I Designed an Aerospike Rocket Engine - Is it any Better?

Aerospike rocket engine. Is it any better than a bell nozzle? A couple of months ago, I built my first fully functional rocket engine that ran on compressed air and propane. It made supersonic exhaust and was able to demonstrate the function of a basic bell nozzle, but it had one critical issue. As it ran out of air, it would suffer from over-expansion and flow separation, meaning its efficiency was pretty terrible. Thinking of ways that I could improve this, one option would be to increase the tank pressure and use a regulator, which would be dangerous and very expensive. I'd also need a gigantic regulator to deal with the ridiculously high flow rates in the feed plumbing. Another idea would be to optimize the nozzle for a wider range of expansion ratios. This could be done with a special geometry known as an aerospike nozzle. This particular nozzle can operate at almost any pressure ratio, as long as the chamber pressure is high enough to produce supersonic flow, about 28 psi or 2 bar of relative pressure at sea level. Aerospike nozzles work like an inside-out bell nozzle, where the accelerating gases are held to the wall of the spike by atmospheric pressure. If the feed pressure-atmospheric pressure were to change, this would have little impact on the expansion efficiency because conditions such as flow separation are nearly impossible with the right nozzle geometry. Aerospike nozzles still have problems of their own, though, primarily in the dissipation of heat. The high surface area of the design means that the materials need to withstand extreme temperatures and are limited to short-duration firings. I started designing something that I would be able to produce myself in hopes that I could achieve the required tolerances on my homemade machines. Some of the operations also required committing the ultimate machinist crime, using my lathe as a milling machine. Predictably, this operation failed in the worst way possible, ruining both the part and my little homemade lathe. This was one of the many reasons why I had to build a new lathe, although I already planned to invest in a better machine. With machining my own parts not being an option until I built the new lathe, I was kind of out of luck with this project, not being able to move forward. After sitting on this project for a while, a company called Husky Machining reached out to me and were able to offer their services to machine the nozzle. The 5-axis mill meant the nozzle could have complex geometries with tolerances that I could only dream of achieving. I designed an idealized nozzle with the identical throat area as the previous one that would easily fit into the existing combustion chamber. It would be machined out of stainless steel so it could survive the full test duration without being damaged by the heat. When the parts arrived, they were beautifully finished with all dimensions being extremely precise. The parts fit well together and were attached with number 4 machine screws. Before I installed the new nozzle, I needed to get some baseline test data for the old nozzle. So I dusted all the spiderwebs off my test stand and made sure it still worked. Supposed to go on there. This other one popped off too because this hasn't been used in a couple of months. I need to retighten all the hose connections. Aside from a few loose hoses blowing off and the ignition being a bit inconsistent, it still worked reasonably well and I could run the setup the same way as I did a couple months ago. The old nozzle achieved the same performance as before with around 60 seconds of specific impulse. I removed the old nozzle and slid in the new nozzle with a supposedly high temperature gasket instead of RTV silicone. Once it was back on the stand, it was pretty much ready to test. However, before testing with fuel, I needed to get an idea of how the gases would expand in the new nozzle. I used an imaging technique called Schlieren imaging to see the variations in air density within the exhaust and where shockwaves can be seen along the aerospike. The shockwave pattern that can be observed appeared consistent with most sea level aerospikes and had the typical alternating expansion and compression waves. I also gathered the cold thrust data for these tests for an impulse comparison later. With the cold testing done, it was now time for fuel. Like last time, I used gaseous propane in a fuel-rich mixture to help flame stability and slightly lower the exhaust temperature. Running stoichiometric would have some performance gains but almost certainly destroy the uncooled aerospike. Now that everything is ready to go, it was time to put this aerospike nozzle to the tests. This engine appeared quite spectacular throughout testing, producing toroidal mock rings, just like some large-scale aerospikes. How did this nozzle perform? From initial observations, it appeared that it had 29% more peak thrust but it also ran out of propellant a little bit faster. This is quite interesting because with the same throat area, they should have had roughly the same rate of propellant consumption. The difference is likely due to the aerospike having a higher coefficient of discharge over its annular throat as opposed to the original circular throat. Just like before, the hot gas tests were performance-wise much better than the cold gas tests in terms of both thrust output and efficiency. This high peak thrust is very promising for the engine's performance because its firing duration wasn't that much shorter than with the bell nozzle. Now looking at efficiency, I could calculate the average specific impulse just like with the original rocket engine, and by doing that, I got about 64.9 seconds, or an 8.5% improvement over the previous design. This is a significant improvement, and the thrust gains would make it even better suited for flight. A possibly better approach to measure efficiency uses the pressure data to get an instantaneous specific impulse and see how quickly the efficiency changes over time. I used the tank pressure to find out how quickly the air mass changed and was able to calculate the mass flow of the engine throughout its test. Adding the constant fuel flow rate and using the instantaneous area under the thrust curve, I got this messy looking graph of instantaneous specific impulse. Comparing the aerospike to the bell nozzle, the peak impulse of both is much higher than the average and decreases as the pressure drops. Considering how noisy the graphs are, I can speculate that the peak efficiency of the old nozzle was likely around 80 seconds, with the aerospike possibly peaking as high as 95-ish seconds. However, the analog nature of my pressure readouts makes it extremely difficult to accurately determine this peak value. This could be solved in future iterations by incorporating an accurate pressure transducer and a load cell into the same data logging system to avoid some of the errors in analog gauge reading. The engine also had some interesting soot deposits after firing, being left over from excess propane, and they happened to line up with the mock diamond patterns. It was easily wiped away revealing some colored oxide film on the stainless below. Overall, this aerospike nozzle did see a noticeable improvement over the conventional bell nozzle with this particular testing setup. This could possibly transfer to a flight-worthy setup, but weight, simplicity, and cooling are factors that the bell nozzle will always be better in. The best type of nozzle really comes down to the intended application, where the extra performance could outweigh the drawbacks. One area of propulsion that is particularly well suited for aerospike nozzles is rotating detonation engines. The annular detonation chamber means that an aerospike nozzle is the obvious choice, with good performance at any altitude making it that much more efficient. Expander cycle engines may also benefit from the additional heating because they can boil more propellant to drive the turbopumps. My extremely simple pressure-fed aerospike is a good proof of concept and shows the experimental process of hobbyist rocketry. Eventually making something flight-worthy needs that high thrust as well as the better efficiency. I think it's quite possible to design a light enough version of this engine to be able to lift a small rocket someday. Even though aerospike nozzles aren't some magical new propulsion system, they definitely are promising at making more efficient and versatile launch vehicles as long as the drawbacks are minimized. This is a very fun and exciting project that ended up being another great research and learning opportunity. Let me know if you want to see any other interesting nozzles or rocketry concepts tested with the setup.