Stronger FDM 3D prints, using "brick layering"

[subscriber]

[rsterne]

Haven't watched the videos yet, but it seems to me all that is needed is the right slicer program.... Not a new idea, but I don't know of a slicer that will do it yet.... Maybe in the videos?....

Those links didn't work, maybe try these.....

youtube.com/watch?v=5hGm6cubFVs

youtube.com/watch?v=9IdNA_hWiyE

So this gem is tied up in Patents?.... 

Bob






Bob

[subscriber]

Bob,

Your links work.  Mine display the video thumbnails in my Chrome browser.  Clicking the image runs the video.

Anyway, it seems that brick layering is only 10% stronger than conventional FDM filament layering.  Significant for those test pieces, but not massively stronger.

I think if one prints a tensile test piece at 45 degrees off vertical you induce the brick layering effect.  You might even have made and tested such test pieces.

I think that if you have proper fusion between layers, then mechanical interlocking is not required.

One disturbing aspect is that after the 1995 patent for such printing expired the USPTO issued another one for essentially the same invention.  As if the examiners were not on the ball.

[rsterne]

Some test coupons for 3D filament are printed with alternating layers at 45 degrees, just like what you see when you print infill.... IIRC, they are quite strong, but I have not tried them, the lines are printed on the X-Y plane at +- 45 deg. but then the force is applied in X or Y, so not really the same thing.... It's like pulling on a piece of plywood cut on the bias....

I assume you are talking about having the test section printed in Z, with a 30 deg. overhang on one side and a 30 deg. slope on the other side, where the layer height and thickness are equal?.... That produces a hexagonal matrix of round beads, but you end up with every other layer one bead shorter (producing a zig-zag edge).... Every bead touches 6 others (except the edges).... If you print beads that are not round, they must be laid down horizontally, and are usually wider than they are tall, like a brick (like the pics above).... The angle along the "edge" of the layers depends on the ratio of width to height, those appear to be about 2:1 and the angle is roughly 45 deg. along the contact patch.... The contact patch lines up if you go up 3 layers and over 2 bead widths.... However, if you turn that pic 45 deg. the beads are no longer horizontal, unless they are round.... Clear as mud, right?.... 

Bob

[subscriber]

Thanks, Bob

My suggestion was closer to what you described in the second paragraph.

Perhaps the metric that is more important is that when a part is overloaded in the Z axis, it fails through the material, rather than between layers.  That would make for a strong print, up to material strength. 

How to achieve that is the question.

[rsterne]

I think laying down alternating round beads to form a Hexagonal pattern, with a slight amount of overextrusion (to fill the corners of the Hexagons) would be the ideal....



However, I still think the failure lines would the along the cracks that are the closest to perpendicular to the force applied....

Bob

[subscriber]

Bob,

This a bit like the strength of sintered iron, as a function of density of the part, compared to density of base material.  Except, then semi-random metal grains are welded to grains around them, compared to the welding of linear plastic filament layers.

I should probably go and re-read the experiments you did on the factors that affect test specimen tensile strength.  A lot of good data and pointers there.

[rsterne]

Another thought.... If you used squares instead of hexagons, alternated in printing position, and each row raised up 1/2 the height of the bead, that would tend to shear through the square beads instead of following the slopes of the hexagons.... This is how the original videos show the idea, but the beads are not extruded enough to fill the corners because of the rounded ends of each bead.... Another unknown is how difficult it is to shear through a bead compared to pulling the apart....

The strongest Z coupons I printed in whatever filament show partial tearing from one layer to the next instead of a perfect straight shear between layers.... If you start printing at a low temperature, the coupon shears straight across between the layers, and you can clearly see the gaps between the circular walls (the test area of the Z coupons looks like a target when broken).... As you raise the temperature, the lines between the rings of the target get smaller (better fill out into the corners between each perimeter (wall line).... At some point, when the temperature is just right, the sheared face of the coupon is no longer flat, the shear line jumping up and/or down from the major shear failure to one layer above and/or below.... The gaps between the printed lines are barely visible.... As you increase the temperature further, the material seems to look more "crystalline" with an appearance like sandpaper where it shears, with almost no layer lines visible, and the strength seems to decrease again.... This may be because the test section is so small (5mm diameter) that it is not cooling sufficiently between layers, and the plastic is getting overheated.... Much larger coupons, and a LOT stronger test apparatus would be necessary to prove that definitively.... It takes about 200 lbf to break the stronger Z coupons, and 300 lbf to break the stronger X coupons (which have a 3 x 5mm test section).... The force is applied at a speed of 5mm/minute....

