Tensile Strength Testing

[rsterne]

OK, curiousity got the better of me, and I did one more test with a cone-shaped test piece, 50mm at the base and at 11mm up, tapering to 24mm at the top (25mm high).... It was PETG with only 3 threads, and took 340 lbf. to break.... and it broke in a way I did not expect!....



I should have known that is what would happen, comparing it to the previous test with 3 threads, which broke at 240 lbf.... They both broke in a shallow cone, with the deepest part at the bottom of the test threads, right where the tension is the greatest, and tapering up towards the end where the bolt was inserted.... In the case of the 23mm diameter cylinder (on the right), that left a tapered "washer".... In the 50mm diameter test cone, it ripped out a tapered wafer, where the OD was between 30-32mm in diameter.... The original sample, tested above, was limited by the OD, but the cone was limited only by the depth from the bottom of the threads to the end.... Having those 3 threads set down further from the end (or having more threads, so that the bottom thread was further from the end) would increase the difficulty in tearing that tapered piece out, and very likely make it larger in diameter, following more or less the same angle, which is roughly 20 degrees.... The ultimate strength was roughly proportional to the OD of the tearout.... (32 / 23) x 240 lbf. = 334 lbf. and the cone failed at 336 lbf.... Coincidence, or maybe a rule of thumb (for the same number of threads)?....

Following this logic, a 2" diameter solid (100%) print of PETG would support about 6 threads before the cone torn out would reach the OD and the strength then start dropping because the tear-out becomes affected by the OD.... That should happen at about 680 lbf., using double the 3 thread of this test piece.... Call it 100-110 lbf. per thread for PETG, but limited by the OD of the printed part splitting along that roughly 20 degree angle....

All of this has too many variables to be relied on for safe design work.... but it is VERY interesting to know that the limiting factor to thread strength is for the most part the OD of the part the threads are in.... If the part has Infill, the the diameter to where the infill starts would become important....

Bob

[subscriber]

Very useful data, Bob

Thanks for all your effort.

[rsterne]

Thread testing done, so back to testing the filaments I have available.... Today it's Anycubic Silk Silver PLA....



The curves are very flat from 210-230C, and while the X-direction strength is pretty good, the layer to layer bonding is quite poor....

Bob

[rsterne]

Next up is Overture PLA Professional in white....



In tension (X-Axis), the highest Yield measured was at 210C, but the curve was essentially flat from 210-240C.... The layer adhesion (Z-Axis) was virtually constant from 230C up to 250, which is as high as I went.... The Vertical sample was virtually identical in strength to the standard Overture PLA, and in the Horizontal direction, it was actually less.... However, all the horizontal samples stretched after the yield point, rather than breaking right away.... whereas the standard PLA broke above 240C.... The Z direction had 63% of the strength of the X direction, which is pretty good from my results so far....

Bob

[rsterne]

Now we get to the interesting stuff I built this setup to test!.... I ran some Polymaker PA612 Carbon Fibre today, and it did not disappoint, being about 50% stronger than the best PLA.... Here are the results.... BTW, I am using a layer height of 0.25mm for the CF-Nylons, as a 60% minimum of nozzle diameter (I'm using 0.4mm) is recommend for carbon fibre....



The tensile strength in the horizontal (X) direction exceeded 13,000 psi....  .... The vertical (Z) test coupon got to 5300 psi before breaking.... All samples broke cleanly, with no noticeable stretch, as you would expect with carbon fibre.... The pieces were very stiff.... The X-direction peaked in strength at 270C, while the layer adhesion was pretty flat from 260-290C.... Within those temperatures, the lowest yield occurred at 270C, but at 5080 psi, within 5% of the 5313 max, which on data from single coupons is probably within the region of being statistically equal....

Bed adhesion was a bit tricky.... I used Garolite (FR4) which is basically printed circuit board, laminated glass fibre and epoxy.... In fact, I got a piece of copper clad one-side (and used the back), I figured the copper back couldn't hurt the heat distribution on the bed.... I used 70C, with the first layer at 80C with a nozzle temperature of 270C, and had no problems once I found that combination.... I didn't try my PEI bed, but I probably should have.... I had trouble with the retraction settings, the ones that gave the greatest strength were very stringy, and when I increased the retraction distance to clean up the prints, the outer surface was no longer smooth and the strength dropped in both axes.... Strings are not that hard to remove, so for an engineering material where strength is everything, I'll put up with them until I find a solution that doesn't lose strength....

More CF filaments to come, it will be interesting to see if price corresponds to strength!.... 

