Material Shrinkage and Accuracy

I've been reading "Rapid Prototyping--Laser-based and Other Technologies" by Patri K. Venuvinod and Weiyin Ma that Sebastian suggested for the book club.

In this book they talk shortly about the accuracy issues of FDM due to material shrinkage. One of the concepts they mention and give a source for is 'elongational viscosity' due to the polymer being squeezed out of a tip. They say that:

"""
The nozzle squeezes out the fused filament in a manner analogous to a toothpaste tube and deposits the material at the desired location on the partially constructed part. Polymer flow through the FDM tip is not by simple shear (as in SL and SLS) but a combi8nation of shear and elongation, so it becomes necessary to recognize the viscoelastic nature of the polymer (Tadmore and Gogos 1979). In particular, the resistance to stretching needs to be modeled by using the notion of 'elongational viscosity' (Bird et al. 1977). Further, part properties can be affected by the flow orientation of filament placement during layer buildup (raster direction) relative to the part loading direction (Fodran et al. 1996 and Bertoldi et al. 1998). It has been found that the tensile strength of FDM-built specimens aligned diagonal to the testing direction are generally lower (by about 18%) than those obtained by aligned specimens, thus suggesting that interbead bonding may be a limitation. Further the actual strength values my fall significantly short of the values claimed by the manufacturer.
"""
(section 8.3 pgs 288-289 paragraph 3)

Sorry for the long quote--but get used to it in this post... ;-)

The point is we can model the extrusion through the concept of 'elongational viscosity'. I put in the rest of the paragraph because the FDM strength bit was interesting.

A few pages later it goes on to say:

"""
Material shrinkage associated with temperature and phase changes can mar the accuracy of FDM parts. However, the resulting warpage and curl are somewhat attenuated by the ability of polymers to stress relax. To reduce the warpage, Stratasys deposits the material onto a removable foam base that is firmly anchored to the bed of the machine. THe foam lowers the cooling rate, thus allowing the viscoelastic polymer to relax out some of the stresses generated during crystallization. Further control of the cooling rate is achieved by controlling the surrounding air temperature.
"""
(section 8.3 pg 290 paragraph 2)

This paragraph is key to room temperature extrusion. I understand that a lot of people have been using fans and whatnot to cool down things. This is counter-productive in that it may cause faster cooling of the part and thus more warpage. Perhaps a better strategy would be to use a hair dryer in order to *slow down* the the cooling and thus attenuate the warpage some more.

Further, I think it was Forest that had been using foam to extrude onto but it has been Nophead that has been doing the tests as to shrinkage. Perhaps a combination and comparison of the two methods together (ie printing on foam and not) would be instructive.

Okay, this post has gone on long enough. Tell me what you think and if you have any ideas or have read anything else relating to this and how we can apply the info to make RepRap better.

Demented

Hi Demented,

... there already were some threads with tis issue in past - different methods of re-tempering the topmost slice or using heat-controled housings for extruding short beneath the melting-point of the polymer should do great ...

I agree: - for FDM-fabbing the easiest way is a housing, tempered at some centigrades beneath the melting point of the FDM-material.

But i'll try with dispensing pastes at room temp, where the shrinkage-issue will bite uniform for every slice when hardening with UV or heat, or at once for the complete object, when burning in an oven or such ...

Viktor

Well, it's not that we need to enclose everything in an oven to get the temp to a few C below melting, what we need is to calculate an acceptable error for this project and then use directed heat application--think hot air gun--to maintain the delta-T for that error.

Demented

I don't think 'elongational viscosity' has much bearing on this at all, perhaps it affects ABS. The four plastics we are playing with all behave very differently. I think some are crystalline and some amorphous.

>This is counter-productive in that it may cause faster cooling of the part and thus more warpage.

It might be true for ABS, which does not need a fan, but I don't think it is true for HDPE. You get better results with a fan. If you cool it down fast it becomes firm to resist the pull of the layers above. If not you get more rounded corners etc. I don't think it makes any difference to the warping. The last set of tests were without a fan, because it stresses the extruder too much, but that limits the speed you can extrude at. If you try to make a small object it just becomes a molten blob unless you add interlayer cooling delays.

>Further, I think it was Forest that had been using foam to extrude onto but it has been Nophead that has been doing the tests as to shrinkage. Perhaps a combination and comparison of the two methods together (ie printing on foam and not) would be instructive.

I have done tests with foam board as well. It is not strong enough to hold the object as flat so you get significantly more warping than you do with a PP board.

I think re heating the object while still attached to the board to just below its melting point will reduce the warping you get after removing it.

Interestingly I have just done a test with PLA and a fan and that gives very low warping. Probably the best figure for 100% filled object I got so far, i.e. better than HDPE, ABS and PCL. I have to leave it a few days to be sure.

ABS is good because it has a low thermal expansion so doesn't contract much.
PCL is good because it only goes hard when it gets down to about 40C so it doesn't shrink much afterwards to get to room temp.
PLA is good because although it has a high melting point (175) it has a glass transition temperature at around 65C. Below that it is very hard so the layers that have cooled can resist the contraction of the cooling layers until they get down to 65C, below that, again, it's not far to room temp, so little further contraction.

HDPE is bad because it goes hard pretty soon below its melting point and then has a long way to cool to room temp and has a large thermal expansion coefficient.

