Tensile Strength Testing

[rsterne]

I keep hoping that I will find a Carbon Fibre filament that has a greater layer-to-layer strength (Z-axis) that is closer to what can be achieved in the horizontal (filament) X-axis.... I can't afford to play with the $500/roll blends, and it seems to me that CF strands (not ground up dust) should be capable of beefing up the stiffness in a filament like PETG that is not known for it, and they are not terribly expensive.... So, I ordered a couple more brands, and today I tested the IEMAI PETG-CF.... Here are the results....



It was slightly stronger in the X direction than the other PETGs I have tested, at just over 7,000 psi, but the vertical layer-to-layer bonding was the same, at about 3,700 psi.... This is weaker than PLA in both directions.... One other thing I noticed with this filament is that in line with the filaments the strength increased with lower temperatures, which I have not seen before.... I suppose it might be even stronger at 210C, but the layer bonding (which peaks at 250C) is so poor at 220C there was no point in going lower.... I have some Kimya CF-PETG on the way, hopefully one of these will be better....

I was using my new coupons to test this, after confirming that they don't influence my results, and I am very pleased with the consistency and mode of failure the provide....

Bob

[aluminumfetish]

https://www.youtube.com/@CNCKitchen
This guy has done it all. A great resource for 3DP testing.

[rsterne]

I agree, Stefan's videos are classic.... However, he has not tested filament strength at various temperatures, which is what I am doing....

Bob

[WhatUPSbox?]

There are some interesting discussions on the affect fan cooling has. Here is a summary.
https://3dprinterly.com/how-to-get-the-perfect-print-cooling-fan-settings/
In the embedded videos Stefan does his usual thorough job of strength testing and CHEP shows some parameters in CURA that can be used to optimize. Note the comment in Stefan's video on the Creality fan design. Also reducing fan cooling to improve strength brings geometry and ambient temperature into the mix.
I think, while there are knobs that can be tweaked for Z axis strength, establishing a specific value for strength margin calculations should be approached with caution.

[WhatUPSbox?]

I modified my test rig to accept Bob's coupon configuration and printed a couple of checkout coupons. Unfortunately in the excitement to make the modification, I forgot about the leverage ratio in Bob's test setup. Surprisingly, the small, geared hobby motor/leadscrew was able to pull the XY coupon (PLA) to failure at 83.54 Kg. A reasonable data point but I don't think I want to load the brass leadscrew nut to those levels. On the other hand it provided a new proof load value for the setup . I'll need to add some leverage.
Bob, thank you for the coupon design.

[rsterne]

Not sure how I missed posting it, but a week ago or so I tested Eryone CF-PETG.... Here are the results....



The horizontal (X-axis) strength was pretty flat regardless of temperature, but at 6000 psi max. it wasn't as strong as Overture PETG with no carbon fibre!.... The vertical (Z-axis) was also significantly weaker than the plain PETG, until the temperature got to 260C, at which point it was basically equal.... I was not impressed with the Eryone....

Yesterday and today I tested the French made Kimya PETG-CF, which is pretty pricey at  $50 per 500g roll at Digitmakers.ca (it's not easy to find), which puts it in the same price range as 3DXTech's PETG-CF at $76 per 750g at the same store.... I really liked this filament, finally I am seeing the X-axis strength bump up significantly more than straight PETG, into the same range as PLA.... Here are the results....



At some point I hope to test the 3DXTech offering in CF-PETG, once my wallet recovers a bit!....

OH!.... One more thing.... I took a closeup photo of the broken Z-axis coupons at various temperature, from 220-260C.... Here they are, left to right.... Click to enlarge....



You can clearly see that at 220C the layers are distinct and separate, and the hotter you go, the more melting and fusing that occurs between the layers.... By the time you get to 260C, you really can't see the layer lines any more.... This shows why the strength in the Z-direction increases with temperature....

Bob

[rsterne]

It is results like this that make me very glad I am putting in the effort.... Today I tested the plain-jane version of Eryone PETG, in white.... It is a fairly translucent colour compared to other whites I have seen, so I am guessing a minimum amount of pigment.... Here are the results....



