Material Properties and Strengths (Filaments)

[gr5 2,167]

I MADE A BETTER AND INTERACTIVE GRAPHS HERE:

 

 

 

http://gr5.org/mat/

 

 

 

Okay well this post will grow when I add temperature properties.  I've been thinking about posting this for a few years.  Here is a graph showing some mechanical properties of some common filaments.  Please read explanation as this is a complicated subject.

 

Click to zoom in on chart (click 3 times - first click zooms, second click jumps to actual image, third click zooms into that).  Higher up is stronger.  Farther to the right is stiffer (not steel).  Both axes are logarithmic.  Most data is published by various manufacturers.  I have personally verified a few of these numbers with my own equipment.

 

 

VERTICAL AXIS

So the vertical axis is tensile strength.  It's measured by pulling on a cylinder of material from each end until it breaks.  Divide the force by the cross sectional area and you get the strength in psi (pounds per square inch) or MPa (mega pascals) for those who prefer metric (like me).  Anyway this is kind of complicated because the materials towards the left are quite stretchy and long before their breaking point the parts are damaged.  The point where it won't bounce back is the yield strength but I choose the ultimate aka breaking strength.  Or whichever was higher (some materials actually start to get weaker again - like steel.  Or PLA).  Now the weakest material shown, PP is actually showing the yield strength - it is actually much stronger than you would think so this is unfair.  But the machine that UM used to test PP wasn't long enough to test this value (the part never broke).

HORIZONTAL AXIS

This is the tensile young's modulus wherever possible (sometimes it's the flexural modulus which is close enough). This is tricky and complicated but for the most part indicates how stretchy/flexible a material is.  So ninjaflex on the left edge has very similar flexibility to a rubber band.  Most nylons (except shapeways) are much more flexible than PLA or ABS and this makes them very tough.  Materials to the upper left will be tough as hell.  In fact anything to the left of and including "nylon UM" can probably be driven over by a car and come out just fine afterwards.  Or a tank.  Or you can throw it against a brick wall with full strength or hit it with a hammer.  Most of those materials in most shapes can handle it.  Tough.  Materials to the lower right are more likely brittle (hence glass is the most brittle).  XT is probably somewhat brittle among filaments (I've never tried it).  Materials to the right tend to be hard.  The hardness scale and the modulus are closely intertwined.  Things to the right are harder, to the left are softer.

Specific Materials

The table that created the graphs here is at the bottom of this post.  Red materials above are for comparison and are not filaments.  ABS is shown in green above - this shows how different people testing the same material get different results.   Most of these tests (maybe all) were done on printed parts which will be a bit down and to the left of injection molded parts.  Two different companies tested Taulman Nylon 645.  With professional equipment and also got different values hence the two data points.  UM=Ultimaker in the chart.  XT is colorfabb.  POMC is delrin.  I'm very skeptical about nylforce CF specs.  If someone wants to send me some I'll print and test it with my stress/strain machine.

 

 

 

 

 

In the graph above there are a few points to keep in mind.  Materials with low softening temp are the easiest to print because they don't warp much in the temperature range from this temperature to room temp.  It's only about 30C difference.  As you move to the right the yellow group of materials is a little harder - parts are more likely to warp off the bed so you need to learn some tricks.  Maybe. They really aren't much worse.  The orange area with ABS and other materials are tricky now for a few reasons - they don't stick as well to the bed, you are now getting into materials with layer adhesion issues so you need to lower the fans, the bed takes much longer to heat up - you really need to enclose the printer to raise air temp to around 35C to get decent quality.  The red group needs nozzles that can go over 300C (no teflon please!) and print beds that can go to 150C and ambient air in the printer at 80C.  So this requires special equipment.

 

Also as you move to the right your materials can handle working environments of higher temperature.  The green materials can't handle a car with windows rolled up on a hot sunny day (neither can a human for that matter).  The yellow materials can handle this but can't handle boiling water.  The orange materials can handle boiling water.

