Reprap Extruder’s: A few possible explanations.

Following on from my previous introduction to all things plastic, I thought I would try to help explain why home built plastic extruder act the way they do. The idea being that if you have a better understanding of the processes that are causing you problems, you can more likely design usable solutions. Please note that I am by no means an expert in this area, im just using what knowledge I have to give my best estimation of what may be taking place.

  • Why does the plastic pop, smoke and out-gas if it gets too hot?

I covered this briefly in my last pot when discussing thermosetting plastics. When you heat a plastic up high enough, you not only start to break down the weak bonds between chains, but also the stronger bonds holding the chain together. When they start to fail you end up with some very reactive chemicals being exposed to the atmosphere. Some of these compounds will stay gaseous while others will react with oxygen to produce a dark sludge/charred material. The simplest and most obvious way to prevent this is to keep the temperature low.

Remember that the temperature you extrude at has a direct impact on the properties of the object your printing.This is especially true for ABS. If ABS is extruder in the upper temperature range the object it will be more resistant to high temperatures but have a lower strength. If its extruded in the lower temperature range it should be a stronger object.

  • Why do some plastic filaments curl back on themselves when you extrude it into mid-air? Why not all plastics?

This is a more interesting, although not very important question. Not all polymers are created equal, in fact far from it. During the plastic manufacturing process (polymer syntheses or ‘polymerization’) there will always be different polymers of different lengths. Some will be very long chains, while others may only be a few units in length. These different polymer chain lengths have important implications for when they are extruded. Once melted in the heater chamber of a reprap device, each polymer will have the freedom to move around a little as its pushed out of the chamber. Due to friction with the chamber wall, the longer polymer chains are dragged along the chamber wall while the shorter ones migrate to the centre of the filament thats being extruded. This effectively forms a skin of long chain polymers at on the outside of the filament with a short chained core. Due to random distribution, at any one time there will always be slightly more long chain polymers on one side of the filament than the another.

After the filament of plastic is extruded it begins to cool and all polymer chains start to contact a little. The longer the chain, the more it contracts. So you now have a filament of plastic that has a uneven distribution of contracting long chain polymers on one side. This puts a stress on the filament causing it to curl upwards in which ever direction happens to have the most long chain polymers at that time. This problem should be more pronounced in plastics that have low levels of cross linking such as ABS and less so in HDPE. Take what I have just said with a gain of salt as this is just my take on the problem and not something I have inferred from experimental results…

Although this is interesting, it has little importance for a reprap extruder as its normally extruded so close to the bed/object that there is no noticeable effect. However it could become very important if you were trying to extrude long filaments onto a spool from plastic granules or shredded milk bottles.

  • How does each extruded layer bond to the layer below it?

As previously discussed, all polymer chains are moving. The hotter the plastic, the more the chain will move. When you extrude a hot layer of plastic onto a cold layer of plastic below it, it will melt surface layer of the plastic below. As you now have an inter facial region of molten plastic you end up with the polymer chains diffusing into each other due to their random movements. Once cooled, the polymer chains that were at the surface are now entwined and bonded to form a seamless join.

On a side note, this is the same process used for solvent welding of plastics. When you apply a solvent it replaces the lose bonds between the polymers with its self. This frees the polymers to move and in effect lowers the plastics Tg until the solvent is removed, normally through evaporation. Solvent welding could be explored as a way to join different parts printed on a reprap device that are either too large or have complex geometry (overhangs) and as such can not be printed as a single piece.

  • What exactly causes warping and why does a heated bed or chamber help?

As a polymer chain cools, it contracts. A very simple concept to imagine. So why then does a printed object want to peel up at the edges? Surly the whole object should contract equally, leading to an object that is a few percent smaller than what your printed when cooled to room temperature.. This would indeed be the case if an object had a random distribution of polymer chains in all directions and was laid down all at one time. However, when extruded in lines there is an uneven tension in the horizontal directions as it cools.