Bob

[Baco]

one variable I'd like to see tested and saves a ton of time is thicker than nozzle extrusions comparing how they would scale with the layer height to width ratio when using bigger nozzles, we know that lower LH:W ratio and so bigger nozzles are way stronger but how that plays out when simply bumping up extrusions to 125%?
aside from printing faster since you print less walls the Z adhesion would improve due to this extra 25% squishing
 that's a setting I've been using for some time and doesn't affect surface quality at 125%

e.g I'm using a 0.6 nozzle and that comes out to 0.8mm extrusions by itself faster but in case strength scales up like with bigger nozzles that means you can use higher LHs too like 0.4 and that almost doubles print speed without sacrificing strength or messing with the hotend

ref:
https://www.cnckitchen.com/blog/the-influence-of-layer-height-on-the-strength-of-fdm-3d-prints

[rsterne]

I'm a bit confused.... You say you are using layer heights of 125% of the nozzle diameter (eg. a 0.5mm LH with a 0.4mm D).... Yet the text with Stefan's video (and the chart) show a significant loss of strength when the layer exceeds 50% of the diameter (eg.0.2mm with a 0.4mm nozzle).... I haven't watched the video yet, just read his text....

Bob

[TorqueMaster]

Interesting results -- the CF-Nylon excels in XY, but is barely stronger than PLA in the Z. 
At first I thought your 60% minimum of nozzle diameter was a typo, but sure enough I can't find where anyone suggests less than 50%, or less than 0.20mm layers.  60%/0.25mm minimum seems to be the consensus. One source also recommended staying under 80% of nozzle width.

This follows with Baco's great suggestion about layer height vs. linewidth (as opposed to nozzle width.)  Whenever I see these percent of nozzle width recommendations, I think they should be saying percent of line width.  Not everyone with a 0.40 nozzle prints at 0.40 linewidth.  In fact, I have read in more than one place, that the filament naturally spreads beyond the nozzle orifice, and printing lines at ~10% larger than the orifice should be closer to what naturally wants to happen.  So most should be printing at 0.44mm, not 0.40mm.  I routinely did 0.50mm lines with a 0.40mm nozzle.

And good results can be had going way past that -- I've taken a 0.40 nozzle up to 0.80mm (with 0.40mm layer height.)  No strength testing was done, but the parts were quite functional.   It's not ideal to push that much plastic through the tiny orifice, but it can be done.  Most nozzles are flat on the bottom, with a tiny hole, so as long as your linewidth doesn't exceed the width of that flat, it should get squished quite nicely. 

Printing lines thinner than the nozzle size can be done to some degree, but generally not recommended.

[Baco]

I'm a bit confused.... You say you are using layer heights of 125% of the nozzle diameter (eg. a 0.5mm LH with a 0.4mm D).... Yet the text with Stefan's video (and the chart) show a significant loss of strength when the layer exceeds 50% of the diameter (eg.0.2mm with a 0.4mm nozzle).... I haven't watched the video yet, just read his text....

Bob

sorry had a strong headache then, i see now my comment was ambiguous...
 like Torquemaster said by extrusions i meant layer width and actually 133% the nozzle width without losing visual quality

THEN assuming that strength scales proportionally with LW one could bump up LH to go with it and print functional parts in roughly half the time

edit: interesting that the overture PLA did better above the brand's recommended temp

[rsterne]

In my experience so far, most filaments reach highest strength above the temperature where the prints look nice!.... Stringing is a signficant problem at the higher temperatures....

The 3DXTech PA6-CF-G3 I am testing right now is being a real bear to get to stick to the bed.... I have tried 3 different beds (Garolite, PEI and the original black Creality bed), and the only one that I can get to work well is the black Creality bed at 110*C, with Elmers PVA stick.... They recommend a bed temp of 80-110C, and even 100C doesn't stick good enough.... They recommend printing at 240-270C, and at 260 the stringing is horrible, and the surface very rough.... Not at all pleased, considering the price of it!.... Yes, I did dry it, 16 hours in a food dehyrdator, that made no difference over as opened....

In theory, given a large enough part, there is no difference in the density regardless of how many ROUND strands you have, because smaller threads have more, but smaller, holes between them.... for the same reason (again in a large enough container), #12 lead shot weighs the same as buckshot.... However, our printed strands are not round, but flattened, and the more you spread out each strand, the fewer the gaps, and for a given layer thickness, those gaps should be the same size (assuming 100% flow, and that the slicing software is doing its job on that).... That should mean that wider strands should be more dense, have more contact area with the layers above and below, and therefore be stronger.... However, layer height is related to nozzle diameter, and it appears from Stefan's video that you don't want to go larger than 50% of the nozzle diameter.... Comparing larger nozzles (when printing the same relative to diameter height and width), should produce the same density (as hole size should be relative to nozzle size), and appearance should be the only loss.... In theory.... 

Bob

[rsterne]

Well, after a week of testing, I finally got good data for the 3DXTech PA6-CF-G3.... I did my best to dry the filament in our oven (4 hrs @ 90C, as recommended) and keep it dry during printing (dry box with 50 grams of silica gel), but I have no doubt a heated dry box with the filament feeding directly into the extruder might yield better layer adhesion.... I had a terrble time getting the print to stick to the bed, and finally ended up with a Garolite (FR-4) sheet, sanded with 220 grit sandpaper, and then a thin layer of Elmer's PVA glue stick at 70C.... The horizontal (X-axis) test was nearly as good as the previous CF-Nylon I tested, but the vertical (Z-axis) was much worse.... At 230C, the first layer of the horizontal print stayed behind on the print bed, and I dropped one of the vertical coupons and it broke when it hit the floor!.... The recommend temperature range is 240-270C, and I can sure vouch for not going lower!.... Anyways, here are the results I got....