One other thing, the strongest Z coupons had the thickest line walls, which means the fewest lines and gaps (ie the ratio of gap to line width was better).... If I used a 1mm line width, there were only 2 circles and a dot in the cross section (and only 2 circular gaps).... With a 0.5mm line width, there are 5 circles of filament (and 4 gaps)....

Bob

[subscriber]

Bob,

If you are stating that minimizing the gaps between filament beads makes parts stronger, I have to agree.  If you zoom in on the second image below, it seems to have a granular fracture, with very small gaps between filament beads:

Here is a picture of a thin walled light weight custom moderator, printed from PETG for a Huben GK1 pistol.  While the mod turned into confetti as a result of the valve sticking open, and the air tank dumping, the fracture pattern seems not to have followed the print layer lines.  So, apparently very good layer to layer adhesion.

Perhaps the failure patterns for a tensile test specimen, and a pressure vessel that saw axial and hoop stress cannot be directly compared.  The guy that printed this mod and unintentionally popped it reported that the failure was disconcerting, but that no plastic bits were projected towards him.  So, fail safe.

[Scotchmo]

I think laying down alternating round beads to form a Hexagonal pattern, with a slight amount of overextrusion (to fill the corners of the Hexagons) would be the ideal....



However, I still think the failure lines would the along the cracks that are the closest to perpendicular to the force applied....

Bob

I agree. Compression, tension, shear.

Probably no difference for compressive strength.

Small improvement in tensile strength.

I’d predict a more significant gain in shear strength.

[rsterne]

subscriber.... I don't think the primary force was trying to pull the end off (ie tensile in Z).... I think it simply "blew up" from the internal pressure.... and ripped/cracked from weak point to weak point.... The baffles reinforced the shell where they contacted it, and the initial failure looks to be the walls between baffles.... Hard to tell for sure from that photo, though....

Bob

[subscriber]

Bob,

Surely the pressure differential across the first baffle and mounting thread creates an axial force, trying to blow the mod off?  Ditto for the pressure differential across the second chamber, but the first chamber "works the hardest".

The plastic bits were all found ahead of the pistol muzzle when it blew.  There must have been a force applied to drive them in that direction.  Were they further forwards than sideways is the question?  I don't have the answer to that question.  There probably was a dispersion of bits where some travelled sideways more than forwards.

I could ask the guy who had the experience, although he may wish to keep it to himself.

[TorqueMaster]

I think using anything approaching round printed lines is getting off track.  Circles, hexes, squares, will all be close to the same when laid down in reality.   And what's worked best so far is using printed lines with a roughly 3:1 width to height ratio.  2:1 is commonly used, and 4:1 iirc is about (almost?) as good as 3:1.  This is using tensile strength along the Z axis as a the weakest direction, and the one needing the most improvement.

So I don't see throwing away the 3:1 ratio as a good idea.  The alternating bricks on alternating layers is a unique approach -- disappointing that it did not offer more improvement. 

My thinking is -- since Cura5+ has variable line width, they should use that to "brick" the layers horizontally, to mechanically bond ALL the lines better than simply stacking them:

A is traditional 3:1
B is bricking using the 3:1 lines, using "half-lines" to fill in the gaps (as a mason might do...)
C is alternating 1:3 and 1:4 layers.  There may be even better patterns, "brick width" is infinitely variable, but this was easiest to suggest.
D is using the alternating layer start heights using the strong 3:1 print lines -- probably not much more improvement than the video test got, though.

I really don't know if any of these will improve Z-axis tensile strength (well, D should if the video test did.)  But every time I see print lines stacked directly on top of the previous lines (A) I believe some overall strength improvement is to be had by using something like B or C pattern.