Possibly a lot more important than this is the temperature of each layer with respect the layer below it. Imagine a single line of plastic laid down on a printer bed. If it was left to cool it will want to contract and shorten. If it was not stuck to the printer bed it would just shorten in the same way that a rope shortens when a load is  removed. However as this line of plastic is stuck to the printer bed it can not move. Now imagine a second line of plastic laid down directly on the first, while the first layer is still hot, and then a third and forth and so on until there are 5 lines stacked on top of each other. Logically, you may imagine that if you left this to cool it would shorten like the unloaded rope analogy.

However in practice it seems that the lines pull up at the edges. I’m guessing this stems from the uneven way in which the object cools. The middle of the line of plastic cool slower than the out side edges and the whole object cools from the bottom up. Remember that where ever it cools there is the tension pulling inwards. So you now have an object that is pulling more at the edges and bottom, while also being constrained in the middle.

Now imagine this same process but for a 3d object and you can start to understand why it lifts at any sharp corners. This is actually a lot harder to visualise than I first thought and I doubt I will be able to fully understand it until I get to play with my own Mendel..

A number of solutions exit to this problem of warping that are currently being used and that I’m aware of. They include a heated bed that keeps the object and the air above it warm. This helps speed the movement of the polymers slipping past each other to remove the built up stresses internally. A heated work chamber also has the same effect.

Another way to reduce these stresses is to design objects with warping in mind. Any sharp changes in direction will be a strong point for warping. So using rounded edges could help. I wonder if changing the orientation or shape of the infill could also help? Maybe a honeycomb or triangular infill instead of the standard hash pattern?

For more info google injection moulding warping and there are plenty of guides that give good tips for low warping product design.


Well I hope someone was able to glean some useful information out of this. If nothing else it cemented some of the ideas more deeply in my own mind.

When I get time for my next post I will hopefully shed some light on the theory behind hot melt adhesion, and more importantly, what properties could make the best printing bed.

About Richard

I am a Materials Engineering working in the field of Magnetic Materials in Melbourne, Australia. This blog covers my personal interest in all things CNC.
This entry was posted in Thinking aloud (Theory) and tagged , , , , , , , , , . Bookmark the permalink.

12 Responses to Reprap Extruder’s: A few possible explanations.

  1. Pingback: Plastic Science! | MakerBlock

  2. Jonathan says:

    Hi, here is my take on curling:

    I don’t know if there is an uneven distribution of chain length within the filament, but I don’t think it would properly explain curling.

    I think that the curling is caused by the layered structure of the printed parts:
    As one layer is printed, it starts cooling immediately, and therefore contracts a little.
    Then, another layer is printed on top, which is hotter, then cools and contracts as well.
    However, although the two layers will both continue contracting, the top layer will contract more relative to the bottom one’s length at the time the top layer was extruded.
    The only way these two layers can stay adherent to the other is by curling upwards, the amount of curling depending on the amount of contraction of the top layer and the amount of contraction the previous layer had gone through when the top layer was laid down.

    In >2-layer objects, the problem just compounds itself.

    A heated build platform (or chamber) mitigates the problem by stopping the lower layer(s) contracting too much by keeping them warm, it might also allow stresses to be relieved by letting the warm layers slip on each other while they equalize temperature.

    I expect non-curling plastics either cool down slower(higher thermal mass) and/or contract less.

    I do agree though that the extrusion process causes the filament to contract differently in different directions (polymer chains are probably more aligned along the length of the filament) causing curling and warping dependent on the geometry.

    What do you think?

  3. Richard says:

    Hi Jonathan

    I’m a little confused, when you refer to ‘curling’ are you referring to the way an object lifts at the edges after its printed or the way a filament curls back on its self when extruded into mid air? I must admit I didn’t make it very clear in my original post so I have gone back and changed a few things.

    I think your spot on with your description of how each layer contracts more relative to the layer below. This is no doubt why I have been reading that the speed and size of an object is so important with regards to warping at the edges. I will update the post to reflect this. The alignment of the polymer chains most likely also plays a part. I wonder if changing the orientation or shape of the infill could help? Maybe a honeycomb or triangular infill instead of the standard hash pattern?

    This is why I can’t wait to build my own Mendel so I can test all these things!