The results for the Carbon-Fibre filled version are in the previous post.... Note that the vertical (Z-axis) is almost identical.... However, the important thing is that the plain PETG version is as much as 30% stronger than the CF version in the horizontal (X-axis), where the carbon fibres are supposed to be "aligning with the extruded plastic" and therefore increasing the strength and stiffness.... Well, not in my testing it didn't.... In fact this everyday Eryone PETG is the strongest I have tested so far along the printed strands!....

Bob

[TorqueMaster]

Some filaments just add CF powder for the looks and it causes a possible loss of strength, rather than adding strands of CF that should/may improve the strength.  Does Eryone claim one way or the other?

[WhatUPSbox?]

There are a number of technical papers looking at PETG-CF. They tend to get into the weeds and their data is all over the place but the SEM pictures are interesting.
Example: https://www.sciencedirect.com/science/article/pii/S0142941823000296

[rsterne]

Nobody claims to use CF powder, of course.... Kimya and 3DXTech claim to use milled fibres, I don't know about Overture.... I couldn't find a claim on the Eryone website, but right now you can pre-order 30 rolls at ~half price!.... 

Apparently they use recycled CF in most of the filament....

Bob

[rsterne]

I decided to run a test on the Elegoo PLA that I bought for my 4-player Chess set.... I had a bit of extra red, and here are the results....



I have modified the coupons yet again, reducing the hole for a 5/16" pin instead of 3/8".... as I was still getting an occasional failure through the side of the eye, where my FEM analysis said the stress was high.... I think the bending at that point as the coupon stretched was the reason, and reducing the diameter of the hole straightens out the stress lines there.... The coupons are still giving the same values, but they are even more consistent, and I have not had any more fail except right where they should....

The Elegoo was typical for the Yield strength in the XY plane, and all 5 coupons stretched in the middle after they yielded.... All the Z coupons broke cleanly right in the center.... The layer adhesion was not as good as some other PLAs, but better than the wood-filled one.... That may be a function of the colour, if I have enough of the blue, white and black left over I will test them at 220C (where the red was the strongest in the horizontal) to see if colour makes a difference....

NOTE:  I tried the three other colours at 220C (the strongest temp) and found VERY little difference between them.... They were all within 10% in X and 12% in Z of the average yield strength.... In these tests, the White was the strongest in X but the weakest in Z.... If you want to be sure about your particular batch and colour of filament, testing it would be the only sure way....

Bob

[rsterne]

I received a sample of Duramic 3D TPU from a reader, and ran the tests on it.... twice!.... The first couple of coupons were OK, but as I got deeper into the (re-rolled) sample, it was wet, and the coupons became rough, pebbled, with a frosted appearance, and the strength dropped.... I dried the filament for 12 hours at 50-55C, and then it was fine.... Here are the results for he dried filament....



There was a huge increase in strength between 220C and 225C, after which it leveled off.... At 220C, the horizontal coupons failed because the layers were separating across the coupon, particularly where they were bonded to the sides of the "eye".... At 225C and above, that did not occur, even at maximum load.... The peak strength occurred both in X and Z at 230C, and all samples had a high gloss....

Boy, does this stuff stretch!.... I increased the travel rate to 20mm/min. (from 5), and the stronger horizontal coupons stretched 170mm in 8.5 minutes before snapping.... That is to over 4 times their original length!.... Elastic materials certainly do not follow the rules, as when you take into account how small the section was (about half size in width and height), when they failed the stress was about 4 times that shown in the graph above (ie approaching 30,000 psi).... Of course that is not how you do it, but it shows why things like tennis racquet strings can stand the tension that they can, by stretching....

One thing that occurred to me was the thought that this material might be usable for transfer ports.... It recovers most of its stretch, unless you way overdo it, and it's tough!.... If the OD was contained in metal (such as the recesses in a barrel and breech) it might be possible to make custom, stand-alone transfer ports from it.... Print them vertically, so that the layers are in compression, and the outer walls are contained by the recesses, and they might work really well.... Just a thought....

Bob

[rsterne]

Here are the coupons from the TPU testing.... First a horizontal (X) one at 220C (wet), before and after testing.... Note the delamination, which did not occur above 225C....



The next is one on the torture rack, again a horizontal one....



Here are all the Coupons after testing.... The left column is horizontal (wet), second column is vertical (wet), third column is horizontal dry, right column is vertical dry....



