Annealing printed parts directly in the Ultimaker?

[geert_2 558]

In another post user Artiz mentions as a side note that he anneals his polycarbonate prints by moving the glass plate directly from the printer into an oven, and let it sit there at elevated temperature for a couple of hours, to relieve the stress due to uneven cooling caused by the printing process.

This seems like a very good idea. The advantage of doing this on the glass plate, immediately after printing (thus without the glass cooling down) is that the models do not warp as long as they sticks well to the glass.

I have done annealing in an oven too, but only after removing the parts. In which case they tend to warp in the oven indeed. I did not do this to relieve stress, but rather to increase temperature resistance a bit (which seems to help indeed, although it does no miracles).

But not everyone has a suitable oven, and not everyone likes to remove the glass plate from the printer every time.

So I just got this idea: why not anneal the parts directly in the Ultimaker itself?

After the print finishes, as soon as the build plate has lowered, immediately set the build plate temperature to the desired level (depending on the material). And keep it there for a few hours. Also immediately put a little plastic box, or cardboard or isomo box, on top of the model, so the heat is contained well. This should be easy to do, should increase temperature resistance of the parts, and should avoid or at least reduce warping. Might be worth experimenting with?

I haven't tried it myself yet, but I thought I would let you know the idea already. Otherwise I might forget.

[SandervG 1,521]

Hi Geert, thank you for your post.

Great that you are sharing ideas, that is what everything starts with!

I can't say I heard about this technique before, so while reading I had some questions.

If I understand it correctly, the idea is among others to reduce warping, right? Warping that occurred during the print? I was wondering if that wouldn't require any physical force to 'deform' your model back in its original shape? And wouldn't the entire model get soft and potentially loose some of its dimensional accuracy?

I think not being able to use your Ultimaker for a few hours is an appealing solution to many users, except if it indeed solves some problems and you don't have a regular over available.

Could you explain the affect on uneven cooling and how that affects the strength of your print? Does it result in unequal (and bad?) layer-adhesion? And can this be fixed afterwards by this method?

Curious to hear your thoughts on this!

[geert_2 558]

The idea is not to reduce warping while printing, but to reduce internal stresses that would cause problems later on during the part's service life.

A bit more about the background: when any parts are moulded in metal, glass or plastic (for example via injection moulding), they have huge amounts of internal stress moulded in due to the force of the flow that was used, and due to the always uneven cooling that follows. Edges and thin areas cool faster than the inner parts and big volumes.

If you watch transparant parts of glass or plastic under polarised light, you will see al those nice color changes due to these internal stresses. If you bend the part, you see the colors changing in the areas of stress. (I 'll see if I can find a few pictures I made years ago.)

These stresses will later on cause fatigue effects such as cracks or deformation of the part, although it may take months to show up. I have seen polycarbonate parts with fine details starting to totally crack after a year. If you leave such a part in a hot car, it is likely to deform due to the part getting closer to its melting temperature, and those stresses acting up.

To relieve these stresses, after production the parts are subjected to an elevated temperature for several hours, or sometimes even for days (especially for glass lenses). This elevated temperature is around the glass transition temperature, so that the molecules have a little bit of freedom to move, but not too much. Then they will slowly "relax" and remove stress.

When I do that with 3D-printed parts in a normal oven (I have a lab oven adjustable from room temp to +100°C), the parts will warp during the relaxation of the stress. So I need to put a weight on it, or clamp them while annealing.

However, when printed freshly, the parts are still firmly "glued" to the glass plate of the printer. So there is no need to clamp them to avoid deformation, as they can't move anyway if the bonding was good. That is the beauty of Artiz' idea to move the whole glass plate into an oven *immediately*, before it cools down and the part falls off. So he can anneal it without corners lifting.

We could use this same idea, but do the annealing in Ultimaker by using its heated build plate as an oven.

To reduce heat loss, and to get an even heating, the model has to be covered. That could easily be achieved by putting a box over it, or isomo, or something similar, if the model is not too big. I would recommend only covering the model on the glass plate with a box that is a little bit bigger than the model. Not the whole printer, so that the electronics do not get subjected to the heat longer than necessary.

In case of printing with PLA, annealing the parts may increase temperature resistance with 10 to 15°C, which may be enough for some applications. So they experience less warping when you use them in your car, for example. I have tried that: it did not eliminate warping in the sun, but it reduced it by at least 70% for my parts, which was enough. And we had a couple of very hot days this summer.  

Practically, I think a good starting point to test this concept would be to use the same bed temperature for the annealing, as used for printing that part. And then see if it sags or not, and adjust temperature accordingly for the next parts. If I have time next week or the week after, I will try.

