We own a print farm and run hundreds of prints in PETG that are subject to high daily stress loads. We found that larger nozzles with slight over-extrusion increasing the XY plane contact between rows along with optimal temp gave us the highest strength. There is an upper limit though, forcing us to design based on the print direction. Stefan, I’ll have to run some torture test samples at the larger nozzle diameter to see if it’s worth implementing for our prints. Thanks for the info!
Coming from concrete strength testing, you want the test cylinder to explode under compression. If just a part snaps off, or it cracks just by one cleavage point than the test cylinder is not homogeneous. Similar to how your parts exploded it likely means more parts of your model were experiencing stress at the same time tell the plastic itself had to give. For ridged materials optimal strength testing often ends with an explosion.
Yep That one PETG test gave me flash backs to concrete cylinder test in Material testing 101. The rebar reinforced ones would shake the building when they went off it done right!
Pardon me, but I struggle to see the similarity in material properties concerning Concrete versus Plastics. Plastics usually behave worse in strength tests when particles are introduced to the structure, whereas concrete derives its strength from the aggregates mixed in. Injection molding of PETG should preferably make the part string out when subjected to tension (as in these tests). However, since these parts aren't entirely homogenous due to the process of 3D printing rather than injecting liquid material into a mould, I'd argue there's an increase in strength due to the parts being more homogenous than traditional parts. I base this off of the parts breaking in several directions, rather than just cleanly snapping off. If they were less homogenous, they would have broken sooner
@@Arterexius Ideally in both cases neither should end up with a plasticity failure. plasticity failures are nice in that it gives you a safety factor, but fractures in multiple planes means the part took all it could handle in the current composition. Plastics can benefit from aggregates as well. Carbon fiber and glass reinforcement is valuable in plastics. This printing style is getting a lot closer to what you would expect with an injection molded failure. Concrete, steel, plastic, wood. It's all just materials and when engineered *chef's kiss* just right that's key.
@@brianhunt6943 Ah, I must have misinterpreted what you meant. I thought your argument was that it was further from what could be expected with an injection molded part, rather than closer. Apologies for that and thanks for the explanation that aggregates work well in plastics too.
@@Arterexiusconcrete paste can have strengths upwards of 400MPa and commercial strength products and commercial products around 120MPa but aggregates are suddenly the weak point. Sheared granite aggregate looks cool though
Why? When I implemented and tested it, it came out with weaker parts. I think a lot of it has to do with what kind of printer you use and quality of filament. On my bamboo p1s this made things worse on strength. But on the Bamboo the layer cohesion is already extremely good. I get the concept, it’s not new or revolutionary by any stretch, it’s rather cute and has been done many times. But on higher grade printers, this reduces layer cohesion, resulting in high stress points forming as layers don’t fully bond together. The reason this method is useful for bricks, is because those bricks are not all one solid object, they are multiple objects that are not actually connected, so you have to rely on mechanical connections like interlocking to make them strong. Good printing is not that way though, it’s not multiple separate parts, it’s one solid part, welded together basically. Interlocking like this does not get done in welding applications because it has no practical benefit or effect, and hurts weld integrity even. 3D printing is inherently a welding process. The material is heated up to a critical point that allows it to melt together with other material to form a single solid.
A 10% increase in strength with negligible changes in print times sounds pretty good. Seems like prints using this method may be more water tight as well. We need definitely need more testing. Great video!
It sounds conceptually bankrupt since MIT published well on this and of course the issue is running a heat pump and/or thermal IR panels to get the heat bay good and hot for adhesion and mixing (where my ultrasound hot end at?) Though on the other hand you don't want the heated baseplate overperforming or motors and controls overheating. Weird good news; looks like NIST (in Nature Communications ystrdy) found a way to use cellie compasses to do clutch NMR or at least some low key chemical measurement, though maybe outgassing things like PETG losing the glycol to atmosphere are too small.
3:03 There is a 2020, patent but it's a copy of a 1995, patent which had expired so this process is public domain now despite the 2020 patent which should easily be challenged in court.
I tested this method last Auguest for my Master's degree. I implemented it in PrusaSlicer with some CAD trickery. In addition to the things tested in the video I also tested a parameter that I call overlap. Overlap is the amount that the adjascent extrusion lines intersect with each other. resulting in a more solid part. My results showed that at higher overlap this new approach had no advantage over traditional printing (where higher overlap can be achieved my simply extruding more). The max interlayer strength I acheived was ~35-36Mpa in PLA for both printing methods, but the interlocking layers were printed slower because of double the amount of layer changes.
Share your results and explain why Stephan measured an increase of 10 to 14%. If you trade speed for strength this is a good thing to know. If the increase in strength is not worth the effort it is also a useful outcome.
@@chipcode5538 In this video neither method (interlock and normal) have been pushed to the limit in terms of (over)extrusion. My objectives were a bit different in that I only wanted to test the case where both have maximum extrusion. The nice thing is that I realised both can achieve essentially solid parts, but these solid parts do not have uniform properties as I hoped probably due to the thermal history. Another difference I forgot to mention was that I used ASTM 638 samples which are a different geometry and are solid instead of hollow. However I doubt this would make a huge difference. Ultimately I moved on to other projects. But I wish someone with the equipment could test this idea with more other parameters including material, geometry and other things. Maybe some other materials that are natually good at layer adhesion can benefit from completely solid parts. Another thing I would like to mention is that if you recall the transparent part video, the strength gains were better than this iirc, which kinda is in line with what I found. Overall extruding like crazy will give you higher strength IMO.
one idea i had was to mount the extruder motor to the side of the printer and have a carbon fiber hexagonal drive shaft with a Teflon gear that can slid along it making the hot end super light but still direct drive
This is amazing!!! This is the exact reason I watch this channel. Not only to learn how to 3d print but to better the process of 3d printing. I'm not an engineer but was always fascinated by how things were made and how to make something better. This is exactly what this channel offers. Thanks you Stefan!!!
The crack features on the brick printed parts strongly implies increased vertical and horizontal bonding, leading to increased homogeneity, which is why the failures were 3d and not 2d. The 45 degree oriented bonding in the brick prints is notable in this regard.
I wouldn't be suprised if printing sample horizontally wouldn't bring any better sample resistance. I would expect this feature in slicers in near future after this video.
@@Martial-MatI believe it’s good because it’s sharing the stress on bonds across both planes rather than just horizontal (layer adhesion) I could be wrong though
But if that is the case, the failure plane would still be perpendicular to the force vector. The parallel cracks would suggest that there is now a different travel path for stress, along a secondary plane of anisotropy. I bet this would be confirmed by prints made flat on the bed.
Yes but this is different than brick. A brick pattern, as he illustrates in the beginning with the blocks, is when the vertical surfaces are distributed over multiple cells above and below it. This successively doubles the surface area that is carrying any load. Instead, with this hexagonal linking, while it connects each course to more courses and of course makes it stronger, is not a brick pattern. he says there is an issue with stacking vertically, but staggering vertically maintains most of the qualities of the vertical stack and only partly links horizontally to more tracks, but arguably it already connected horizontally, and instead of just linking to the nearby track, is now linking to two nearby tracks on each side, but half as much. he should instead do what he describes in the beginning of the video and actually stagger horizontally.
