It’s rocket science
Using Algorithmic Engineering to design and 3D-print the Aerospike
Published 1 August 2022

Project Breakdown

Industry: Space Technology

Product: Aerospike Rocket Engine

From Injector Head to Rocket Engine

In the last case study, we gave a glimpse into how we designed and 3D-printed a co-axial swirl injector head. It was framed as one part of an entire modular rocket engine that we had engineered algorithmically. Today, we are drawing back the curtains on the developmental process behind this very piece: Hyperganic’s Aerospike rocket engine.

Rather than just one design, what we have created is a parameterized set of algorithms that describes how to construct such engines. During its prototype iteration, we generated hundreds of viable variants and some were selected to be 3D-printed. One of them marks the most complex 3D-printed object to date and the first algorithmically designed rocket engine to exist. Standing at 80 cm tall, we printed the engine from copper alloy together with AMCM. It was then accompanied by a 40 cm version printed from Inconel-718, in lieu of our new collaboration with EOS on space propulsion hardware. Skirting the line between science fiction and reality, it obliterates the limits of what can be actualized with Additive Manufacturing when empowered by Algorithmic Engineering.

80 cm tall Aerospike engine printed by AMCM from copper alloy, as seen during the depowdering process. (Image source: AMCM)

40 cm tall Aerospike engine printed from Inconel-718 by EOS . (Image source: EOS)

The Aerospike Concept

The concept of the Aerospike engine is not new. Having existed since the 60s, it carries high interest value from a theoretical standpoint: its altitude-compensating abilities promise a performance advantage over traditional bell nozzles of up to 20 percent during atmospheric ascent.

With that in mind, prototypes were built by Rocketdyne and NASA but historically never left the test stands. The Space Shuttle was meant to be propelled into Single-Stage-To-Orbit (SSTO) by an Aerospike engine, but it was never realized. All this ties back to how the Aerospike concept posed thermal challenges that engineers could not surmount readily with traditional design and manufacturing methods.

Firstly, the central spike that facilitates supersonic exhaust gas expansion is subject to overheating and meltdown if not cooled sufficiently. Secondly, the compact design of an Aerospike, though ideal for small orbital launchers and in-orbit propulsion, increases the heat load even more.

Altitude Compensation – at low altitudes, the exhaust plume is rather narrow. With decreasing ambient pressure during atmospheric ascent, it widens significantly. Traditional bell nozzles are optimized for a specific altitude (mainly sea-level or vacuum) and restrict this gas expansion in every other condition. However, under-expanded exhausts cause a performance hit. The Aerospike engine is supposed to keep the plume in an optimal shape at all altitudes and pressure regimes for best performance.

With modern 3D-printing methods, it’s now possible to build regenerative cooling channels directly into the manufactured part. This gives us a more contained and integrated solution to overheating. But for it to be conceptualized by engineers and therefore, tapped into the design freedom granted by Additive Manufacturing, we have to radically change the underlying product engineering paradigm.

Algorithmic Engineering in Space Propulsion

At Hyperganic, we script rule-based code that generates physical objects automatically according to formulaic principles. Think of it as a list of step-by-step instructions for the computer to mimic an engineer’s process of execution. The difference is that a computer can work tirelessly to solve a problem at a level of detail that a human engineer cannot.

All iterations of Hyperganic’s Aerospike engines are generated through the same algorithm — just with different input parameters to meet other specifications. Each physical component within the engine is represented by a code module: one for the injector head, one for the combustion chamber, one for the flange etc. In other words, both the program and the physical object follow the same architecture. Object-Oriented Programming (OOP) methods allow us to assign attributes and methods to each class of code. The information is synced between components such that when one of them changes, it will be updated throughout the entire assembly to produce a valid design every time.

Program Structure – for each physical component in the final object, there is an analogous software module handling its construction. Its connective architecture allows information flow between the modules. For instance, the chambers adjust to the data they receive from the spike; the cooling channels take input from the combustion chamber module, and so on.

In software, such modular approach is commonplace. As long as one complies with the specified interface, the workflow is easily extendable. This method also helps to break down the complexity of a problem into parts that can be workshopped. Often, a module can be treated as a placeholder that returns simplified values until the engineer finds the time and inspiration to improve on it. Currently, our Aerospike model has been strengthened to produce functionally sound results from a multitude of rocket equations and analytical formulas. And still, it can be easily expanded and refined.

Our Aerospikes are firmly grounded in the capabilities of 3D printing and boast design qualities enabled by Algorithmic Engineering. The two regenerative cooling circuits we have integrated into its structure exemplify this. Watch the animation below to visualize how the cryogenic cooling liquids run through the channels.

