Testing Wall Thickness in 3D Printed Beams
Вставка
- Опубліковано 23 лют 2024
- TRY THIS FILAMENT:
amzn.to/4bMxESP
Today we test the compressive strength of 3d printed rods with variable wall thicknesses.
When designing 3d printed parts, one of the key considerations for ensuring durability is the number of wall loops. But what are the numbers behind this? In this video, we put 20mm rods under extreme compression in order to find out what the optimal wall thickness should be.
🔔 Subscribe to Tangled Testing for updates on our latest videos and discoveries in the world of 3D printing. Together, we'll unlock the full potential of 3D printed materials and ensure your projects are built on a foundation of reliable data.
About Tangled Testing
Welcome to the official UA-cam channel of Tangled Testing - your go-to source for testing and analysis of all things 3D printing. Our mission is to shed light on the true quality of 3D printing materials through various testing methods. Whether you're a seasoned 3D printing enthusiast or just starting out, our channel offers valuable insights that will empower you to make informed decisions about the materials you use. Subscribe for insights that will enhance your 3D printing projects and join us on our journey towards higher standards in 3D printing. - Наука та технологія
View the Data:
docs.google.com/spreadsheets/d/e/2PACX-1vR3WXKnLqu2sP4MOSvqhWclPFsPrqifHWVHk0GzVqYNAY7z-N-5jKEaoamf8R-ajNEWXfqXGSxdog8p/pubhtml
Try the Filament:
amzn.to/4bMxESP
For the next one, the points should be stacked on the x axis for each wall thickness, the current regression line isn't valid. I'd also love to see the curve compared to an estimated failure. I'd bet the step looks more like a sigmoid since adding thicker walls internally have a smaller diameter.
"FILAMENT TESTED" in description, would provide better context vs. "TRY THIS FILAMENT" which seems more like a random affiliate link marketing. Reading "Try THE Filament" gave it context.
Two minor suggestions. The analysis would improve if additional measurements were included to achieve statistical significance. In addition, it would be beneficial if the pictures depicted not only the mean but also the standard error of the mean.
Agree, 5 samples with the std deviations in this test would be nice.
I suspect that with the large deviation at 2mm, some of the nonlinear progression we see is deviation, not real.
With tighter deviation tests, 3 samples would be fine.
Also, I am not sure that the standard error would tell us much here, as each set of 3 samples is from a different population (wall thickness).
Standard deviation might tell us more, but with only 3 samples, you can just look at the break values and see what the deviation is.
To test 2.5mm would be a decent reaction I guess.
As you mentioned, the geometric profile definitely has an impact here. As we know, the area of a circle is exponentially related to its diameter. The more you increased wall thickness, the smaller the diameter of the rings you were adding. That means you were adding exponentially less filament with each increase in thickness, even if you were only printing a partial circle. Circumference itself is only linearly related to diameter, but printing fractions of a circle still creates a small exponential factor. The material being closer to the edge of the part as opposed to closer to the center also affects how the force applied affects it. I would think those are the main reason you don't keep seeing a large spike in strength, unlike with the infill test which distributes the added material more evenly across the whole cross section.
A = pi * r^2. It's a polynomial relationship, so that's mildly incorrect. I don't know where you got exponential from, as those take the form y=k*a^x, y=k*e^x,... Funnily enough, the area of an annulus wold still be quadratic, as A = pi*(R^2-r^2) (R is major radius, r is minor radius)
@@danielstarr2483 Fair enough! Been a long time since I had a standard math class and I always hear people say "The area of a circle grows exponentially with the radius." While more accurately it grows polynomially or quadratically. I think it should be obvious where I get exponential from since the expression has an exponent. Doesn't mean I was correct. Would you believe I've done 3D graphics programming? :P
My overall point was just that the area of each ring grows by a larger amount in proportion to its radius as opposed to in a linear fashion. For example a ring with 20mm OD and 0.4mm width has an area of 24.63mm² while a ring with 10mm OD and 0.4mm widh has an area of 12.06mm². So rather than being 50% of the area it is just under 49%. Not as drastic as I expected, to be honest.
