Annealing 3D Prints: A Scientific Approach

We’ve all been taught the scientific method: Form a hypothesis, do some experiments, gather some data, and prove or disprove the hypothesis. But we don’t always do it. We will tweak our 3D prints a little bit and think we see an improvement (or not) and draw some conclusions without a lot of data. Not [Josef Prusa], though. His team printed 856 different parts from four different materials to generate data about how parts behaved when annealed. There’s a video to watch, below.

Annealing is the process of heating a part to cause its structure to reorganize. Of course, heated plastic has an annoying habit of deforming. However, it can also make the parts firmer and with less inner tension. Printed parts tend to have an amorphous molecular structure. That is to say, they have no organization at all. The temperature where the plastic becomes soft and able to reorganize is the glass transition temperature.

Common glass transition temperatures are 65C for PLA, 75C for PET or PETG, and 105C for ABS, in fact those are the minimum temperatures. Until the plastic melts, the temperature is technically a glass transition temperature. However, the hotter you get, the more likely the plastic is to warp, shrink or deform.

There’s a lot of data in the post about how different temperatures affect PLA, ABS, ASA, and PETG. They measured the change in dimensions, how much force it takes to break a part, and the resulting part’s tensile strength.

After reviewing the data, the post and the video talk about how you might do the same thing in a home electric oven. We had to wonder whether a convection oven would be better for eradicating hot spots.

This isn’t the first time we’ve seen this, but it might be the first one to use a few hundred test prints. Even so, the results are consistent with some we’ve seen in the past.

28 thoughts on “Annealing 3D Prints: A Scientific Approach

    1. A lot of “air fryers” maintain temp with pretty decent granularity already, and move the air around to eliminate hot spots! I grabbed one to cure pcb epoxy for a repair I needed to do. So the short answer is, these may be better platforms to start from….not that any of us on here need more projects! :)

      1. @Surprised I can still put any name in here!

        Yeah, right? We’re still open to anonymous comments. We’d like to think that what people have to say is at least as important as who they are!

        (Which is not to say that I don’t know your IP address, credit card details, and whether your socks match… OK, well one out of three ain’t bad.)

      2. Ah, mine was horrible. And from decades ago, used. So the little thermal dial bottomed out at 220F (in reality) and it had way more heating elements than desirable for it’s main applications. Made it more prone to overshoots.

        Both of which were fun to remedy! So I gutted it, cleaned it, rewired, grounded, and AVR/thermocouple controlled it too. It just needs more complete software now.

    1. In industry that’s what fixturing is for. Metallic parts bend, twist, and warp from residual stresses left by machining and welding if you anneal. It’s common place in industry to fixture the parts so they relax into the desired shape when heat treating.

      It’s likely you could print a fixture through support material with a higher glass transition temp than the PLA or PETG you’re annealing.

      You’ll still have deflection, but it should be significantly reduced.

      1. I’ve experimented with annealing PLA parts in sand. Basically pĺastic casting where the material is already in the mold and is just heated up.

        The idea was that the sand prevents the part from bending. But A: the play sand I used was not the ideal material for this. It melted into the plastic and I couldn’t get it unstuck again. Basically a sand paper texture.

        B: 3d printed objects tend to have hollow spaces inside and just crumble in on themselves in the mold. Printing at 100% infill helps somewhat, but does not solve the problem completely.

        1. If the issue is that the part is melting into the void inside of the parts then can you heat the part till its fluid and spin it like a slipmold. That way the plastic just moves to the sides of the part.

    2. There are build simulation programs for metal additive manufacturing such as Magics and Netfabb that can produce a model that is the inverse of deformation so when you print it comes close to nominal. If enough data was gathered on plastic annealing I would assume something similar could be done

  1. Can you get similar results by putting the parts in a vacuum sealed bag and putting them in boiling water, or in a sous-vide cooking pot? I tried the oven method once for restoring overly hydrated pla filament and my wife and kids were JUSTIFIABLY mad at me regarding the smell. I would also liketo know if it is viable for annealing?

  2. I’ve thought about, for prints requiring more strength, packing them in something like green sand then heating it all up….obviously 100% infill required, and it would take a lot of time, and I suspect the part would shrink….someone else want to try, or try something similar? :). I’ve already got too many projects!

    1. Oh yes. The guy has quite the ego. I met him once, and he didn’t want to talk to me because I didn’t recognize him.
      “If you don’t know who I am. You are not worth my time” kinda.

      The groupies don’t help of-course, they are to happy to praise him.

    2. I know right! I met Prusa once and his ego was so big I couldn’t fit in the same room. I would never buy one of his 3d-printed printers for the insane amount of money he is charging for them, and besides – even if you do buy one you can be waiting 6 months! thats 6 months he’s held onto your money and made interest on it, meanwhile you have no printer… Honestly do not know why people have their nose so far up his behind.

  3. There’s been some interesting research from the medical side about print strength after sterilization. High Pressure Processing, which unfortunately is out of reach for hobbyists, was beneficial. Autoclaves seemed to cause slight decrease in strength, but nonetheless it would be interesting to run prints through an instant pot.

  4. You should use box/dot-plots to display this type of data, or at least error bars. I am assuming that variation was significant because the parts are printed and its important to visually display difference of means in this specific type of mechanical test.

  5. PLA having a HIGHER heat deflection point than any of the other plastics mentioned after annealing is somewhat surprising, but only because the video makes a mistake: common 3D printing filaments are *almost* all amorphous. PLA is the exception: it is semicrystalline. The annealing process increases the crystallinity. Crystalline PLA is able to retain some stiffness well beyond the Tg of amorphous PLA. Very neat trick.

    I’d be interested in different annealling schedules. Instead of a simple single temperature, a very slow ramp in temperature (or holding at lower temperatures for longer periods before increasing to slightly higher temperatures) could allow PLA to achieve a similar Heat Distortion Temperature as the highest given here but without as much distortion.

    PLA really is fantastic. The Material Science gods were kind to us.

    1. I almost wish it wasn’t called annealing since in metal annealing softens the metal while tempering improves strength by making it less brittle.
      Also tempering chocolate changes its crystalline structure which increases its melting point, and is what gives chocolate that snap.
      Too late now.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.