---

[www.hydraraptor.blogspot.com]

I think the type of information this was written about was ABS and other "standard" plastics used in FDM. Since HDPE isn't much used in commercial machines--is it in any?--the information probably doesn't apply to it directly as you say. It is likely, however, that the levels of shrinkage that Stratasys and other companies are worried about is less than what we will be worried about.

As to the foam thing...are there other types of foam board out there that we aren't looking at? Obviously there is the stuff that the commercial companies use but there could be some other stuff that has the properties that we are looking for.

Demented

Hi Demented,

for my first LOM-milling-test ( [forums.reprap.org] ) i used a foam-sheet called "Foamalux", which i can buy in different colours at my hardware-store for decoration-issues and such.

It has a solid surface and is nearly as hard as a 'normal' plastic-sheet, but it's foamy enough to be a good thermal insulator.

Look if you can find a similar product in your local stores ...

Viktor

That is PVC foam sheeting. Pretty cool stuff. I don't know where to get it locally but I can get some. I'll see what I can get and try it.

Demented

I've heard of this material as "expanded PVC" or "foamed PVC" and I have bought it locally from [www.tapplastics.com]; (they are located near me). I'm sure you can pick it up in other places.

It is reasonably stiff and has a smooth surface finish, but when you cut it you can see that it has millions of microscopic bubbles, which is where the "foamed" part of the name comes from. I assume that this makes it a good thermal insulator.

The stratasys insulation foam appears to be fairly thick and fragile:
[www.umaine.edu]

So apparently material shrinkage might be an issue

I have been using 1 inch thick open cell Styrofoam and it has been working very well with the exception that it does melt a little causing the part to sink. ABS sticks to it very well being it is made from styrene and so is ABS. I have been tring to find out what Stratus uses for foam?

Bruce W.

I don't think stratasys machines need to use foam. I've seen one in operation and I recall it printing the parts directly onto a hard black plastic insert (that could be pulled out of the machine to facilitate finished-part removal).

Was the black plastic UHMW? I've been messing around with UHMW as a thermal break to replace the PTFE on my extruder. Seems to hold up pretty well to most abuse and machines nicely.

Demented

I want to know about HDPE material shrinkage.How much % and how to calculation about it.

If you look here: [www.ides.com] you can see HDPE mould shrinkage is given as 0.7% to 3%.

To calculate it you need the thermal expansion coeficient which is given as 120 x 10-6 m/m C here: [www.engineeringtoolbox.com]

So when HDPE cools from its melting point of 130C to room temp 20C it will shrink by 110 x 120x10-6 which is 1.32%. That lies in the range stated above. I don't know if there is an additional size change due to the state change from liquid to solid but as the figure is in the right ball park I guess not.

---

[www.hydraraptor.blogspot.com]

nophead Wrote:
-------------------------------------------------------
> I don't think 'elongational viscosity' has much
> bearing on this at all, perhaps it affects ABS.
> The four plastics we are playing with all behave
> very differently. I think some are crystalline and
> some amorphous.

Elongational viscosity sounds like it's part of coping with the fact that melted plastics show a lot of non-newtonian behavior; deformation is not purely linear based on applied force. I'd expect it's for modeling things like die swell and the related behavior inside the nozzle. If you're going purely experimental without quantitative theoretical work the number probably is irrelevant. Also, melted plastics aren't crystalline; that's a behavior that shows up on cooling - and probably has substantial impact on warping. Has anyone tried LDPE? it's much less crystalline than HDPE...

> >This is counter-productive in that it may cause
> faster cooling of the part and thus more warpage.
>
>
> It might be true for ABS, which does not need a
> fan, but I don't think it is true for HDPE. You
> get better results with a fan. If you cool it down
> fast it becomes firm to resist the pull of the
> layers above. If not you get more rounded corners
> etc. I don't think it makes any difference to the
> warping. The last set of tests were without a fan,
> because it stresses the extruder too much, but
> that limits the speed you can extrude at. If you
> try to make a small object it just becomes a
> molten blob unless you add interlayer cooling
> delays.

What demented is talking about doesn't mean that you want to change the temperature at which you're depositing the next layer onto a previous one; it's reducing the derivative of temperature as the layer is solidifying, or otherwise keeping the new layer warm longer. The idea is to leave the solidifying layer time to be stretched by the layers below it, while it's still less solid than them. You wouldn't want another layer on top of it during that time, and it'd slow the build down a lot - but you could set it all in software, so low-warpage builds just take longer. Essentially, it's suggesting that because plastics slowly deform permanently under stress, and quicker when warm, cooling a layer quickly with a fan is not equivalent to letting it cool on its own or slowing it's cooling further.

Also, it occurs to me that cooling a layer slowly doesn't require the attention of the extruder - so adding a long inter-layer delay could just mean that the lead time on a part is longer, but not that the machine can produce less volume.

> I think re heating the object while still attached
> to the board to just below its melting point will
> reduce the warping you get after removing it.

That'll certainly get rid of the internal stresses, but it might do it just by speeding up the warping process you're observing over several days - although the surface of the part would relax faster than the core if you did the heating relatively fast, which might make it warp less troublesomely for solid parts. But controlled relaxation during the build - when we can use where the heat is to control which layer deforms, like a blacksmith - might perform better.