I'm a bricklayer. Problem here is your stacking your "bricks" again, just on a different plane. Real bricks on running bond sit so each brick sits equally above two other bricks. The point of this is to divide the weight of the wall equally down, so the compressive load is much greater. Shear strength is not it's specialty. Your thumbnail image would be more correct if you rotated it 90⁰. In the thumbnail all I see is four stacked columns. But that would only be for compressive strength, which is all real bricks do.
Yeah comparing it to bricks doesn't work anyway since here we are testing pulling forces, not compressive, 3d prints already do fine under compression. But the thumbnail is correct if you think about layer adhesion, with normal layers, a line only adheres to it's sides and down, with this pattern I does slightly melt into and adhere to the ones diagonal below it, so the thumbnail is sorta correct in showing the contact points between each line.
Even though the direction is different, you also bind the brick, and here also the fibers. That's the point. If you stacked the bricks on top of each other, the wall would fall.
Regardless of if the layers are stacked like bricks this technique increases the amount of layers each strand is touching; in theory increasing bond strength between layers.
I don't know how this video ended up on my dash given that I don't own a 3D printer, but I still want to acknowledge your craft. This was both eloquently and succinctly explained, your theory and tests show a strong grasp of the underlying physics, the graphics are clear and informative, and whenever you presented data you did so in a clear and complete manner. Content entirely aside as I don't care about the topic, I still stuck around to relax in the presence of charts with titles, honest axes, and even error bars. Thank you!
Stephen is an excellent content producer. Explains enough but not too much and refers you to other videos for further information you want to know more.
@@wormball Wow!! Great suggestion! If you could just adjust the flow to be higher on every inner layer, it'd make a much more homogeneous piece. Print the outside layers first for a barrier to keep the over-extruded insides from oozing out, and you might have yourself a winner!
yep, a lot of us gave up on them about 5 years ago when they told us to pay AGAIN as our free life time updates didn't give us the new version, that and the advancement of free slicers just killed the hobbyist interest in it.
I actually submitted this as an idea for Bambu Studio/Orca Slicer almost a year ago. It is nice to see it in action. My use case was to make thin walled parts, such as a simple bin/cup stronger by using the offset layers. Imagine a part with a wall three lines thick. Now offset the middle line in the wall, as you have demonstrated. There is no longer a single layer line cutting through the wall. line
@@leocurious9919 The next "low" layer will be partially on top of the previous "high" layer. The mushyness of the plastic still gives you some room to work but it'll probably limit slopes a bit.
Also, couldn't you make the inner lines in walls have higher extrusion so they fill in the space more? You could print the outer lines first then with higher extrusion the inner ones.
This exists in Cura.. partially. Set infill layer height higher/lower than regular height. Select extra infill wall count. Also increase the outer/inner wall overlap. You should get at least 3 walls overlapping. That's the limitation as far as I know.. but most people dont print with more than 3 walls.
If it's a 202x era patent, it's nuked from orbit from a patent from one of the pioneers of 3D printing that expired globally in 2016. The only thing keeping us from doing this is, bluntly put, cowardice on our part. ;)
Alternately, since the middle perimeter is supported on two sides, you could try disabling part cooling for that perimeter. That should give it way more time to bond to the other two perimeters, and maybe even the next two put down on top.
@@mariusbendiksen163 interesting, I usually print my PLA at 225 for that purpose. People point out that I loose the luster of the PLA at that temperature. But I print functional parts. Not display pieces. I need it strong, not pretty.
Great idea. If combined with the techniques you described in your video about 3D printing clear "glass" (slower, hotter), I think you'll basically fill the voids between lines and the part would be very strong.
I was suprised to see any voids in the corners. If I don't get my first layer a near perfect rectangular cross section I have adhesion problems with PETG
@@TheRealPlato What kind of build plate are you using? I've heard so much PETG slander about adhesion. Some people say it adheres so well to bare glass that it can chip the build plate on removal... Other say it never adheres at all. I really don't get it, as I've had only a few adhesion issues printing PETG. And that's only on very long prints. I tend to just used glue stick or Magigoo's PC adhesion promoter on every print these days just so I can forget about the whole problem.
This channel consistently produces some of the best 3d printing content on the whole platform. Not just in form but in function as well, advancing the community farther and farther with ever video!
I print only in high strength filament. I don't bother with ABS or PETG, etc. That your brick laid model exploded (rather than cleanly detach from a layer) is ample proof that you have achieved injection level strength. A 10 to 14 percent increase in strength is very significant in functional parts. Great work Stefan!
I can't believe I haven't seen this tried before. It feels extremely obvious in hindsight, especially after seeing how fdm prints tend to split. Thanks for the great work as always, Stefan!
Impressive results! I wonder if increasing the flow rate for the middle, higher, wall layer could fill the triangular gaps between bricks more completely, like filling a trough
One thing I couldn't help but notice was that the contact patches for the brick slicing seemed smaller, even though there was a larger number of them. I can't help but wonder if this might have reduced the potential strength gain and if it's possible to increase the extrusion multiplier slightly to increase the squish of each layer and try and increase the contact patch size
You identified the main problem with this test: his layers were too high to start with. With low layers they squish into rectangles (in cross section) and as you get to high layers they look like circles. Layer bonding area drops towards zero and tensile strength falls off a cliff. In my experience, nozzle diameter to layer height ratio needs to be 2 or greater. I very much like this idea but I wish he’d done this test with a more conventional layer height (say 0.2mm).
That extrustion multiplier makes sense too since you are imparting more "void" space since printing lines naturally tend to round out. I wonder how much a small amount of extrusion multiplication would add to final strength
In PrusaSlicer’s default settings, it uses a 0.45 extrusion width on a 0.4mm nozzle. This plus a layer height of 0.2mm or lower creates perimeters that don’t have the giant gaps.
@@Tyler-wg5xh This plus extruding at the high end of the material’s temperature range so that it really squishes itself into the voids might get even closer to near perfect contact!
Just had a mighty row with a friend about this very subject. How to make sure that the circular filament fills out the square “channels” in which a slicer “thinks”. He said I was too lazy to make a better = thicker design. I said there is something fundamentally wrong if slicers can not adopt to this problem. Until then 3D printed parts will be very much weaker the those made with conventional molding technology. You’ve taken an important step in the right direction with this very experiment. Well done!
the solution would actually be quite simple. Print outer walls first, then for inner walls increase flow so it fills out the empty space neatly squished between the outer walls.
@@marcbrasse747 I suspect if you created a shell, then filled that with hot filament, which, please remember, is a non-Newtonian fluid, in that it readily expands after it leaves the nozzle, to its "natural" diameter, that you'd end up with a shell that was being pushed on, constantly, by the filler. I think this would likely result in a print that tended to change its shape over time. To warp like green wood does. It would probably be more obvious in thinner-walled prints. But, that's only my speculation.
Fascinating. This is a common source of strength at the molecular level too! Metal becomes more brittle and less likely to slip when the atoms form a tight lattice like this.