The top row (L only) is at 210C, horrible finish, virtually no bonding.... Second row at 220C, third at 225C, fourth at 230C, fifth (R only) is at 235C, and the bottom row is at 240C.... Note the rough finish and stringing on the wet vertical coupon at 240C.... That is when I decided I had to dry the filament....

Bob

[rsterne]

I got a sample pack of Spectrum (Polish) filament, with 5 different plastics, all with a carbon fibre fill.... The first one I tested was the PLA-CF10, which as the name suggests has 10% carbon fibre content.... Here are the results, plotted vs temperature.... Printed with a 0.4mm nozzle, at 0.25mm layer height and 0.40mm line width.... pretty much standard settings for CF....



This material is a bit stronger than any other PLA I have tested in the horizontal (X) direction, but quite weak in the vertical (Z) direction.... The layer to layer adhesion is not great.... I was curious about how layer height and line width affected the strength, so I did some more tests.... First of all, the horizontal (X-axis) strength does not appear to change when these are varied, total variance was within 3%.... However, the layer to layer adhesion changed quite a bit.... All the tests were done with a 0.4mm nozzle at 220C.... above which the horizontal strength drops.... Here is what happens when you change the line width.... 0.40mm is 100% of nozzle diameter, 0.48mm is 120%, and 0.56mm is 140%.... The number of walls in the test section (0.56 x 0.56mm) is also given, along with the line width, in the X-axis labels....



Note that with the 0.25mm and 0.30mm layer height there is virtually no change in the yield strength in Z as you change the wall thickness.... However, with the 0.2mm layer height, increasing the wall thickness greatly improved the strength.... With this in mind, I then tried four different layer heights, again at 220C with a 0.40mm line width, and got the following results....



It is quite obvious that as the layer height is increased above 0.20mm, the layer to layer adhesion increases a lot.... I'm guessing this is why a 60% of nozzle diameter minimum layer height is recommended for carbon fibre filled filaments.... Possible the inclusions in the matrix from the particles requires a thicker layer for optimum strength.... One wonders if the same thing applies for wood filled, silk, or matt filaments, which tend to have weaker layer adhesion?....

Bob

[TorqueMaster]

I'm curious what temperature range the vendor recommends for the PLA-CF10?
Would it make sense to do the tests at 240C, or higher, where the Z-axis strength is strongest, rather than the XY?  XY is already quite strong, the Z is the weak link.   Granted, that is very high for PLA, but if it still prints well, why not?

They suggest 60% or more layer height vs line width, yet the best results for Z strength were at 36% of your widest 0.56 line at 0.20 layer height.   Are their recommendations based on 0.40 wide lines only?   Seems like they should have a recommended minimum layer thickness value rather than percent based on your findings.

I'm also curious what happens on the tri-color chart if you go further right -- 0.64 0.72 0.80 wide?   If 0.25ish continues to be the strongest layer thickness, 0.70ish (or more) is looking good for a line width...

[rsterne]

Spectrum only recommend 190-220 for the nozzle temperature for the PLA-CF10.... On all my testing I always go 10-20 deg. higher.... Print quality is usually pretty bad at the highest temperatures I test (so much stringing I have to clean out the holes in the vertical sample to get the pins through, and rough surfaces)....

It is my understanding that the limits for layer height are 25-75% of the nozzle diameter (some sources say 80%), so with a 0.40mm nozzle that is 0.10-0.32mm.... I have never seen a maximum width listed, but it apparently depends on the outer diameter of the nozzle tip for the absolute maximum.... I agree I should test even wider widths, but I have never seen anyone publishing data at over 140-150%, so there must be a reason for that.... Have you tried it?.... I did do some at 150% (0.60mm) which printed OK, but on the bottom layer I started seeing gaps between the lines.... Increasing the flow might cure that....

The test spools are only 250g, and I ran out before I could do any more testing.... I certainly plan to push the limits on the PETG-CF filaments to increase the Z strength, I just tested one where the X was over 10,000 psi, and as you say the layer-to-layer adhesion is definitely the limit.... I am also seeing a trend that as you get a stronger X, you get a weaker Z.... Perhaps the material becomes so rigid the stress gets more and more concentrated on the inter-layer adhesion?....

Bob

[TorqueMaster]

Agreed, PLA at 240C, 20 over their recommendations, and poor print quality -- you've gone far enough.  I can't really explain why XY tensile strength must go down as Z goes up -- hopefully you will find a material that upsets that trend.