 

SOFTENING TEMP (horizontal axis)

In the graph above the horizontal axis is a mythical characteristic called "softening temp".  For many of these materials in the green and yellow area I have tested them myself personally by sticking them in hot water. Above a certain temp they can be easily bent and when they cool a few degrees they stay in the new shape.  That's what I call the softening temp but in reality this value came from HDT (heat deflection temp) or glass temp in other cases or functional temperature in other cases.  It's a mixed bag.  So it's very approximate!  Normally you want the heated bed at a temperature a bit above this temperature such that the material is soft enough to flex a little and spread out the warping forces. 

 

PRINTING TEMP (vertical axis)

Also somewhat arbitrary as some materials like PLA have a wide range of printing temps.  Also variation in heater block design and variations in nozzle length, filament diameter, airflow touching nozzle, and more - affect what this temp should be.  But it is a good place to start.  Mostly I'm just showing that the red materials need special equipment to print them.  The green, yellow, and orange materials can all mostly be printed by most printers no problem.

 

In table below, take all values with a grain of salt.  Especially temperatures.  For tensile modulus notes read "horizontal axis" paragraph far above.  For tensile strength read "vertical axis" far above.  For softening temp - please read "softening temp" paragraph above.

 

I've already fixed several mistakes in the table below but ONLY on the website - please go my website for better data!

gr5 materials

 

[SandervG 1,521]

Thanks @gr5 , very interesting! Thank you for your time in putting this together. 

I was wondering; for tensile strength is it relevant to mention with which print profile it was made, like layer height? (Do you know since you did not perform all tests yourself, but also collected data left and right?) Or was it tested in parallel with the direction of the printed layers? (Or are these pure material strengths, so not performed on a 3D print but just a string of filament?

[gr5 2,167]

Most (maybe all) of these tests were perfomed on a printed part shaped like a bow tie but they were printed by different companies and using different printers.  The UM filaments were all printed by UM so you'd have to talk to them, lol.  Mostly each manufacturer printed the parts and sent them off to be tested or they had their own machines.  Anyway each company chose different nozzle sizes and layer heights and so on.  I've printed these "bow ties" myself and the best results come when you make the thickness a ratio of the nozzle width and you do all shell (no infill).  Thick layer heights (e.g. 0.2mm or more) and large nozzles (0.8) might give slightly stronger results but I suspect most companies used their default nozzle of the printer that was handy so probably mostly 0.4mm or 0.5mm nozzles.

 

More data to come!  Watch this space, lol.

[rajilpahuja 38]

This is awesome research @gr5 and very helpful in determining what specific material should be used for the corresponding application. 

[geert_2 557]

Thanks. These are very interesting data, and a very good starting point for selecting materials. Maybe this thread should be sticky?

 

Actually, I am a bit surprised at the positions of Taulman 618 nylon , and colorFabb XT and HT.

 

A suggestion: as you accumulate data, maybe you could add a sort of "error bars" to the values in the second graph (with green, yellow, and orange ovals)? Both horizontal and vertical lines. These error bars would indicate the useful range. Then for example PLA might sit at 210°C, but its error bar might go from 190 to 220°C. While another material at 210°C might only go from 205 to 215°C. This would give an idea of the useful range of the material, and how critical temperature settings are.

 

[SandervG 1,521]

4 minutes ago, geert_2 said:

Maybe this thread should be sticky?

ok!

Also, good suggestion about the range bars! 

[gr5 2,167]

The "softening temp" is the least accurate because it's usually glass temp (which can be less than 0C for something like nylon or ninjaflex which is useless info) or heat deflection temp which is also sometimes useless.  I'm pretty sure the softening temp is wrong for some of the items in the green area but I don't have any samples so I don't know for sure.

 

Most of these materials did indeed come with a printing range and so I picked a temp near the middle.  Some ranges were tight (e.g. 210-220) and some were loose like PLA (180-230).  It would indeed be great to have a min/max printing range pair of columns.  