Dear Stefan, maybe you could host a competition? You already have equipment for testing and this way people would be encouraged to innovate unique slicing methods and verify them. Just an idea.
This would be super cool. Have a standard model, material, and print temperature. Then test the tensile strength to weight ratio using different gcodes people submit, along with their slicer settings and post processing scripts.
I'd love to watch something like that. Sometimes people have revolutionary ideas and don't even realize it, or don't share them thinking that someone else probably beat them to it.
I'm very late to this comment section, but hopefully you read this and give some thought. Something similar to this, that I have been thinking about for a long time is printing the even and odd perimeters at different extrusion rates. So, as an example, let's consider a five perimeter print. The outermost perimeter, let's call it perimeter 1, then perimeter 3 and subsequently perimeter 5 would be printed first. After perimeters 1, 3 and 5 are finished, perimeter 2 and 4 are printed at a higher extrusion rate, lets say 115%, calculated to fill in the star shaped gaps between layers 1, 3 and 5. Speed might also be reduced to allow the melted filament to fill said gaps. Maybe this is something that already exists en slicers, but have just never seen it. Hopefully you can give this a try, Cheers!
The essence of this channel for me! All these incremental improvements and optimizations, one of these days Stephan is going to hit upon a critical advantage that can't be ignored for its improvements.
I remember trying something like this in 2017 by taking a page out of my composites experience with alternating patterns and rotating layers. It was a pain in the butt getting the printer to run them concurrently without inducing a concentrated repetition zone where the initiation point wasn't a blob in a corner. Gyroid infill works in about the same way and doesn't require manual code adjustment, but it would be worth trying again with modern advances of the last few years on the entire layer slice.
I'd be very interested in testing this approach towards air/water tight prints. Those interlocking lines should also help with reducing gaps between extrusion lines.
PURE GENIUS! Stefan, you just figured out the best Draft settings ever! This could be the best 0.32 layer heigh way of printing fast and strong parts. 3D Printing will be better because of this video. Way to go man!
This was my conclusion after reading the post about over extruded solid parts being equally strong with and without stagger or overlap. This is faster and stronger, for rapid prototyping only.
It's better, but it's not nearly the best. The best would be to print with the fibers aligning with the flow of tension and compression and leave no hole in the center of the cross section of the model as seen in the demo. Instead of a mere 10% improvement, there could be a 20x improvement.
The use of dimensional adjustments in simplify was pretty smart! I think most people don't realize how powerful the process settings in Simplify can be. I would have never thought to use them this way. Nice video 👍🏼
True, I used Simplify 5 years ago to achieve the adaptive layer quality almost without time loss (outer perimeter is 0.05mm, inner and all other model parts - 0.3mm) using the same technique with multiple printing processes. Even 5 years later there are no slicers which can offer me this incredible functionality.
How about printing the lines of each next layer onto the seams of each previous layer? Some tweaks would have to be made to get rid of the missing half line at the outer perimeter, but the overlap/interlock would be wider than with the method you presented here.
Really interesting analysis! I've been 3D printing for several months now and I'm learning new things everyday. I had thought about how layers are stacked on top of each other and have broken a few prints by mistake, usually due to poor layer adhesion. This is a really interesting approach - thank you for sharing!
Hi Stefan, what an amazing idea and execution! One add-on I've come up with... MORE FLOW. Let me explain. AFAIK most printers (extruders) are set up to feed slightly less than 100% material. Like 98 to 99%, and that is BEFORE any manual flow rate adjustment you apply. That's because you need to compensate for the inevitable air gaps between the round edges of the "bricks". But with this staggered layering, there's going to be less gap, so you should be able to feed more and yet maintain the dimensional accuracy. And if you really want to take it to the limit, you could keep the reserved flow for the outermost perimeter and push more into all the other perimeters in between. This can make the print much stronger I believe, and I really hope you have time to try it!
And also i think it is worth to slightly increase the temperature cos the hot sausage is contacting more cold plastic, and it has to slightly melt all of this.
Awesome stuff man. Came back to this because my professor for principles of AM mentioned this exact concept. Surprised it doesn't have more support yet.
How about implementing a random number generator in the slicer, so that the thickness of a small section of any print line is different from it's neighbouring lines in all directions. Potentially introducing a lot of internal asymmetric patterns creating additional area for adhesion.
The secondary cracking looks to me like you are approaching isotropy. This is definitely a positive development. I’m excited to see what can be done with ideas like this. Varying deposition width and thickness to direct forces to more advantageous areas. Would it be possible to print a part with small-diameter z-channels, then, when the print head gets to the top of the part, instruct the nozzle to pause over each channel and inject material down into it? I’m curious if you could increase tensile strength by adding z-oriented structural members in this way.
Good idea. Bah, in reality might be difficult to achieve. To inject some molten plastic for some length in such a channel you must create a pressure all along already filled region of this channel. Thus, you need to seal somehow the channel entrance and a nozzle. Nozzle is hot. It will melt the walls and plastic will go all around the nozzle but not in the channel. However, maybe it depends on the channel length, and having such a stitches between 3-10 top layers is achievable.
@@bogdan1543 consider the possible use of a second material which has a lower melt temperature. It likely wouldn’t bond as well to the surrounding “jacket” but it might offer additional shear resistance none-the-less.
This just popped in my RUclips so I'm a bit late... I've tried this by creating a part with z-axis holes and modifying gcode to do the 'extrusion'. 1) Superheated nozzle (260 degrees for PLA instead of 230) 2) Pressed nozzle tight onto the hole (half a layer into the surface) 3) Quick extrusion with some tuning to get timing right (hot PLA was dripping from the nozzle) Managed to extrude about 10 layers deep, which is promising, but I thought same as @robblincoln2152: One should use a second nozzle with lower melting point & viscosity material. Even something like cyanoacrylate if one could prevent it blocking the nozzle.
5:33 I would love to see an impact test along the layer lines like this as well. I fee like that's where the biggest improvement might be with this method. Also, what if you could over-extrude all the inner layers of the wall to fill in those gaps? Could lead to even stronger layer adhesion and maybe more water resistant parts.
Like this idea. 7:56 - certainly introduces other unwanted results but to learn about this we need more testing. PETg should never explode like this. It’s just so difficult to quantify FDM part strength because there are soooo many variables - and these variables become clear from one part to the next. When I want strength - I print my parts one at a time, rather then layer by layer. Your brittle parts were probably the ones that cooled the most between layers. The heat of your previous layer is really important for strength. Printing one at a time not only allows for reduction in complexity- but also allows you to think more critically about the part you are printing - especially in achieving strength. Printing one at a time also reduces the chance of inconsistency in extrusion. I think we need to start breaking down the variables one by one and addressing them. An ideal job for machine learning. But to start with it we figured out… 1. Optimal previous layer temp vs nozzle temp. 2. Extrusion multiplier vs cooling Obviously the above points interact with one another. But if we start breaking it down we will learn. Then we can decide what technology should be added to make it work together. Just to add - keep going! You can find solutions if u keep going! Ur very good - keep going
They have alot of videos on similar tests. The point here wasn't maximum strength, it was to show the difference with this singular change. That being said, I would love to see what this change does when trying to achieve maximum strength of a specific part. Enclosures, reduced fan speed, hotter temps, slower print speed.