However, if you take it in 1mm steps as he was, the 20mm OD 1mm wide step is 59.63mm² while the 19mm OD 1mm wide step is 56.55mm², so a bit over 5% less material. The final step from I'm assuming 16mmOD at 1mm wide is 47.12mm² so almost 21% less material than the outer 1mm wall section. That's a pretty significant change all things considered.
I don't think that accounts for the full decrease in added strength for each step. Also doesn't explain the drastic jump and then reduced rate of increase. As I said I think where material is inside the structure also matters. Something to do with how the force is distributed across the material and total distance from point of force and I'm sure more I'm not considering.
@802Garage
This "jump" is actually quite expected in static analysis of beams. The most useful number would be the second moment of area, which essentially gives beam stiffness in the axis of loading. After some reading through a wiki article, the second moment of area is (pi/2*(R^2-r^2)). This means that as the minor radius decreases, the second moment of area will increase quintically (on the order of x^4). What is likely happening is that the step size is too large to capture that behavior.
Also important for consideration: the beam is under shear, compressive and tensile loads in the weakest axis. The bottom of the beam is being compressed, and the print's Z axis (printer prints a layer, then moves up in the Z axis) is capable of handling that just fine. On the top side, the print is under tension, almost like something is being pulled apart. Since the layers are like a chain, the weakest layer will cause the entire test to fail.
Another factor, in addition to shear (which many prints arent susceptible to unless they are very thin) is the bending moment. This is 0 at the point of loading, and as the distance from the point of loading increases, the moment increases (this is in addition to all the other forces) This peaks exactly at the cantilever holder, which explains many of thr failures at the base of the cantilever beam.
@@danielstarr2483 Yep that all aligns with what I was thinking. I obviously coulda read up on it, but I enjoyed speculating. ;) Appreciate the info and insights!
I'd like to see a similar test done with a rectangular cross-section rather than a circle. I'm part of the Astrodroid builder community, and fully 3d printed droids are becomeing mroe and more common, but there's a lot of argument about strength and durability, with strong opinions on both ends of the argument. The strength is obviously dependent on settings, and the key points of failure tend to be the legs and shoulders. R2-D2's legs are essentially just boxes, and having some real measured data on how a box performs in this situation would be very useful.
Also, thank you for describing this in mm rather than number of perimeters. Swapping nozzle sizes keeps getting easier and it's getting really frustrating how some channels keep talking about how many perimeters they used without any information on the nozzle size. Getting this summary with everything in mm makes it useable regardless of how big one's nozzle is.
I think it would be more useful to control for the total amount of material used by varying the infill densities such that each part has the same overall weight.
Fascinating how much of a jump between 2 and 3 there is, though I usually go for number of walls instead of thickness directly 😅
From my experience, a layer height also makes a huge difference when it comes to strength. A part printed with a .8mm nozzle will be stronger than a .4 mm nozzle
Nicely done! Thanks
It'd be interesting to see the same column tested but printed on its side so that the filament carries the bending, rather than the layer interface. It'd also be good to see this same column wall thickness test done in buckling, especially to see what the strength to weight numbers look like.
Interesting test results. Would like to see test done in terms of wall print layers. With 0.4mm print head, 1mm, 2mm are not really integer multiples. (graphics seem to indicate 3 layers for 1mm, and 5 layers for 2mm, but no closeup view of actual parts being tested, after they break)
It would be nice to see a quick calliper measurement of the wall thickness. A helpful reference to any other party duplicating, but understand, that for youtube, one of the objectives being to keep videos short and concise.
BTW: noticed the different wall thickness broke at different distances from where force was applied.
Excellent, thank you, very informative
It would also be interesting to see the strength to printing orientation test like this is printed standing but then print at 15,30,45,60,75 and 90 degrees i.e flat.
Thanks for the information! I'll be putting it into action!
Great video. Looking at the test rig, you might want to change the jig holding the cylinders/rods. It could lead to varying results due to compression varying the tests.