I just looked something up for grins -- structural lumber, the highest grade according to what I saw, has a tensile strength around 1650psi, and compressive of 2250psi.  Not that we'll be building houses out of this stuff, but interesting that when printed well, it's at least as strong, oftentimes many times stronger.  Kind of puts it in the layman's perspective.

I have printed at 200% width -- I was flow testing with a 0.40 nozzle, and did 0.80 wide x 0.40 high layers.  I had to print fairly slowly to not exceed my hot end's volumetric maximum -- the results were less attractive, but functional.  I did not do any strength testing.  There was a point where I could request say 12mm3/second, but it would only deliver 10mm3/s -- and of course quality suffers.  I did "game the system" once to command 12mm3/s on a slice that really only needed 10mm3/s, and it worked, but not something I'd recommend.  You're right, the line width limit would be the diameter of the flat area around the nozzle, which is usually a lot bigger than the nozzle diameter.  The smart, but inconvenient, thing to do past 150% or so of nozzle size would be to put on a larger nozzle. 

Keeping layer height at 25-75% of line width is a good practice.  I have done 0.08 on a 0.40 line/nozzle, and 0.12 on a 0.50 line/nozzle, so I've successfully been in that 20-25% zone.  I rarely go over 50% with concerns of how well it will squish together, and more recently because of knowing how overhangs will suffer.

Keep on testing Bob, this is great stuff!

[rsterne]

Carrying on with the Spectrum Carbon filament sampler, the next one is the PETG-CF10.... I started printing the coupons as it was wound on the roll, but after a few I decided it probably needed drying.... I dried it at 65C for 6 hours, and indeed it changed the results I was getting.... The strength in the X direction (horizontal) increased from 10,400 psi at 250C to nearly 14,000, which is the highest I have recorded, even stronger than CF-Nylon!.... However, the dried filament LOST strength in the Z (vertical), the layer to layer adhesion dropped from 3300 psi to only 1000 psi !!!!.... The fan setting, at 70% (recommended by Spectrum) seemed high for PETG, so I ran the tests again with the fan at 30% and got a small improvement in both directions.... Here are the results....



The filament was the strongest in the X direction at 250C, and I had lots left, so I did some tests at 250C, with the 0.25mm layer height I had been using, and also with 0.30mm layers.... The variable this time was the line width, and I only printed the vertical coupons, with the following results....



In a similar manner to the test I ran with the PLA-CF10, with the thinner layer, the strength improved as I increased the width.... but with the thicker layer, the width of the lines made little difference.... I still had some filament left, so I installed a 0.6mm nozzle (all the above were done with a 0.4mm), and using a 0.6mm line width I printed samples at various layer heights.... Here are those results....



The 0.42mm layer height was not that great a quality, and the 0.48 was pretty awful, to be honest.... The strength in the vertical (Z) direction peaked at a layer height of 0.42mm (70% of the nozzle diameter), and I can't imagine using more than about 0.4mm for anything that you care about the appearance of.... The strength in X (horizontal) peaked at a layer height of 0.30mm, which is 50% of the diameter, just like printing at 0.20mm with a 0.4mm nozzle.... The interesting thing about this test is that it pretty much disproves the "60% rule" for layer height with a Carbon Fibre filled filament.... However, this and the previous test using PLA-CF10 did show that a 0.25mm minimum layer height (60% on a 0.4mm nozzle) was a lot stronger than the "standard" 0.20mm layer height.... My conclusion, based on these two tests, is that you want a layer height of 0.24-0.32mm when using a CF filled filament.... I may refine this as I do further tests with other materials....

Bob

[rsterne]

Next in line in the Spectrum Carbon sampler is their PCTG-CF10, and no that is not a typo, it is not PETG.... This material prints at a higher temperature than PETG, and I had high hopes for an improved layer-to-layer bonding.... only to be disappointed!.... Here is the plot of Yield vs Temperature....



The Horizontal (X) strength is wonderful (almost 12,000 psi), but the Vertical (Z) strength was well under 2000 psi.... The material had been dried at 65C for 6 hours prior to testing.... I had lots of material left, so I decided to try and answer some nagging questions about whether the cross section layout mattered on vertical prints I had data from the first round with all walls in the test section, so I used the same layer thickness (0.25mm) and line width (0.40mm), and reduced the walls to 4, then 3, then 2, with the interior filled with a 100% line infill, as in the Cura slicer drawing below....