@@radioactivesdesigns3554 - yes I agree. I have followed for a good while. But the over-arching goal is to increase strength of FDM parts - which is a great goal because it’s tricky. I really liked the clear printing results. I learned from that myself - helped me print stronger parts. I feel like the clear printing parameter would be a really good place to start on achieving the holy grail of layer bonding. I would love to see more deep diving into this.
PETG is actually known for flash crystallisation under stress, which causes it to fracture and explode in all directions, forming sharp edges. I have no clue what causes it, it's a weird material. It's also quite capable of producing almost isotropic prints with no clear weak plane even under normal printing algorithm, with just the outer perimeter bead corners and model geometry causing stress risers.
@@SianaGearz - do you mean 3D printed PETg? - PETg sheet / stock is pretty consistent and ductile. But I have seen all kinds of weird results with PETg printed - sometimes really ductile - sometimes really brittle.
All of my petg parts I've made that underwent prolonged cyclic loads exploded just like that, multiple mounts and adapters I made to attach stuff to my motorcycle (nothing safety critical!) Eventually failed spontaneously and catastrophically just like that. Different filaments and geometries, never failed along a layer line either since I printed to not have significant loads in those directions. Reprinted with CF nylon and they've lastest more than 2 times as long without a single sign of wear so far.
I was just thinking about this approach a few weeks ago, thanks so much for doing this! I'd love to see a comparison across a range of temperatures, maybe with & without being annealed.
There seems to be a lot of air in the part at the 5:40 mark, would it not be beneficial to try and close those gaps? Maybe by creating perimeters 1 and 3 first and then squishing perimeter 2 in between them?
It would be super interesting to see how the brick method affects shear and bending strength. A lot of as printed lugs and bosses have that failure mode and I think the brick method could provide a big improvement there.
I've tried to do something this before but never got it to work properly. My attempts never got past the first 10-20 layers when the print was irrecoverable. I'm happy to see you actually did it!!
If some of the inner layers' height was intermittently doubled, such that the cross-section of some lines is lenticular in shape, it may produce more strength along the diagonal, reducing the catastrophic failure incidents. An example might be at 7:53, the exact frame where it breaks. It breaks nearly symmetrically along the diagonal in the middle, and the two pieces fly away in opposite directions. However, the fractures occurred at nearly the same spots mirrored vertically along the item, which happened to be diagonally along a fault line. The breaks occurred at nearly the same angle which appears proportional to the slope of a slanted failure. My thinking is that failures like that might be caused by the near homogeneity in the size of the material, just like how concrete with near homogenous aggregates can tend to be weaker due to the lack of anisotropic strength due to the accumulation of variances of the aggregates, similar to the fundamental problem of laying your bricks on top of each other. Could be wrong though, but this is just an idea.
Great video! I think a lot of people thought about this because it makes so much sense, but you actually did it. I hope in future this becomes a standard setting for slicers. To get the most out of this, I think it would be a good idea to slightly increase the pressure so the gaps get filled better, creating basically a hexagonal pattern in the section.
A potentially even better solution could be a system of channel and fill. but require 5 layer at least, so it creates something like this: nomeclature: layer x : sequence of the height of the extruded plastic layer 1: 10111 layer 2: 10101 layer 3: 13001 layer 4: 10031 layer 5: 10301 goto layer 3 this way you can create more planar decoupling, but it is a lot harder to implement. the other thing here is that the channels where the multi line is filled could suffer thermal failure an suffer some bucking (or eventually, re melting and welding better the vertical interlayers). Also, this solution only works in close-to-vertical walls. and would reduce printing time as some of the load (machine limiting factor? idk) would move from the kinematics to the extruding system.
Nice experiement! How do the results change when the samples are printed at 90º (i.e., z axis becomes x axis)? So obviously, while you'd have increased vertical strength, horizontal or sheer strength is reduced, but still an interesting idea to explore IMHO. Then there is the 45º modification that you mentioned also. For comparison, would've been interesting to see tensile tests for this as well. Furthermore, I could imagine tests with mixed filaments or part metal part PLA filaments.
Interesting, thanks for the video. I just don't understand why you test the "pure" layer-adhesion and not a shear load. The benefits should be much better there.
i feel like increasing extrusion rate in the central layer(s) may fill in the cracks way more and increase homogeneity while maintaining external layer accuracy
As a welder : what about FILLING these gaps ? what about pushing the material down between the previous layers instead of gently droping plastic and hoping for a weld ?
You can increase the flow of material in your printer settings, that way it spills out more from between the previous layer and nozzle, potentially filling gaps. But you can't really push down since the filament is liquid when printed. What would create a stronger bond is printing at a higher temperature (or reducing fan speed to like 50%) and printing slower for better temperature control. What I think this method improves is adhesion between two rows on the same layer as well as an increased surface area between consecutive layers.
Is it possible to drop the filament from a height slightly greater than the filament width? So that you can lay a filament over an uneven surface without needing to move the print head up and down over the surface as you go? If you can do that then you can weave the filament into the third dimension to effectively knit the layers together. For example, consider a spool of string which is wound with that familiar herringbone pattern. The spool turns continuously while the string winds onto it by moving back and forth across the length of the spool, following the ridge created by the previous time the string went in that direction. But the spool completes maybe four to six full revolutions in the time it takes the string to reach the other end, and so there are several points where the string crosses windings which were travelling the other way. Where it crosses a the string going in the other direction it locks it down, and it will itself by locked down when the string comes back the other way. I _believe_ this means that overall all the layers are tied together at regular intervals. So maybe like that, but rather than going around a spool, you would lay one continuous thread zig-zagging across the width of the wall, and continuing indefinitely around that wall.
This idea needs to be tested with z “wobble” this would drastically increase the surface area and should result in stronger layers. Ps a series on maximizing layer strength would be awesome.
Check out additive manufacturing of non-planar layers with variable layer height (Pelzer, hopmann 2001), they attempted something very close to this with fairly promising results.
If this gets implemented into slicers, it's something I would definitely use on occasion. Sometimes part strength is more important than printing speed.
Have you considered the option of perpendicular pouring of layers? If the first layer is filled with a spiral, and the next one with a snake. The threads between the layers will be perpendicular to each other. This can help to increase the strength of the adhesion between the layers.
Very cool. I hope this gets more traction. Could you use ironing to fix the inconsistent top layer? And slightly overextruding in the mid layers also seems like it would be helpful, maybe especially if it could be alternated off and on between alternating layers.
In trying to see if I could do this in Orca Slicer, I found that they do something similar by default. The outer wall is printed 0.42mm while the inner wall is printed with a 0.45mm thickness. This is the 0.2mm standard layer height. It also allows the setting to use percentages. There is also an extra wall every other layer option that is supposed to enhance the fusion between the wall and the infill. I know it's not the same as the brick method, but it is a step in that direction.