Animation of the cooling sequence:
Scene 1: Liquid Oxygen cycle that cools the spike interior before being injected for combustion.
Scene 2: Cryogenic Fuel cycles that cool all the combustion chambers individually.
Scene 3: Fuel and Oxidizer are injected into the combustion chamber through the injector head before they combust, accelerate and exit the chamber at the throat.

The use of multiple combustion chambers enables thrust-vectoring. Individual chambers can be throttled differently to steer the rocket without excessive, heavy gimbal mechanics. The structures of the chambers are driven by a desired mach number distribution and calculated based on isentropic flow conditions. This is just one instance of the many physics principles that inform the construction logic by which the object is built up. It allows us to pack rich engineering knowledge into the creation process and the resultant code.

Thrust Vectoring – the Aerospike consists of multiple combustion chambers that can be throttled differently. This shifts the angle of the exhaust stream slightly and changes the direction of the spacecraft.

Regenerative cooling circuits are automatically routed across the double-curved walls of individual combustion chambers. You may think of the cooling channels as paths created by little bots that crawl along the surface, dictated by rules the engineer specifies (e.g. constant distance, swirl angle over height, channel cross-sectional area and shape over height). This method allows us to map a wide spectrum of cooling layouts across the chambers that could not be achieved with conventional methods in CAD. The design intricacies of meandering cooling channels extend to the central spike where thin, radial cooling rips are employed in combination with a bio-inspired differential growth pattern that increases the surface area for heat exchange.

Regenerative cooling channels of the combustion chambers.

The injector head has also been empowered by Algorithmic Engineering in adapting to varied temperatures. It consists of mass-customized co-axial swirl injector elements, each optimized based on its location within the power head for a given distribution of temperature during combustion. This makes it easy for engineers to tweak the mixing ratio of injector elements positioned close to the chamber walls to reign in high temperatures. If you want to dive deeper into the injector module, check out our previous case study and see it being printed in titanium.

Tight Integration of the Print Process

Here at Hyperganic, our objects approach high levels of complexity but are anchored within reason of production. The algorithms should never lose sight of a printer’s constraints so that we don’t end up with something that cannot be manufactured on a respective machine.

In our case, all manifolds and internal routings adapt their shape according to the local overhang angle to ensure printability. Their cross-sections smoothly transition from a circular shape when vertical, to a rhombic shape when horizontal. Such considerations are what made materializing our Aerospike engine with the printing technology of Titans of CNC and EOS possible.

Fast Iteration

Algorithmic Engineering allows for quick iteration cycles within seconds and minutes, as opposed to weeks or months. This way, engineers can explore a much larger solution space and focus on system level optimizations rather than spending their time on manual, repetitive CAD drawings.

Each of the Aerospikes below took just a handful of minutes to generate from scratch. Over the course of a weekend, one can effortlessly generate hundreds of variants. The task of an engineer no longer lies in finding the best design in the first attempt. It’s about asking the right questions to find the best solution out of many opportunities.

Algorithmic Variations – these Aerospike variations were automatically generated over the course of a weekend. Each of these geometries is roughly 30 cm tall and took about 10 minutes of program runtime. All of them were generated using the same construction logic. Yet, by applying a different set of input parameters every time, the resulting geometries look notably different.

“They built an aerospike vending machine. Push some buttons, get an engine. Different every time. That’s an incredibly powerful capability, and also one that demonstrates how the future of additive manufacturing may unfold. Imagine a world where every object is, at its core, an algorithm. Bolts, furniture, appliances, rocket engines, or whatever, are all systems where a request with some parameters generates the precise object required.”

Kerry Stevenson, Founder and Editor, Fabbaloo

The Future of Spaceflight

Algorithmic Engineering is a critical enabler of ever-more complex geometries that are perfectly suited for Additive Manufacturing and fast iteration cycles. Being able to fail fast and learn fast within an engineering sandbox will spark creativity and technical innovation. In particular, novel propulsion systems like the Aerospike concept become possible.

Rocket engines are by far the most expensive parts in an entire launch system, as development consumes copious amounts of time, cost and human labor. By allowing engineers to build on top of existing software modules, Hyperganic’s platform approach means that they never need to fully reinvent the wheel. As a consequence of dramatic design acceleration, development cycles will run more cost-effectively. This unlocks affordable access to space and high frequencies of testing and launching.

Right now, we are onboarding engineers from all over the world onto the Hyperganic platform to supercharge their design capabilities and enable new innovative solutions.

Check out our Forum posts to learn more about how the Aerospike was constructed in Hyperganic Core:

Don’t have a Hyperganic account yet? Sign up here:

Featured media articles for further reading:

Next Button
We Are Hiring!
Visit us at our here