2021, 9 Asteris 1B render April 1, 2021, Uplink 2 engine thrust structure render April 2, 2021
While reviewing the calculations for the R7 engine I discovered several major flaws including chamber length and nozzle ratio. I scrapped it entirely and found a less expensive way to manufacture the engine. It would regeneratively cool the nozzle of the engine since it would be experiencing the most thermal load, allowing me to 3D print just the nozzle. This significant reduction in cost gave me the flexibility to scale up the size of the engine. It went from producing 100 pound-force of thrust to upwards of 800 pounds of force if everything worked. The engine would consist of three main parts: a 3D printed nozzle; a steel chamber with custom milled flanges; and the injector head. This new engine would be named Asteris 1A. It runs off ethanol and lox or gaseous oxygen.
Image: Asteris 1A first render April 25, 2021, Asteris 1A Thrust sled mount renders April 30, 2021
After final evaluations, I sent the nozzle design to get printed out of a steel copper alloy. Two weeks later, I had my first metal 3D printed part. I machined down the top flange to be flat, milled the o-ring channel, polished the inside of it, and tapped the holes for the compression fittings.
Image: Post-processed Asteris 1A engine nozzle, May 19, 2021
My work went from design to hardware quickly. First, I began accumulating cryogenic plumbing hardware and parts for the custom digitally actuated ball valves I had designed. Next, I started assembling the test stand by welding together the main steel base of the stand.
Simultaneously, I was developing a computer application in Java to control the engine computer. It acts as a bridge between the hardware and the software. This is extremely important since it controls ignition, main propellant valves, pressurization valves, and all the data collected while testing. The GUI development proved to be a challenge; however, I was able to get the two systems communicating through serial communication. The main engine computer is an Arduino Mega, a high-powered microcontroller board. Once I pressed the fire button after arming, the computer would go into an auto-ignition sequence and take complete control, monitoring tank pressures. It would abort if it saw a drop or an anomaly out of nominal range.
Image: Mission control GUI June 18, 2021, Final Asteris 1A design, June 25, 2021
I finalized the engine's design and the rail system the engine sits on so it can move freely to record the thrust data. The engine's final estimated thrust at optimal performance would be 350 pounds of force, with a specific impulse or Isp of approximately 230s.
The valve actuation system I designed is a high-powered stepper motor with a 15:1 orbital gear reduction to give it the torque required to open the cryogenic Swagelok valves. I built two of them, for the fuel and the oxygen sides. At this point, I had not received the injector elements or completed milling the injector manifold or chamber, so I 3D printed a mock-up for fit checks.
Images: Valve assembly and engine mockup with nozzle, June 29, 2021, plumbing fit checks, July 2, 2021
I continued testing all the subsystems of the stand, valve tests, pressure tests, tank qualification, igniter tests, etc. The plumbing setup is a simple blowdown setup with the fuel tank being pressurized by gaseous nitrogen and the gaseous oxygen being its pressurant. Finally, I finished milling the injector and welding the chamber for my first fire-ready liquid bi-propellant rocket engine.
Images: Test stand assembly with engine fit checks, July 5, 2021, Water flow test July 12 2021
Image: Asteris 1A fully assembled, August 2, 2021, Engine igniter test, August 7, 2021
I ran injector igniter tests to dial in the timing of all the valves and ignition to avoid a hard start, which is where un-ignited fuel builds up in the chamber and explodes. The igniter is an electric sparkler that can be lit using the engine computer. All was in place for a hot fire static test of the engine. Next, I set up a camera mount on the test stand that I dubbed “danger cam” since it would be in the direct line of the engine fire.
Image: Danger cams view, August 20, 2021
After struggling to secure a location to fire the engine, my high school principal graciously permitted me to fire it after a detailed safety review in the junior parking lot. Finally, on Saturday November 11, I packed everything up and proceeded to the test site. The stand was weighed down with 400 pounds of sandbags. I installed a load cell to measure the thrust produced right behind the engine.
I powered up the engine computer and data collection computer and armed the igniter. The countdown proceeded nominally. The igniter lit, and then it fired. A huge fireball blasted 25 feet out with a bright roar. It was quite something. I saved the test stand by purging all systems with nitrogen, disabling valves, and shutting down the system. Because of excess ethanol leakage, there was a small fire on the stand, which I proceeded to put out with the fire extinguisher. After reviewing the data, I discovered the oxygen valve was lodged halfway open, throwing off the mixture ratios and resulting in the fuel-rich burn. However, it produced upward of 210 pounds of force when working correctly. The following are photos from that night.
Image: Test site setup, September 11, 2021, Me adjusting danger cam, September 11, 2021
Image: Firing, September 11, 2021, Post-fire, September 11, 2021
Twitter rocket fire.mp4Hot fire video
After a more detailed inspection, I discovered slight melting occurred in the chamber, but the injector and regeneratively cooled nozzle remained remarkably cool and sustained minimal damage. Further reviewing test data and hardware allowed me to generate a list of improvements and modifications to build the next stand and engine generation. Some critical list items included: Switching the entire valve actuation system over to pneumatic with a fail shut state; secure and adequately mount plumbing; improve independent redundant main propellant valves; upgrade safety and failsafe mechanisms ranging from burst disks to check valves; overhaul control GUI; create more stable and robust sensor data channels; independent purge lines for both lox and fuel; better abort capabilities; feedback loops watching for abort scenarios and many, many more.
Image: Injector post-fire, September 12, 2021, Post-fire chamber and nozzle, September 12, 2021