Found this video via this one which goes into the patterns and that they expired years ago: ruclips.net/video/9IdNA_hWiyE/видео.htmlsi=TufSmKKeFKtUUQcs Would love to see this happen
Im new to this 3d printing and thought of this idea as well, with regards to the interlocking lines of print. I may have a little suggestion or maybe a question as well. Is there a way for the parts at 150% at @4:50 time stamp, can the temperature be modified instead and have that part make the filament material hot to make it "more gooey" that it acts like a mortar that it seeps in with the gaps of the other side of walls? maybe make the 150% parts have a higher temperature and leave the rest at a hot but lower temperature to retain the shape and not make a hot mess/gooey outcome overall? or its not worth it because of time of printing when changing said temperature? (heating and cooling, etc). Its just a thought though, I'm an architecture student and currently been researching about 3d printing. I have been watching a lot form your channel and it inspires me to continue with this journey.
I have tried a similar approach printing relatively basic, flat designs in vase mode. For example a flat profile piece of say 2mm thickness, I use a 0.6mm nozzle and set my perimeter width to 1mm. The ramp effect essentially criss crosses the layers and provided your perimeter width is half the wall thickness both sides of the "vase" bond well. I've used it to produce som nice strong parts with the added bonus of no seam lines..
couple opinions of mine: 1. great fricking video! 2. great fucking idea 3. In metal-based material science the idea of a crack path is used to analyze the short and long-term strengths of a material. There, one wants to "force a crack to change direction as often as possible". Instead of extrusion lines, metals are composed of tiny little crystals that can deflect the crack in just the same way. I think in English they are called "grains"...Long story short: The idea of increasing part strength like this is already investigated, so research could be built on what has already been figured out for metal grains. 4. Apart from anisotropy, porosity is also a rather significant mark of quality for not only FDM but all of additive manufacturing (except for SLA i guess). The volume the molten polymer has to fill is "framed to a lesser degree" in your approach. This should impact solidification and therefore porosity. IMO these tests should include density measurements + cross section images of the actual part to give some insight into this. That might explain the PETG vertical crack. 5. one last thing...: to truly give a transparent representation of this approach, you would need to disclose the number of filament lines in your test parts + the number of additional interface areas due to the high offset. I would have loved to see that. One might even set the raw crack area of both versions in relation to the part strength... sorry for my incoherent sentences. i love the idea.
I love it - your statistical variance showed on the first results showed that the weakest of your brick layer samples was stronger than the strongest of the "normal" layer samples.
As a bricklayer, block layer, stone mason; we also add steel horizontally and vertically. As concrete has great compression strength and not such great pulling strength. By adding steel to our work; ie. rebars and or wire lock; we transfer the pulling strength of steel to our masonry, and the compression strength of concrete to the steel. 🇨🇦🙂👍
I love this idea. I am also curious on temp increases for the middle layers if that would also help as well since it being in the middle shouldn't drastically effect stringing or part shape. That, or going slower for that layer so more radiant heat can help.
My work around current is place the parts in 45 degrees so the plane is not parallel to the build plate. It often results in low buildplate holding force, so I design often breakable support in the file. FDM is similar handling to wood. One direction is much stronger than the other. Be aware when you’re designing your part. You can also not change the behavior of wood.
If we would heat up a box (air-tight) to certain degree and the draw the air out, would we possibly get the gaps between the layers to get closer together and make it even stronger? 5:50
When I first got into 3D printing, I had that very thought, alternating layers for better interlocking of layers. Seeing your microscopic view show what is lacking, that being the filling of the empty areas. I see that simplify 3D does not permit individual wall flow changes. That being the case, the voids can be calculated as a percentage of theoretical solidly printed parts, then the flow can be increased as well as printing temp, to get the plastic to flow into those areas.
Very interesting result for the interlocking PETG sample. I wonder if the same sample could somehow be produced via injection molding if it would have the same secondary cracks and at what point would it fail.
Great work! Having 10% stronger parts with different slicing setup (without increasing fill) seems like a great result and I hope open source slicers implement this trick.
@cnckitchen, what about combining this offset method describe in the video to slow printing speed to increase the heat transfer to neighboring layers, hence layer adhesion? For structural components, I don't see slower printing speed being too much of a problem, but not sure if it would worth the extra printing time..
Stefan, when you analyze data, do you apply Analysis of Variance (ANOVA)? I learnt this recently in Engineering school, but if you don't know, it's one of the most important tools in statistics ever, and it's used for quantitatively determining whether the variance of your data is insignificant enough to draw a conclusion for your tests, or in other words, whether there is too much overlap between the error bars of a bar chart for the analyst to draw a conclusion. ANOVA's tedious to do by hand but I'll bet there is an excel spreadsheet out there that does it all for you already.
Correct me if I'm wrong, but I noticed that the test you conducted was tensile stress test. As an analogy you used bricks wall pattern, I'm no engineer so I might be wrong, but I think brick wall pattern is meant to increase resistance to other type of stress like compression and maybe bending. I believe you would see interesting results by testing for other types of stress. Anyway 10% increase is no joke tho, nice video.
I have an idea to maybe also help with layer adhesion...but no time or skills to implement it. How would turning off part cooling for ONLY interior layers work? Still cool the outside perimeters, but let the interior layers have more heat and time to bond.
In Simplify3D, is there an option to set a horizontal overlap of perimeters ? The main strongness is still in vertical connections and the brick connections add the observed improvement. Bigger overlap of perimeters would distribute the stress more to shear stress of the brick connections.
Did you also test UCS, or tried to pull test in another direction? For example, instead of printing them "standing", have you tried printing them "laying" from l to _, and then put in the same machine? Nice video, congrats!
Those gaps between the extrusions layers could surely be reinforced with some kind of resin and a vacuum chamber, right? I know those gaps are really tiny but I’ve seen resin impregnated into lignin using a vacuum chamber. Lignin gaps are surely smaller.
would it be possible to force more material through the nozzle in addition to this method ? This would probably result in more rough surfaces, but less air inside the sections so more adhesion and wway better contact/surface area for it to adhere.
![BLACK BAG - Official Trailer [HD] - Only in Theaters March 14](http://i.ytimg.com/vi/Du0Xp8WX_7I/mqdefault.jpg)
We own a print farm and run hundreds of prints in PETG that are subject to high daily stress loads. We found that larger nozzles with slight over-extrusion increasing the XY plane contact between rows along with optimal temp gave us the highest strength. There is an upper limit though, forcing us to design based on the print direction. Stefan, I’ll have to run some torture test samples at the larger nozzle diameter to see if it’s worth implementing for our prints. Thanks for the info!
Get back to us with your results please
I have a much smaller operation but my use-case is similar in scope, functional prints of PETG. I'm also curious for your results.
Is there a way we can follow you for updates? Lol
Very interesting. By “slight over extrusion” do you mean flow rate or line width?
Try z axis fuzzy skin, half layer height on all walls
Coming from concrete strength testing, you want the test cylinder to explode under compression. If just a part snaps off, or it cracks just by one cleavage point than the test cylinder is not homogeneous. Similar to how your parts exploded it likely means more parts of your model were experiencing stress at the same time tell the plastic itself had to give. For ridged materials optimal strength testing often ends with an explosion.
Yep That one PETG test gave me flash backs to concrete cylinder test in Material testing 101. The rebar reinforced ones would shake the building when they went off it done right!