I hoped to find out if there was any difference in the way different line orientations bonded, layer-to-layer, or if it was just the cross-sectional area that mattered.... These tests were done at 270C, the upper end of the recommended temperature range for PCTG.... Here are the results: virtually a straight line....



To me, this answers the question whether you should favour walls over 100% infill when the load is only trying to separate the layers, and there is NO bending load applied to the part.... it doesn't matter.... NOTE, this is not the case when the part is bending (as in Stefan's test hooks), where the outer layers are under more stress, so of course more walls is better....

I next wanted to find out what difference line width made on Vertical (Z) samples, so I varied the line width (which changes the wall count in the test section) from 7 walls of 0.40mm down to just 4 walls but of 0.7mm.... I did this using a 0.25mm layer height, and then repeated it with a 0.30mm layer height.... I then did some Horizontal (X) coupons at a 0.3mm layer height as well.... Testing again done at 270C.... Here are the results....



Once again, we see that with the 0.25mm layer height, there is an increase in strength going from a 0.4mm wall to a 0.5mm, then not much difference going to 0.6mm.... There was another increase in layer-to-layer bonding going to a 0.7mm width (which is 175% of the nozzle diameter), this may be due to the high pressure require inside the nozzle to squeeze the filament out to that extreme width.... When retested with a 0.30mm layer height, the line was much flatter, there being little difference in the strength by increasing the line width.... We have seen this before....

I had some filament left, so I did some Horizontal (X) coupons at the 3mm layer height, again at 270C.... Interestingly, the strength decreased slightly as the line width increased from 0.40 to 0.60mm (100-150% of nozzle diameter).... The 0.25mm layer with a 0.4mm line width tested initially was slightly stronger than the same width with a 0.30mm height.... The difference was not great, and with the Z direction being so much weaker than the X, would not matter on most prints....

My conclusion on this material is that unless you need specific properties where PCTG excells, the Spectrum PETG-CF10 is a better choice (and cheaper)....

Bob

[rsterne]

It seems that the more testing I do, the more questions I have.... I guess that this early on in learning about 3D printing, that is typical.... The first testing I did with PETG-CF filaments, I used them "out of the box", as I thought they should be fine when first opened.... The results are earlier in this thread.... Now I am drying them and retesting.... Just like the Spectrum PETG-CF above, when I retested the IEMAI PETG-CF filament, the strength in X increased but the strength in Z decreased after drying at 65C for 6 hours!.... I don't understand the mechanism by which this is happening, but I need to figure out how to get back the lost strength in the layer-to-layer bonding.... So, I started trying various combinations of layer height and line width, testing at 250C, with the following results....



At the "threads" of filament get bigger, I am losing strength in the horizontal (along the threads), but gaining greatly in the vertical (layer-to-layer) strength.... In fact I managed to double that by going from a 0.25mm height with 0.40 mm width to a 0.30mm height with a 0.60mm width.... I then tried printing slower, 40mm/sec. instead of 80, and the horizontal got a bit stronger (10%), but the vertical lost 20% of its strength.... I then tried a speed in between, and turned off the fan.... I lost 10% in the X direction, but made a huge gain in the layer-to-layer bonding, and recorded 5340 psi, which is 75% of the X strength at the same settings.... The failure point showed almost complete fusing of the layers (which is great), but the sides of the square test section were bulged out and no longer straight lines.... Most of the coupon looked OK, but the smallest section, that takes the shortest time to print, was obviously getting too hot and distorting badly....

I think I have zeroed in on a "minimum" height and width for CF filled filament when using a 0.4mm nozzle, however.... There is no question that you don't want to use less than 0.25mm for a layer height, and there is a pretty decent improvement in layer-to-layer bonding by going to a 0.5mm line width (125% of the nozzle diameter).... Further gains in layer-to-layer bonding appear to be available by increasing the layer height, but at the expense of appearance of course.... If you don't mind the layer lines, you can increase the height to 0.3mm and the width to 0.6mm.... At that size, you could (should?) be using a 0.6mm nozzle.... and then you can go even larger, at the expense of pretty terrible looking prints.... For me, the practical maximum would be a 0.36mm layer height, with about a 0.72mm width.... Thicker than that looks pretty ugly, IMO....

Bob