I had read that this product gained layer adhesion as the temperature increased, at it was indeed the best at 280C, but less than half what I got with the other CF-Nylon I tested earlier, and worse than all the better PLA's and PETG's I have tested so far.... Considering the cost (in Canada it is $86 per 500 grams), I was hugely disappointed.... The print quality was very poor, especially as I increased the temperature, starting to string badly at 260C and getting worse the hotter I went.... If the trick with this filament is to feed dry filament directly into the printer, that is beyond what I can do at the present time, and it was so fussy, I'm giving up on it....

Bob

[rsterne]

I tested some Anycubic standard PLA today, in Grey.... It ended up being the strongest PLA I have tested so far....



The yield point on the Horizontal (X axis) coupon was over 9000 psi at all temperatures from 200-240C and a very flat curve.... In the Z-axis (Vertical), it peaked at nearly 6000 psi at 220C, the next best so far was the Overture PLA at 4750 psi.... At 63% of the Horizontal strength, the layer bonding of this Anycubic PLA is great performance.... I don't know if other colours are the same, of course, but the Anycubic Silk Silver I tested earlier was about 8500 in the X-axis, but only just over 2000 layer-to-layer in Z.... I presume the "Silk" additive reduces the bonding strength between layers....

Bob

[TorqueMaster]

... but the Anycubic Silk Silver I tested earlier was about 8500 in the X-axis, but only just over 2000 layer-to-layer in Z.... I presume the "Silk" additive reduces the bonding strength between layers....

Bob

Your findings agree with all the anecdotal evidence I've read -- PLA "silk" filaments may have their visual appeal, but are notoriously lacking in strength.  I'm actually surprised it did fairly well in the XY.  Matte (PETG at least) filaments seem to suffer a similar issue compared to non-matte -- the matte additives negatively affect layer adhesion. 

[Baco]

i thought from reading Stephan's tests that silver was the strongest on Z axis but maybe silk silver is something else,?

edit: I've been printing suppressors in gray

[rsterne]

I wish I could find a filament that approached or exceeded the X-Y strength on the Z axis.... That would be very interesting!.... If you have a link, please post.... 

Bob

[rsterne]

I made a new X-axis (horizontal) coupon today, that is stronger, uses less material and prints faster.... It has a very slightly narrower test section, 4.8mm instead of 5.6mm, still at 3mm thick.... I double checked the results from the Anycubic PLA above, and the numbers are virtually identical, although one coupon at 220C actually exceeded 10,000 psi when it yielded.... the previous test was ~ 9400 psi.... That is only a 6% variation, probably within the expected accuracy of the testing.... but still, 10 Ksi for PLA is pretty stout!....

Bob

[Baco]

I wish I could find a filament that approached or exceeded the X-Y strength on the Z axis.... That would be very interesting!.... If you have a link, please post.... 

Bob

I don't think Z>X/Y strength is possible on FDM i meant the strongest between colors tested


i got it backwards, silver weakest in Z but strong on X/Y + heat deformation

[rsterne]

New test coupons....



Less material, better load distribution (so hopefully no more weird failures), no stress risers adjacent to the test section, and print faster.... Loads on both the "X-Y" and "Z" coupons (over twice the test cross sectional area of the "X-Y") are in the 50ish lb. range (in the middle of what my setup can take).... Testing shows that so far the results are unchanged....

I ran the old coupons through my FreeCAD FEM Analysis, and there is a high stress area on the inside of the holes at the sides.... I assume as the coupon starts to stretch, the wall around the hole (which extends to the middle of the side part) bends and stretches on the inside.... which sometimes causes the eye to fail straight across, at 90 deg. to the load!.... The total area there is almost 4 times the test coupon area, but according to FEM, the load is the highest there, even with the extra area.... The change to removing the triangle between the hole and test section is to change the inner wall from wrapping around the hole (and bending) to carrying the force more directly to the test area.... So far, it seems to be working, although without the extra thickness on the "Z" coupon, it still breaks.... The test section is 5.6mm thick, and if the whole coupon is that thickness (like the X-Y, which is 3mm throughout), it can still break across the hole.... However, so far I have not seen the 9mm thick one break anywhere except the middle of the test section....

Having the test section curved, so the thinnest part is right at the center is forcing them to all fail near the middle of the coupon, no more failures where the (straight) test section met the larger ends.... so I know that part is much better....

Bob

[WhatUPSbox?]

Bob, do you mind posting the files for your new coupon design? I think I can pull them in my setup. I ordered some of the Anycubic grey PLA ($14 for the next few hr on Amazon). It would be interesting to compare.

[rsterne]

Here they are, separately and together.... The test area on the X-Y is 4.8mm wide by 3mm thick, and the coupon is 3mm thick throughout.... On the Z coupon it is 5.6mm wide and thick, but the coupon is 9mm thick on the ends.... They are both intended to locate on a 3/8" pin....

Bob