Pardon me, but I struggle to see the similarity in material properties concerning Concrete versus Plastics. Plastics usually behave worse in strength tests when particles are introduced to the structure, whereas concrete derives its strength from the aggregates mixed in. Injection molding of PETG should preferably make the part string out when subjected to tension (as in these tests). However, since these parts aren't entirely homogenous due to the process of 3D printing rather than injecting liquid material into a mould, I'd argue there's an increase in strength due to the parts being more homogenous than traditional parts. I base this off of the parts breaking in several directions, rather than just cleanly snapping off. If they were less homogenous, they would have broken sooner
@@Arterexius Ideally in both cases neither should end up with a plasticity failure. plasticity failures are nice in that it gives you a safety factor, but fractures in multiple planes means the part took all it could handle in the current composition. Plastics can benefit from aggregates as well. Carbon fiber and glass reinforcement is valuable in plastics. This printing style is getting a lot closer to what you would expect with an injection molded failure.
Concrete, steel, plastic, wood. It's all just materials and when engineered *chef's kiss* just right that's key.
@@brianhunt6943 Ah, I must have misinterpreted what you meant. I thought your argument was that it was further from what could be expected with an injection molded part, rather than closer. Apologies for that and thanks for the explanation that aggregates work well in plastics too.
@@Arterexiusconcrete paste can have strengths upwards of 400MPa and commercial strength products and commercial products around 120MPa but aggregates are suddenly the weak point. Sheared granite aggregate looks cool though
let's hope this goes viral and the big slicers start implementing stuff like this
Yeah! We need to share this, definitely.
If the big slicers aren't watching this channel with every new video I question their commitment to the industry. @@ldeadpirate9432
after seeing how arachne caught on, i'd say give it 6 months
Why?
When I implemented and tested it, it came out with weaker parts.
I think a lot of it has to do with what kind of printer you use and quality of filament. On my bamboo p1s this made things worse on strength. But on the Bamboo the layer cohesion is already extremely good.
I get the concept, it’s not new or revolutionary by any stretch, it’s rather cute and has been done many times.
But on higher grade printers, this reduces layer cohesion, resulting in high stress points forming as layers don’t fully bond together. The reason this method is useful for bricks, is because those bricks are not all one solid object, they are multiple objects that are not actually connected, so you have to rely on mechanical connections like interlocking to make them strong.
Good printing is not that way though, it’s not multiple separate parts, it’s one solid part, welded together basically.
Interlocking like this does not get done in welding applications because it has no practical benefit or effect, and hurts weld integrity even. 3D printing is inherently a welding process. The material is heated up to a critical point that allows it to melt together with other material to form a single solid.
maybe they never implemented it, because it is not worth it?!
A 10% increase in strength with negligible changes in print times sounds pretty good. Seems like prints using this method may be more water tight as well. We need definitely need more testing. Great video!
I suspect the boosts would be even more significant if you tested for twisting or bending.
It sounds conceptually bankrupt since MIT published well on this and of course the issue is running a heat pump and/or thermal IR panels to get the heat bay good and hot for adhesion and mixing (where my ultrasound hot end at?) Though on the other hand you don't want the heated baseplate overperforming or motors and controls overheating.
Weird good news; looks like NIST (in Nature Communications ystrdy) found a way to use cellie compasses to do clutch NMR or at least some low key chemical measurement, though maybe outgassing things like PETG losing the glycol to atmosphere are too small.
@@Cineenvenordquist Even if true that MIT published this, it's not "conceptually bankrupt" to have the same idea.
What about sandpaper thing that scratches up slightly the top of printed piece for better connection like in pottery?
@@uncertaintytoworldpeace3650Maybe a micro jagged edge on the nozzle tip to get a similar effect with greater bonding surface
3:03 There is a 2020, patent but it's a copy of a 1995, patent which had expired so this process is public domain now despite the 2020 patent which should easily be challenged in court.
I tested this method last Auguest for my Master's degree. I implemented it in PrusaSlicer with some CAD trickery. In addition to the things tested in the video I also tested a parameter that I call overlap. Overlap is the amount that the adjascent extrusion lines intersect with each other. resulting in a more solid part. My results showed that at higher overlap this new approach had no advantage over traditional printing (where higher overlap can be achieved my simply extruding more). The max interlayer strength I acheived was ~35-36Mpa in PLA for both printing methods, but the interlocking layers were printed slower because of double the amount of layer changes.
Share your results and explain why Stephan measured an increase of 10 to 14%. If you trade speed for strength this is a good thing to know. If the increase in strength is not worth the effort it is also a useful outcome.
@@chipcode5538 In this video neither method (interlock and normal) have been pushed to the limit in terms of (over)extrusion. My objectives were a bit different in that I only wanted to test the case where both have maximum extrusion. The nice thing is that I realised both can achieve essentially solid parts, but these solid parts do not have uniform properties as I hoped probably due to the thermal history.
Another difference I forgot to mention was that I used ASTM 638 samples which are a different geometry and are solid instead of hollow. However I doubt this would make a huge difference.
Ultimately I moved on to other projects. But I wish someone with the equipment could test this idea with more other parameters including material, geometry and other things. Maybe some other materials that are natually good at layer adhesion can benefit from completely solid parts.
Another thing I would like to mention is that if you recall the transparent part video, the strength gains were better than this iirc, which kinda is in line with what I found. Overall extruding like crazy will give you higher strength IMO.
I hope your thesis didn’t result in a company patenting anything - otherwise we will just see it in our lives in three decades 😂
one idea i had was to mount the extruder motor to the side of the printer and have a carbon fiber hexagonal drive shaft with a Teflon gear that can slid along it making the hot end super light but still direct drive
@@GeekDetourdepending where it was published it's probably already been sold
This is amazing!!! This is the exact reason I watch this channel. Not only to learn how to 3d print but to better the process of 3d printing. I'm not an engineer but was always fascinated by how things were made and how to make something better. This is exactly what this channel offers. Thanks you Stefan!!!
The crack features on the brick printed parts strongly implies increased vertical and horizontal bonding, leading to increased homogeneity, which is why the failures were 3d and not 2d. The 45 degree oriented bonding in the brick prints is notable in this regard.
I wouldn't be suprised if printing sample horizontally wouldn't bring any better sample resistance.
I would expect this feature in slicers in near future after this video.
Is that a good thing or not?
@@Martial-MatI believe it’s good because it’s sharing the stress on bonds across both planes rather than just horizontal (layer adhesion)
I could be wrong though
But if that is the case, the failure plane would still be perpendicular to the force vector. The parallel cracks would suggest that there is now a different travel path for stress, along a secondary plane of anisotropy.
I bet this would be confirmed by prints made flat on the bed.
Yes but this is different than brick. A brick pattern, as he illustrates in the beginning with the blocks, is when the vertical surfaces are distributed over multiple cells above and below it. This successively doubles the surface area that is carrying any load. Instead, with this hexagonal linking, while it connects each course to more courses and of course makes it stronger, is not a brick pattern. he says there is an issue with stacking vertically, but staggering vertically maintains most of the qualities of the vertical stack and only partly links horizontally to more tracks, but arguably it already connected horizontally, and instead of just linking to the nearby track, is now linking to two nearby tracks on each side, but half as much. he should instead do what he describes in the beginning of the video and actually stagger horizontally.