 

 

 

[gr5 2,167]

Okay I posted an interactive version.  You can zoom in and hover tip shows all the data for each material.  Here:

http://gr5.org/mat/

[kmanstudios 1,120]

17 minutes ago, gr5 said:

Okay I posted an interactive version.  You can zoom in and hover tip shows all the data for each material.  Here:

http://gr5.org/mat/

Dang! Yer on fire lately

[twiereng 1]

Great work, thank you. Would love to see this as a pdf.

[rowiac 53]

@gr5

It seems as though PETG is missing from these charts, unless it's there with some other name.  I recently purchased a spool of it to try and was wondering how it stacked up against other materials.

 

Great job on this by the way.

[Gigi 43]

Ultimaker PC filament (black - white)  with only 60 degree bed ?

[dextop 0]

Amazing great work! Thank's a lot

[gr5 2,167]

On 3/5/2018 at 3:36 AM, Gigi said:

Ultimaker PC filament (black - white)  with only 60 degree bed ?

Fixed it on my website just now.  Thanks!  Table above is still wrong - not so easy to create the table again as the data is all in json format now.  Although I might write some code to output it into a table some day.

[FalmouthLouis 3]

Just to let you know that I'm working with an investigator from Exeter University to do some serious tensile testing on around 20 variants of a standardised design. We will look at four contrasting materials (PLA, Nylon, Resin and a strong co-polymer), but also at varying print orientations, degrees of infill, infill patterns and so on. At the end of this, I would hope we can produce a guide to the ways of improving tensile strength, starting from PLA at 20% infill, which is the standard entry level setting for most of us. If people are still monitoring this strand, I'll feed results back into it.

[SandervG 1,521]

Hi @FalmouthLouis , that is great news! I would be very curious to read those findings. Looking forward to the results, thanks for sharing!

[FalmouthLouis 3]

SandervG - I'm interested in the views of you guys in Ultimaker as to how you think the various infill designs will affect tensile strength. Assuming infill density is kept to 20%, which infill design should be the weakest and which the strongest? Do you have any other expectations as we play around with the options? By all means feed me with any questions that you think we could throw some light on. My colleague has produced two variants of a testable design - one 8mm deep, one 10 mm deep. We can afford to test something like 24 variants of these designs. What issues do you guys think we could most profitably test within that 24 variant constraint? Remember that we need to focus on issues which matter to real life practitioners like myself and their real life clients. This all started because I am doing a bit of printing for the marine sector and I needed to be very clear in my head about how best to give commercial clients like that the structural toughness they might need, without going over the top in terms of printing unnecessarily in exotic materials.

 

[SandervG 1,521]

Maybe @Tomhe and/or @KristelB have something to add to this?

[kmanstudios 1,120]

I will be keeping an eye on this.

This sounds very interesting. I am curious as to which brand will be used for testing.

[FalmouthLouis 3]

kmanstudios: probably Faberdashery for PLA, Colorfabb for Nylon, 3DGBIR for an advanced co-polyester. Resins probably through the Photocentric setup. The point is that there are now so many filaments out there that it's pointless trying to be scrupulously precise about the differences between filament A and filament B. We want to be pretty precise when charting what happens as we experiment with different print setups using one filament (probably PLA since that is what most of us start of using). When comparing the different materials, we just want to be able to give a general sense that shifting from PLA to the likes of Nylon will produce a perfomance improvement of around so much.

 

If you have views about brands etc, feel free to make suggestions. Within our various constraints (financial, plus the fact that my Exeter University colleague needs to finish her work by a specific date) we are trying to make this exercise as useful as possible to general users as well as the community who come on to this site.

 

I'm an Ambassador for CREATE Education, so am part of the Ultimaker community, hence my willingness to work with 3DGBIRE.

[kmanstudios 1,120]

I would hope that there is a disclaimer then that you used certain brands and how you feel that brands would not matter that much. To buffer that, I would test two brands just to settle that theory. People settle on brands for a lot of perceived reasons and this could lead to an idea of lack of thoroughness or even bias. It would seem that brands would submit for testing; at least their small sample rolls. Also, do not forget the new TPLA's out there. (Tough PLA).