And yes, you are right that a huge drawback is that this method occupies the printer for a couple of hours. So it will not be suitable for people who are doing 24h serial production on their printers. But for others who don't print at night, they could use that time to leave the print stuck to the glass bed in the printer, and keep the heated bed on.

I hope this clarifies the idea a bit more?

I am not sure it will work, but it might be worth trying. Then the Ultimaker would double its function as both a printer and an annealing oven.

Edited October 7, 2016 by Guest
corrected typos

[geert_2 558]

Here are a few pictures: one is from a very small transparant block (about 0.5cm), with a small metal ball from a bearing pushing on it: you see the stress distribution due to this load. The other photo is from a moulded part in Araldite two-component resin, containing a lot of bubbles due to the mixing. You see the stress concentrations around each bubble. And the last shows the moulded in stresses in a dental appliance, especially around the metal pins.

The higher the stress concentration, the closer the coloured lines are sitting together. This is a bit like height-lines on a map (I don't know the correct English term).

Edited October 7, 2016 by Guest

[IRobertI 521]

During a visit to UM a long time ago I remember that one of the material guys there talked about something related to this. He was doing strength tests on prints in different materials and one issue he was having was that if he printed a whole plate of parts he would see differences in strength between the samples. This is of course sort of expected due to the process but what he found was that it was consistently parts that where printed first that deviated from the parts that were printed last.

So, to me, it sounds like something like this could work quite nicely. The only thing I'm wondering is if the plate would reach a high enough temperature for some materials. I'm not a materials expert so I don't know what type of temperatures you need, but that could be a limiting factor.

[SandervG 1,521]

Hi Geert,

Thank you for your elaborated explanation, it is much clearer what you mean now and I have learned something new

Depending on what type of models you have, would annealing your model upside down in an over help battle sagging?

By closing of the printer quite high temperatures can be reached, but it may take some time.

I think you may have spoken to Steve at the time. I remember he was conducting such tests. I can imagine that besides the uneven cooling of the fans the position on the bed alters the draft it may receive from the outside. I think the bed is more or less equally heated, but it is easier to loose some heat near the edges than in the center. Printing with a door should probably make all prints more consistent.

[geert_2 558]

In its simplest form, the "box" around the model would be a simple standard food box, as in the photo. Or two on top of each other, for a better insulation. Just put it over the print at the moment it finishes, adjust build plate temp manually to the desired level, let it sit there for a couple of hours, and then very slowly let it cool down.

Maybe I forgot to stress this point in the previous posts: after annealing, let the prints cool down *very gradually and very slowly*, preferably over a period of a few hours. Don't remove the cover before it has totally cooled down to room temp. So that you don't introduce new stresses due to uneven cooling...

Whether it works or not, will be a question of a lot of trial and error, I think. Too low temperature and too short time, and it won't do much; too high temp and time and it will cause other deformations. So you need to find the exact spot in-between for your models and materials.

@ Sander: Letting an object hang upside down wouldn't prevent this model from warping (see photo). This type of model requires to have a weight or clamp on it while being pushed against a flat plate, or to be glued to the build plate. For other models such as figurines things may be totally different. This particular model has been sitting in an oven for a long time (without any clamps or weight of course), just to see how it would warp and what would happen to it. It first warped to one side, then straightened again less or more, then warped to the opposite side, and became much harder than original. This is colorFabb PLA/PHA.

@ |Robert|: I wrote this idea with PLA in mind, since that is what I use, and it is where these temperature-related warping problems occur most in daily use (as in a hot car). I didn't think of other materials. But indeed you are right, the build plate may not reach a high enough temperature for some nylons or so. In that case people still have the option to move the whole glass plate into an oven immediately after printing, before cooling. Just like in Artiz' original method.

For those of you who print transparant materials, and who have polarising light filters available (or polaroid sun glasses), you could try if this method removes stresses indeed.

The setup to make such photos is: light source - polarising filter - transparant object to photograph - polarising filter - camera. Thus you need two filters, one below and one above the transparant object. By rotating one of both filters, the colors will change. So you can rotate it until you get the best results.

As a light source I would recommend an incandescent lamp, or a high frequency TL-lamp, or LED lamps fed with DC-current. And old-school TL light on 50Hz or 60Hz will give poor results due to the flickering interfering with the camera's sensors. I use a modified transparancy light unit of an old defective scanner, which gives a nice even lighting at high frequencies.

[geert_2 558]

Hmm, the first annealing test in the Ultimaker was not totally convincing.

In the photo above, the flat object is freshly printed, the warped one is annealed. Both are the same PLA.

While it was sitting in the "oven" (=under a cover on the ultimaker bed), I couldn't resist playing with the temperature, and I increased it to 80°C, so it came loose from the build plate and reverted to a more relaxed state. I should have left it at 55°C, I guess...