I'm a bricklayer. Problem here is your stacking your "bricks" again, just on a different plane. Real bricks on running bond sit so each brick sits equally above two other bricks. The point of this is to divide the weight of the wall equally down, so the compressive load is much greater. Shear strength is not it's specialty. Your thumbnail image would be more correct if you rotated it 90⁰. In the thumbnail all I see is four stacked columns. But that would only be for compressive strength, which is all real bricks do.
This was my thought but I’m not a bricklayer and didn’t know how to explain it.
Yeah comparing it to bricks doesn't work anyway since here we are testing pulling forces, not compressive, 3d prints already do fine under compression.
But the thumbnail is correct if you think about layer adhesion, with normal layers, a line only adheres to it's sides and down, with this pattern I does slightly melt into and adhere to the ones diagonal below it, so the thumbnail is sorta correct in showing the contact points between each line.
I thought the gain in interweaving layers would be in lateral shear, not in the way he tested.
Even though the direction is different, you also bind the brick, and here also the fibers. That's the point. If you stacked the bricks on top of each other, the wall would fall.
Regardless of if the layers are stacked like bricks this technique increases the amount of layers each strand is touching; in theory increasing bond strength between layers.
I don't know how this video ended up on my dash given that I don't own a 3D printer, but I still want to acknowledge your craft. This was both eloquently and succinctly explained, your theory and tests show a strong grasp of the underlying physics, the graphics are clear and informative, and whenever you presented data you did so in a clear and complete manner. Content entirely aside as I don't care about the topic, I still stuck around to relax in the presence of charts with titles, honest axes, and even error bars. Thank you!
Stephen is an excellent content producer. Explains enough but not too much and refers you to other videos for further information you want to know more.
This is indeed a low-cost but effective improvement plan. Thank you for promoting this idea!
Thanks! Appreciate it.
@@CNCKitchen i think you should try slightly increasing flow and temperature, the print may be even stronger
@@wormball Wow!! Great suggestion! If you could just adjust the flow to be higher on every inner layer, it'd make a much more homogeneous piece. Print the outside layers first for a barrier to keep the over-extruded insides from oozing out, and you might have yourself a winner!
can someone translate this to dollars please, i’ll match it
Always amazes me how the most basic stuff gets neglected for so long. A child could've identified the problem, given enough information, lol.
"Simplify3D" That's a name I haven't heard in a long, long time
yep, a lot of us gave up on them about 5 years ago when they told us to pay AGAIN as our free life time updates didn't give us the new version, that and the advancement of free slicers just killed the hobbyist interest in it.
Haven't used it in 5-6 years
I use it when I want better support removal. It still works best in that application.
Still using it for non-complex prints (which are in the majority), best and easiest to use interface. I did pony up the discounted update fee gladly.
It's an older code, sir, but it checks out.
I actually submitted this as an idea for Bambu Studio/Orca Slicer almost a year ago. It is nice to see it in action. My use case was to make thin walled parts, such as a simple bin/cup stronger by using the offset layers. Imagine a part with a wall three lines thick. Now offset the middle line in the wall, as you have demonstrated. There is no longer a single layer line cutting through the wall. line
Indeed, it might not be relevantly stronger in tension, but shearing should be much better.
Problem is, with slopes, it doesn't work anymore.
@@supermerill What do you mean? How does it not work anymore?
@@leocurious9919 The next "low" layer will be partially on top of the previous "high" layer. The mushyness of the plastic still gives you some room to work but it'll probably limit slopes a bit.
Also, couldn't you make the inner lines in walls have higher extrusion so they fill in the space more? You could print the outer lines first then with higher extrusion the inner ones.
This exists in Cura.. partially. Set infill layer height higher/lower than regular height. Select extra infill wall count. Also increase the outer/inner wall overlap. You should get at least 3 walls overlapping. That's the limitation as far as I know.. but most people dont print with more than 3 walls.
Geek Detour made a video about a patent that you were searching.
If it's a 202x era patent, it's nuked from orbit from a patent from one of the pioneers of 3D printing that expired globally in 2016. The only thing keeping us from doing this is, bluntly put, cowardice on our part. ;)
Increasing extrusion temp and enclosing the printer might create a more bonded structure using this pattern. Really great work!
I agree, it would probably make it sensible to build a heated chamber for low melting temp polymers like PLA and PETG
Alternately, since the middle perimeter is supported on two sides, you could try disabling part cooling for that perimeter. That should give it way more time to bond to the other two perimeters, and maybe even the next two put down on top.
Maybe also a higher print temp so it’s more likely to still be at a “receptive” temp when the adjacent line is laid down.
@@mariusbendiksen163 interesting, I usually print my PLA at 225 for that purpose. People point out that I loose the luster of the PLA at that temperature. But I print functional parts. Not display pieces. I need it strong, not pretty.
@@oasntet That's a neat idea, and I'll bet he could test it in Simpify3D.
Great idea. If combined with the techniques you described in your video about 3D printing clear "glass" (slower, hotter), I think you'll basically fill the voids between lines and the part would be very strong.
I was suprised to see any voids in the corners. If I don't get my first layer a near perfect rectangular cross section I have adhesion problems with PETG
@@TheRealPlato What kind of build plate are you using? I've heard so much PETG slander about adhesion. Some people say it adheres so well to bare glass that it can chip the build plate on removal... Other say it never adheres at all. I really don't get it, as I've had only a few adhesion issues printing PETG. And that's only on very long prints. I tend to just used glue stick or Magigoo's PC adhesion promoter on every print these days just so I can forget about the whole problem.
This channel consistently produces some of the best 3d printing content on the whole platform. Not just in form but in function as well, advancing the community farther and farther with ever video!
I print only in high strength filament. I don't bother with ABS or PETG, etc. That your brick laid model exploded (rather than cleanly detach from a layer) is ample proof that you have achieved injection level strength. A 10 to 14 percent increase in strength is very significant in functional parts. Great work Stefan!
I can't believe I haven't seen this tried before. It feels extremely obvious in hindsight, especially after seeing how fdm prints tend to split.
Thanks for the great work as always, Stefan!
Impressive results! I wonder if increasing the flow rate for the middle, higher, wall layer could fill the triangular gaps between bricks more completely, like filling a trough
This is something really worth looking at!
I was wondering the same thing.
Same
Like mortar to the bricks.
Maybe even a different, low viscosity filament in between.
One thing I couldn't help but notice was that the contact patches for the brick slicing seemed smaller, even though there was a larger number of them. I can't help but wonder if this might have reduced the potential strength gain and if it's possible to increase the extrusion multiplier slightly to increase the squish of each layer and try and increase the contact patch size
You identified the main problem with this test: his layers were too high to start with. With low layers they squish into rectangles (in cross section) and as you get to high layers they look like circles. Layer bonding area drops towards zero and tensile strength falls off a cliff. In my experience, nozzle diameter to layer height ratio needs to be 2 or greater. I very much like this idea but I wish he’d done this test with a more conventional layer height (say 0.2mm).