 

 

[FalmouthLouis 3]

Understood. However, we probably only have 24 variants to play around with, so we might only be able to do something like test out three differing PLA @ 20% infill versions which would at least give a sense of how different brands of PLA can produce different results (or not). Whatever we're doing is not going to allow anyone to claim that brand X can out-perform its competitors.

[gr5 2,167]

I agree that brands don't matter that much after testing many different materials.  I built my own strength/strain machine that uses these "bow tie" parts to test.  It agrees with the published data pretty well.  A few thoughts.

 

1) For the tensile test - infill will make a HUGE difference.  Also when I build these bow ties I set shell/wall width to 1 meter to be sure it's 100% shell and not cross hatch infill.  This way the filament "grain" won't affect the results.  For ABS if you print the bow tie vertically and if you aren't an expert at ABS layer bonding the part will be much weaker when you pull it apart (ultimate tensile strength).

 

But this infill on tensile parts is not interesting to mechanical engineers.  Normally you don't need 100% infill to make a part stronger.  normally it doesn't help.  Take a beam for example.  You can drill lots of holes through it "sideways" without hurting the strength.  So if you are going to test different infill patterns you shouldn't be doing these butter fly tensile tests but instead maybe you should be doing flexural testing.  In flexural testing the infill pattern shouldn't make much difference on a beam.  In other words I suggest you *always* do 100% infill and don't do it at an angle, do it with walls set to 1 meter so it's a concentric infill.

[gr5 2,167]

The editor got strange so I had to start a new post...

 

2) 8mm versus 10mm deep shouldn't make much difference.  I think it's a waste of your time.  You should hopefully get the exact same results regarding tensile modulus, ultimate strength, elongation at break, etc.  Hopefully you have a proper stress/strain tester.  If not you might want to build the one I built for about $200 in parts.  When calculating modulus and strength you are taking into account the cross sectional area of the part.  So the 10mm part will have a larger cross section and be stronger and it should all work out.  In fact any errors between the two results are probably air gaps - areas of the part where there was air instead of plastic.  10mm should be more accurate.  But at some point the part is too strong for the machine doing the testing.  I do about 6mmX6mm area because my machine only goes up to 300 pounds force and my parts tend to break at around 100 pounds force (sorry for the imperial units - I usually think in metric but sometimes...).

 

3) You might want to print some parts vertically but really, if your results are different than horizontal by more than 10%, well then you aren't printing the part right.  Nylon and ABS and PETG don't always have good layer bonding and you need to turn the fans down to the lowest setting or probably off and also cover the front and top of your printer when printing everything other than PLA.

 

4) PLA has a property that you won't be able to measure with a normal machine - it bends with time.  So if you make a chain link and hang a heavy weight on that chain link, over many months it will stretch out - it will "neck" (it will get thinner) more and more until it breaks.  It might last a month, or it might last a year, but for brutal environments (marine?) where it is under constant load, PLA is not the best choice sometimes.  Anything other than PLA should be fine.  Like nGen/PET.  So if you use it on a boat to hold a door closed (a latch) pla is fine.  If you use it as a part in a winch or a cam cleat it will fail if it is under a calm but steady load for weeks at a time.

 

[gr5 2,167]

I really hope you publish your stress/strain graphs.  I'm very disappointed that these manufacturers publish the modulus on their material but not the graph.  The curve looks nothing like metals - there is no obvious elastic region (straight line on the graph) that transitions to a plastic region (more curved).  Instead, it's almost all curved so the tensile modulus is really only relevant and useful in the first 5% of the graph.  It's a bit misleading.  So if you just plugged only the modulus and yield strength into software that calculates where a part would fail - PLA and these other plastics are actually MUCH stronger when used for example as a beam because they bend more than you'd think such that they are tougher than you would think.  When people try to get the modulus through the flexural method you get a different answer, probably because of the curve shape.  In theory whether you measure Young's Modulus through the tensile method (your bow ties) or the flexure method (like breaking a pencil) you should get the same result but in practice you get different results.  Not that it matters much as engineers usually design well beyond where something will break.

 

Having the entire stress/strain graph is much more useful and interesting.