 

[subscriber]

Bob (TM) makes some good points about not abandoning the well established 3:1 bead width to height ratio for good Z-axis strength, just to enable fancy interlocking shapes - strategy subservient to test data.

Another question would be  about as-printed thread quality when using brick or hex shaped extrusion beads.  Perhaps the simple answer is that to print a good thread, the Z-axis resolution still needs to be 5 or 6 times the thread pitch; regardless of the shape of the bead.

[rsterne]

I agree that a 3:1 bead is better for strength.... Alternative "C" would eventually end up with the vertical gaps aligned, at the "product" of the ratios.... eg. 1:3 and 1:4 at every 9th (12th) bead.... I don't think having the vertical gaps staggered (eg. "B") is going to improve the layer bond much, as you still have straight lines between the layers, but slight overextrusion, or increased temperature, will, we know that....

You are correct that Z resolution when printing threads need to be much smaller than the thread pitch, to get a thread that prints properly....

Bob

[TorqueMaster]

All true, Bob.  That I mentioned for C -- that there are likely better patterns to use -- the slicer could calculate varying line widths that avoid most or all vertical seams.  Variable widths within a specified allowable range.

It would get very complex for irregular shaped models, so perhaps the simplicity of "C" or even "B" would suffice.

The more I think about it, the masonry comparison is a decent one.  Bricks are strong, mortar less so.  In FDM, layer lines are strong, the bonds between them less so.  I think this would mostly be a torsional strength gain, but a gain is a gain.

[subscriber]

Even though I started this thread, lamenting the fact that we are not using an available improvement for 3D printing filament laydown, the topic reminds me a little of the debate about the advantages of polygonal rifling, over conventional spline rifling.

Many people have excellent results with polygonal rifling, but to me its promotion is like claiming a hex nut with its corners rounded over is superior to an undamaged nut.   There is an improvement over the hex nut or socket.  It is called the Torx socket. 

While the above may seem disconnected from the filament bead discussion, I think that the hex bead, while seeming ideal, has the hex flanks add Z-axis stress between layers, when your "wind up" a printed part such as a tube.  This occurs because mating angular surfaces try to cam the layers apart, just like a overloaded hex nut cams open an open socket or adjustable wrench; until the wrench slips and rounds over the corners of the nut or hex socket. 

This suggests that the staggered brick bead, with vertical sides might be superior to conventional FDM printing beads.  And compared to other fancy angular styles.

Now, arguing with oneself might be a sign of being open minded.  Possibly to the extent that the mind has walked off and been lost. 

This is where reproducible test results come in.  Theory is fine; but the results of destructive testing are final.  Just because we study and deliberately vary a parameter, does not make that parameter the most important one in a given system. 

In summary, aim for dense prints with almost no gaps between printed beads.  How to achieve that is the question.  Some degree of over-extrusion, with good welding and crack filling, achieved by a sufficiently hot plastic bead, using traverse speeds that produce the best result   

I suspect that the "width to height ratio" of the common clay brick did not come about by accident.  If bricks are were shorter, the overlapping lengths would become shorter.  Too short, and overlaps may become meaningless.

[rsterne]

The proportions of bricks varies from country to country, but as an example, in England the ratio is 6 x 3 x 2.... If you use the "2" for the layer height, the width would be 150% and the length would be 300%....  That would be akin to printing with a 0.4mm nozzle at a layer height of 0.2mm and a line width (the length of the brick) of 0.6mm ie 3:1.... Our lines are continuous, so the "width" of the brick (equivalent to the thickness of a brick wall) is irrelevant.... but the 50% overlap of the vertical joints (what you see on the face of the wall, or in a cross-section of our perimeters) is obvious....

Concrete blocks (faster to build with than bricks) are typically 8 x 8 x 16, which is a 2:1 ratio of length to height (line width to layer height).... like using a larger nozzle, printing a thicker layer....

Bob

[subscriber]

Interesting

[Scotchmo]

I often use relatively thin walls and 100% infill. With crosshatched alternating layers. (XY rectilinear pattern). So I’m thinking the brick patterns are not applicable in those cases.