Anyway, this shows how high the built-in stresses can be, even in "low shrink" PLA. And what you might get when you leave such a model in your car in summer.

So, when annealing, you really need to clamp the object very well, or it really needs to be stuck very well to the build plate. Otherwise you get what you see above.

[geert_2 558]

I again tried annealing (post curing) a few items in the UM2 printer. These models are very prone to warping due to their long shape and 100% infill, with high internal forces. They also have sharp corners and indents at the bottom (invisible in these pictures) that makes bonding to the glass plate way more difficult than in models with a flat bottom.

Annealing was done directly in the Ultimaker2 printer, immediately after printing, so the printer was used as an oven.

Annealing or post-curing these models seemed to increase their temp resistance by about 10°C, and it reduces warping when exposed to heat after the annealing (for example if left in your car in the summer).

The test procedure was as follows:

The models were printed with standard PLA settings: 210°C nozzle temp, 60°C bed temp, and 50mm/S speed. Glueing to the build plate was insured with my "salt method": gently wiping the glass plate with a tissue moistened with salt water, prior to starting the print. No brim, no raft, no glue, no painter's tape, nothing else.

After printing was complete, I immediately set the bed temperature to 60°C manually, so it did not cool down. And then I immediately put a lid on top of the models, to contain the heat. In this case I used a simple frigo box. So that the models are not subjected to air drafts or uneven temperatures. Then I let them sit like this box in the printer during the week-end (2 days). After that I very gradually reduced temperature down from 60°C to 30°C in small steps.

For this method to work, the models must glue very well to the build plate, so they do not warp or come off during post-curing. The "salt method" works very well for bonding Ultimaker PLA and colorFabb PLA/PHA, but is not optimal for ICE PLA (here some corners do lift in such difficult models).

Disadvantage is that this method occupies the printer for quite some time. In this case I tried two days, but I do not know what the minimum time would be to get good results.

This annealing or post-curing should relax internal stresses in the model caused by uneven cooling during printing. It should reorganise crystal structure and give a harder material. At these elevated temperatures around the glass transition temperature, the long molecule chains have a little bit of freedom to reorganise themselves, but not too much. So the heavily stressed molecules will relax, while the less stressed will still keep their shape. At least, that is if I understood things well, but I am not a materials specialist.

After this test, the models were put in my laboratory oven and subjected to heat: first 60°C, then 70°C, then 80°C, then 90°C.

Untreated models already warped severely at 70°C, bending upwards. The annealed model stayed perfectly flat. At 80°C the annealed model started to warp slightly too, but now downwards, thus in the "wrong" direction. I have no physical explanation for this, but it is like it is. Increasing temp to 90°C did not really make much difference anymore, just a little bit more warping, but not much.

When warping, all untreated models shrunk in length, and expanded in height. The annealed model however kept its length, and almost did not expand in height.

I do not have hardness measuring tools (and I don't know the standard procedures either), but the annealed models all feel stiffer than untreated models.

Photos:

First picture: a box used for covering the printed objects while annealing.

Second picture:

- 1st model (in front): annealed, but no other tests done on it. The thick middle section is 85.5mm long and 6.0mm high.

- 2nd model: annealed, then subjected to heat in oven: began to warp at 80°C, in the "wrong" direction. Middle section kept its length. Height increased by 0.02mm, almost nothing.

- 3rd model: not annealed, subjected to heat in oven: began to warp at 70°C, in the expected direction: upwards. Shrunk in length by ca 2mm. Height increased by 0.25mm, too much for this application (doesn't fit anymore).

- 4th model (at back): not annealed, subjected to heat in oven: began to warp at 60°C, then more severely at 70°C, in the expected direction: up. Shrunk in length by ca. 3mm. Height increased by 0.5mm.

- 1st, 2nd and 3rd model: PLA, Ultimaker, color: Pearl

- 4th model: PLA, ICE, color: white

This is a quick test on a very small number of samples, so it is not scientific, but it gives a rough indication.

In summary: annealing or post-curing increases stiffness, and decreases warping and shrinking due to heat. It increases temperature resistance by about 10°C, after which warping and deformation still start to occur, but less than in untreated models (and in the opposite direction, in the case of Ultimaker PLA and colorFabb PLA/PHA; I don't know for other PLA-brands).

Thus annealing or post-curing does not do any miracles, but it may be enough for some applications. For example for a non-critical decorative part in your car, or a custom cable clamp in the trunk, or something similar. So I would recommend it for this sort of applications, but not for critical applications (too risky) or for everyday prints (takes too much time).

I have no idea what this would do to the PLA's resistance to hydrolysis over a long time span, nor to its biochemical decomposition in nature.