This testing is a little too imprecise to be of value, you need more than 3 samples in some test as you got clearly too many swings. Also you really should have tested more samples between 2-4mm wall thickness as you have an unexplainable jump in load capacity. A 2.5mm and 3.5mm test would give some more data and serve as a sanity check. Honestly given the tiny data provided I'd assume a testing error in the 2mm test as it conflicts with test done by CNC Kitchen and his data is far more consistent and thus reliable.
That's really interesting! I wonder if there's a relation to how well the layers are adhering to each other at a certain critical wall thickness? Maybe there's also a relation regarding how far away the force is applied from where the point is held (ex. applying the force closer to the pivot may cause the strength jump to occur at a greater wall thickness). I'd also guess that you would see a more linear response if the part was printed with layers along the length of the tube instead of along the height of the tube.
Print the part lying and with grid infill oriented 45° to the load direction. I'll buy you a coffee if this will not show the best load to weight ratio so far...
This data is too noisy, do more samples. As is, this in inconclusive
Can u combine this 3mm wall thickness and various infill to see if changes? Maybe the is a proper way there…..
I think you have chosen the wrong best fit line. The radius to the surface area of a circle is not linear. In your case, I would expect the strength of the part to get only logarithmically larger for each linear increase in radius because the amount of material added for each parameter wall gets smaller and smaller with each increase in parameter wall. This is because the part's overall diameter is fixed and the additional parameters are being added radially inward. Therefore, each subsequent addition of the part's parameter gets smaller and smaller, i.e., the 5th layer is significantly smaller than the first layer. A liner relationship would be more like incrementing a rectangular part's top layer.
Your outliers are likely due to a few factors:
1. The force being applied to the part appears variable and inconsistent. For example, you can clearly see force applied, pause, force applied, pause, and so on. The pause time is noticeably different between each test which adds a static force component to the mix that is negatively affect the part's reliability, and therefore skewing the results depending on how long the static force is being applied.
2. There appears to be a mounting issue with the components you used in the test because you can visibly see the clamp shaking with each change in force applied (my guess is its from lever action hydraulic pump being manually actuated), but you can clearly see horizontal movement on the XY plane as the downward Z force is applied.
3. Several oscillations were observed in the graphs of force vs. time as each force was applied. It is my understanding that oscillations can have a variable and unpredictable affect on a structure's reliability. This was likely caused by both the mounting issues and the force being applied variably.
I'd like to see this same test with more stable mounting and a force applied at a steady, increasing rate. Maybe try an electric press or a large bore pneumatic piston?
Great video! Have you considered that the actual printed wall thickness might be thicker than what you specified? Since you print with 0.4 mm nozzles, the wall thickness when set to 1 mm might actually come out to 1.2 mm (or even more, if you extrude slightly more than the nozzle width, e.g. 8-10%, but that’s more nitpicky).
This might help explain the non-linear deviation with the 2 mm wall, compared to the 1 mm and 3 mm settings. The 1 mm wall would be printed with 3 lines/perimeters, 2 mm walls would be 5 perimeters, and 3 mm 8 perimeters, and this nonlinear jumping with 3, 5, 8 could help explain the similar nonlinear jump in strength.
Thinking of wall thicknesses in whole number multiples of 1 mm can help make it simpler for people to understand (especially the case for Slant 3D clients). But, if you’d like, you could redo this test using wall thicknesses in multiples of 1.2 mm, rather than 1.0 mm, to reflect the standard 0.4 mm nozzle width you use.
The gcode shows 3,6,9,12 perimeter so your point is moot.
Ah, I see.
(Though how do they slice it in what would be 1.2 mm intervals but still get 1.0 mm differences? Now that I don't understand.)
Conflict of interest
what are you talking about? they are paragons of integrity! would a dodgy operation scam the community out of 16k to fund the QA department of a new manufacturing facility?
...hang on.
Are those repeating thick layer lines intentional? If yes, why?! Why would you expect a linear relation here?
I think those are individual layers breaking off each other
I would like to see this test repeated with the part lying down instead of standing up.