  4. Jonathan says:

    Hi again,

    I was talking about the edges/corners lifting up; I had forgotten about mid-air curling. (I unfortunately don’t have a machine yet either…)

    Do you think the mid-air curling could also be due to viscosity differences? (I’m guessing these could be due to temperature or indeed chain length differences in the filament)
    I’m not sure uneven cooling contraction would cause the mid-air curling, the filament would have to cool very fast for that no? (That said, I’m not sure how fast this curling occurs, or what the contraction vs. temperature curves look like…)

    Having a RepRap would sure help in figuring all this out! Hopefully one day I will have enough time/spare money to make one.

  5. Richard says:

    I am fairly sure that the problem of the filament curling back when its extruded is due to the longer polymer chains being drawn to the out side of the filament. I feel its about right because I was told in a lecture that a skin can form on extruded plastics, giving its surface a different properties to the bulk material.

    I just took the next step and assumed that this uneven skin over of longer polymers is also what leads to the curl back when they start to cool.

    Thanks again for your first comment. It really got me thinking about how important uneven heating is and how this effects the object. As you can see I ended up doubling the number of paragraphs talking about that one issue and I still don’t feel like I understand it =S


  6. Pingback: Plastic Science! | 3D Printer Parts

  7. Pingback: Tweets that mention A few possible explanations. | Capolight Electronics Projects. --

  8. Pingback: So you want to learn about the thermal properties of plastic… | MakerBlock

  9. Erik says:

    Your analogies are really helpful. I’ve got a lot of hands on experience with FFF (laying down molten filament) through the RepRap but I have no theoretical background at all. Your inferred description seems mostly accurate. Non-uniform cooling equals non-uniform shrinkage. As it shrinks it causes stress which results in deformation and/or delamination (the interface between the layers will be the weakest link).
    When an object held down very well (good adhesion to the substrate) the effects of the stress will not compound because it doesn’t deform much. After a few layers the object itself will have some strength to resist the stress from the next layers. With the first few layers, however, they can bend and curl up really easy because it’s still thin. Getting it not to compound is important. Having a good platform and a raft helps.
    Ideally the raft should sit on little dots of extrusion instead of lines, which reduces the shear force (causing the raft to detach).

    Anyway, keep posting. It’s very interesting 🙂

    • Richard says:

      Rafts definitely seem to help, but they look so time consuming to remove and I imagine waste a lot of plastic. In an idea world the whole printer bed would be in a heated chamber to keep the plastic just above its Tg and so allowing any stresses to be relieved through a controlled cooling process.

      From what I can gather this is what they use on commercial machines. However considering a Mendel is made from the same plastic its printing this wouldn’t end too well.

      I wonder if there would be a way to isolate the bed with an insulated lid after a successful print and then do a controlled heating/cool down cycle to help relieve internal stresses. This could even be all controlled by a heated bed.

      This is where I can’t wait to get my own Mendel to start trying all this things.

  10. Kent McNaughton says:

    I’m having the edge-curling problem with PLA on most prints that take up a majority of the available workspace. Notably (maybe ) is that it seems not to occur on parts that are either more-or-less empty in the middle, or have not-a-lot of plastic in the ‘Z’ direction.
    I’m wondering if there maybe isn’t the cooling contraction effect Richard mentioned.

    I’m printing on a heated bed, set to 60 deg C–though I’ve seen the same effect with the bed at 85 deg C. The sides of warping parts are cool.

    Could it be that the center mass–as it cools from the bottom up–is creating a force too large for the heated bed to resist? This would seem to be especially the case with parts with a lot of plastic–particularly toward the center of the part, and less so for parts with empty or near-empty centers.

    I note that the dates on this post are from 2010. But I still can’t find a definite answer beyond “Try this.” “Try that.” and “this (kinda) worked for me.” searching the RepRap site or the Internet in general.

    • Richard says:

      Hi Kent

      Warping is a problem that I have given some thought to a long time ago (2010) and wrote a few follow up thoughts here and here. I have not had any hands on experience as of yet with the problem so unfortunately I can’t be of much further help.

      I best advice would be to read the blog by nopehead as he has explored this problem in detail in his past posts and has had good sucsess in mostly solving the problems.

      Best of luck and happy printing!

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s