AI-Assisted 3D Printing: Choosing the Right Material with ChatGPT

I found an interesting way to use ChatGPT to help me complete a 3D print project.

AI tools like ChatGPT are so new everyone is still exploring what might be done with them. In many cases quite amazing use cases are successfully identified, and there is no doubt there are plenty more.

My project was to 3D print some “skegs” for an inflatable kayak. Inflatable kayaks are an amazing innovation: they literally fit into a small backpack, yet expand into a full-size 14ft boat when set up. They say the best camera is the one you have with you, and the same can be said for inflatable boats.

One part of the setup is to install the “skegs”. These are stability fins that slide into receptacles on the bottom of the boat. The problem here was that the shipped skegs were pretty short and thus would not provide sufficient stability during windy conditions: the boat would turn around in the wind.

The solution was to simply use bigger skegs. These are available from the manufacturer, but why buy them when you can make your own? It even turned out that I didn’t have to 3D model them, as someone had already done so and provided the model free online.

The question was, what material should I use to print them? I have many different possibly appropriate materials available in the lab, but I needed to make a choice. Printing plastic dragons in PLA is an easy choice, as the major factor is color. But 3D printing a functional object such as a skeg would likely require a different material.

I’m not an engineer, so I wondered if ChatGPT could help. I began the discussion with:

“Of these materials, PLA, ASA, PA6, PETG, PA12, which would be best for a skeg attached to the bottom of an inflatable kayak?”

ChatGPT responded with quite an interesting set of data. It provided pros, cons and a conclusion for each of the proposed materials for this specific application. It did not just dump out generic material properties. For example:

1.  PLA (Polylactic Acid):
•   Pros: Easy to print, good surface finish.
•   Cons: Brittle, poor impact resistance, degrades in water over time.
•   Conclusion: Not suitable for a skeg due to low durability and poor water resistance.
2.  ASA (Acrylonitrile Styrene Acrylate):
•   Pros: Good UV resistance, impact resistance, and weather resistance.
•   Cons: Slightly harder to print than PLA, can be more expensive.
•   Conclusion: A good choice due to its weather and UV resistance, suitable for outdoor and water applications.

And so on. ChatGPT’s conclusion:

“Given the requirements for a skeg on an inflatable kayak, ASA and PA12 are the best choices due to their water resistance, impact resistance, and durability in outdoor conditions. If UV resistance is a critical factor, ASA might be slightly better due to its excellent weather and UV resistance. PA12, with its low water absorption and good impact resistance, is also an excellent choice, especially for its durability in water.”

I discussed several variations, including the use of carbon fiber and glass fiber filaments, both of which I had handy in the lab. ChatGPT concluded:

“For a skeg on an inflatable kayak, PA12-CF would likely provide superior performance due to its enhanced strength and wear resistance, assuming you can manage the more challenging processing requirements. If you need a balance of rigidity, impact resistance, and easier processing, PA12-GF is also an excellent choice.”

In the end I used Fiberlogy’s excellent PA12-GF, which seemed to work very well. The more important matter was that ChatGPT was able to help me with my material choice — and this is without any specific training on materials. I didn’t even say something like “Act as a materials engineer specializing in additive manufacturing”.

My application was pretty simple, but the implications are clear: AI tools can provide some useful help to those who now have 3D printers capable of printing unusual materials.

If you’re facing a material choice question, why not give ChatGPT a try? People get it to make recipes based on a photo of their fridge contents, so why can’t it do the same with your 3D print materials?