Challenge
NASA’s Artemis mission aims to put humans back on the moon, exploring the possibilities of permanent settlements for the future there and, one day, on Mars. For all this to happen, NASA engineers must scrutinise every part on a spacecraft to minimise space and weight, as well as survive the extreme temperatures of the Moon and beyond.
Solution
An experiment conducted at a recent conference challenged engineers to develop a list of geometries and qualities a specific part would need to survive the flight and tolerate extreme conditions on the surface of the moon. These were input into a generative design framework that allowed artificial intelligence to develop a part that met those criteria.
Outcome
One of the advantages of generative design is the ability to see things beyond the human mind’s eye—with a detached sensibility—creating the best solution to a set of parameters. It changes the entirety of your workflow by making rational design decisions incredibly quickly. To that end, what surprised everyone was the speed—less than 36 hours elapsed to go from determining constraints and uploading to the Protolabs digital thread to a machined part.
“For this event to be successful, we knew that any manufacturer we went with [to make the part], needed to deliver the part we ordered on-time. There was no room for error or delay.”
- Matthew Vaerewyck, mechanical engineer at NASA Goddard Space Flight Center.
From Part to Partnership
The PowerSource Global Summit is a technology conference that brings together government, private industry, universities, and more to exchange information, ideas, and best practices.
Attendees came together to identify constraints for the generatively designed part, which acts as an apparatus for holding the equivalent of an upside-down 250mL Ehrlenmeyer flask to capture samples of volatile gases released when the moon warms.
Surviving Space Travel and Environments
Objects on the moon in the area Artemis will inhabit must tolerate temperatures that can range from -193 C to -48 C. According to Vaerewyck, the conference engineers had to ask themselves, ‘What does this part need to do and what environment does it need to survive.’ From there, they developed parameters for generating the design, with an eye on optimising weight and maximising stiffness.
Why worry so much about weight? NASA stats indicate that it costs $1 million for every kilogram (2.2 pounds) launched into space, so every weight reduction saves taxpayers money.
Whenever lightweighting comes up in conversation, affordable aluminium is the first material that comes to mind, thanks to its excellent strength-to-weight ratio. In this case, the CNC-machined part was made with a form of aluminium 6061 that has undergone processes to enhance its strength and stress relief capabilities. It’s a popular choice for aerospace, automotive, marine, and general manufacturing.
Some of the other key design parameters were that the part had to be able to carry 68 pounds of mass and needed to be stackable. The latter enables multiple copies of the part to take up less space on the spacecraft.
The part required footings that allow it to stand upright on the moon's uneven surface.
Finally, the engineers wanted a part that was quickly machinable. While technologies like 3D printing would offer more design freedom, metal 3D printing would not be possible in the 36-hour window to deliver the part.
Given all this data, the computer crunched the numbers. After considering all the constraints that engineers placed on the design, it came up with its best-scenario CAD file. “The awesome thing about it is that we didn’t know what we were going to make in the end,” said Vaerewyck. What they got was between 6-10x improvement in mass, stiffness, and design time.
It quickly moved through toolpathing: 19 hours of milling and 4 hours of processing. “You can’t underestimate the human element here, ensuring that there were no alarms or errors in the milling process,” said Greg Perz, Protolabs manufacturing engineer manager for CNC machining.
The generative-designed part has breaks in the circular design, shedding hours of milling time by allowing for simpler 2-axis milling.
“Our deadline was 5 p.m. and the first parts went out at 4:31. It was quick and painless,” said Perz. From the initial generative design brainstorm to having the part in hand took only 36 hours.