That extrustion multiplier makes sense too since you are imparting more "void" space since printing lines naturally tend to round out. I wonder how much a small amount of extrusion multiplication would add to final strength
In PrusaSlicer’s default settings, it uses a 0.45 extrusion width on a 0.4mm nozzle. This plus a layer height of 0.2mm or lower creates perimeters that don’t have the giant gaps.
@@Tyler-wg5xh This plus extruding at the high end of the material’s temperature range so that it really squishes itself into the voids might get even closer to near perfect contact!
@@joshcarter-com right but we could increase that for taller layers to make faster prints that take better advantage of this too right?
I'm not an engineer, but the science behind this is fascinating! The potential for improving 3D printing processes is incredible. Great work, Stefan!
you're a bot, pushing crypto scams. everyone should report
Just had a mighty row with a friend about this very subject. How to make sure that the circular filament fills out the square “channels” in which a slicer “thinks”. He said I was too lazy to make a better = thicker design. I said there is something fundamentally wrong if slicers can not adopt to this problem. Until then 3D printed parts will be very much weaker the those made with conventional molding technology. You’ve taken an important step in the right direction with this very experiment. Well done!
the solution would actually be quite simple. Print outer walls first, then for inner walls increase flow so it fills out the empty space neatly squished between the outer walls.
@@Janovich In theory? Yes. In practice?
@@marcbrasse747 In practice, I think it'd tend to warp if that's the only change you made.
@@S_Roach Your remark is a bit cryptic. Please explain what you mean in more detail. 🙂
@@marcbrasse747 I suspect if you created a shell, then filled that with hot filament, which, please remember, is a non-Newtonian fluid, in that it readily expands after it leaves the nozzle, to its "natural" diameter, that you'd end up with a shell that was being pushed on, constantly, by the filler. I think this would likely result in a print that tended to change its shape over time. To warp like green wood does.
It would probably be more obvious in thinner-walled prints.
But, that's only my speculation.
The hexagon truly is the bestagon...
i think highly of the hexagon too but i think i like them by chance. why do you like hexagons?
Fascinating. This is a common source of strength at the molecular level too! Metal becomes more brittle and less likely to slip when the atoms form a tight lattice like this.
Dear Stefan, maybe you could host a competition? You already have equipment for testing and this way people would be encouraged to innovate unique slicing methods and verify them. Just an idea.
This would be super cool. Have a standard model, material, and print temperature. Then test the tensile strength to weight ratio using different gcodes people submit, along with their slicer settings and post processing scripts.
I'd love to watch something like that. Sometimes people have revolutionary ideas and don't even realize it, or don't share them thinking that someone else probably beat them to it.
I vote for this idea.
He did something like this a couple years back except for designing a hook to submit to a competition held by some plastic manufacturer.
This is brilliant. Commenting to bump it up.
This is one of the many reasons I love this channel. Not just how and why to 3d print, but amazing explanation of what 3d printing can become.
People like you move things forward. Thank you.
I'm very late to this comment section, but hopefully you read this and give some thought. Something similar to this, that I have been thinking about for a long time is printing the even and odd perimeters at different extrusion rates. So, as an example, let's consider a five perimeter print. The outermost perimeter, let's call it perimeter 1, then perimeter 3 and subsequently perimeter 5 would be printed first. After perimeters 1, 3 and 5 are finished, perimeter 2 and 4 are printed at a higher extrusion rate, lets say 115%, calculated to fill in the star shaped gaps between layers 1, 3 and 5. Speed might also be reduced to allow the melted filament to fill said gaps. Maybe this is something that already exists en slicers, but have just never seen it. Hopefully you can give this a try, Cheers!
The essence of this channel for me!
All these incremental improvements and optimizations, one of these days Stephan is going to hit upon a critical advantage that can't be ignored for its improvements.
I remember trying something like this in 2017 by taking a page out of my composites experience with alternating patterns and rotating layers. It was a pain in the butt getting the printer to run them concurrently without inducing a concentrated repetition zone where the initiation point wasn't a blob in a corner. Gyroid infill works in about the same way and doesn't require manual code adjustment, but it would be worth trying again with modern advances of the last few years on the entire layer slice.
Super interesting! definitely needs to be investigated further and hopefully get something like this implemented natively in a slicer
I've only ever resin printed and I'm STUNNED!!! that you are the first guy thinking of that in 2024 no less... That's crazy! Gut gemacht!
I love that you keep pushing fdm printing forward with your enthusiasm and your testing good job
I'd be very interested in testing this approach towards air/water tight prints. Those interlocking lines should also help with reducing gaps between extrusion lines.
PURE GENIUS! Stefan, you just figured out the best Draft settings ever! This could be the best 0.32 layer heigh way of printing fast and strong parts. 3D Printing will be better because of this video. Way to go man!
This was my conclusion after reading the post about over extruded solid parts being equally strong with and without stagger or overlap.
This is faster and stronger, for rapid prototyping only.
It's better, but it's not nearly the best. The best would be to print with the fibers aligning with the flow of tension and compression and leave no hole in the center of the cross section of the model as seen in the demo. Instead of a mere 10% improvement, there could be a 20x improvement.
The use of dimensional adjustments in simplify was pretty smart! I think most people don't realize how powerful the process settings in Simplify can be. I would have never thought to use them this way. Nice video 👍🏼
True, I used Simplify 5 years ago to achieve the adaptive layer quality almost without time loss (outer perimeter is 0.05mm, inner and all other model parts - 0.3mm) using the same technique with multiple printing processes. Even 5 years later there are no slicers which can offer me this incredible functionality.
How about printing the lines of each next layer onto the seams of each previous layer? Some tweaks would have to be made to get rid of the missing half line at the outer perimeter, but the overlap/interlock would be wider than with the method you presented here.
Really interesting analysis! I've been 3D printing for several months now and I'm learning new things everyday. I had thought about how layers are stacked on top of each other and have broken a few prints by mistake, usually due to poor layer adhesion. This is a really interesting approach - thank you for sharing!
Hi Stefan, what an amazing idea and execution! One add-on I've come up with... MORE FLOW.
Let me explain. AFAIK most printers (extruders) are set up to feed slightly less than 100% material. Like 98 to 99%, and that is BEFORE any manual flow rate adjustment you apply. That's because you need to compensate for the inevitable air gaps between the round edges of the "bricks".
But with this staggered layering, there's going to be less gap, so you should be able to feed more and yet maintain the dimensional accuracy. And if you really want to take it to the limit, you could keep the reserved flow for the outermost perimeter and push more into all the other perimeters in between.
This can make the print much stronger I believe, and I really hope you have time to try it!
And also i think it is worth to slightly increase the temperature cos the hot sausage is contacting more cold plastic, and it has to slightly melt all of this.
@@wormball Great idea, totally makes sense!
10% -14% suggests more adhesion than I initially thought. Neat. Great work as usual.
Neat
Great Idea! You should also try shifting extruder lines horizontally just like bricklayers do. Neat S3D trick!
Awesome stuff man. Came back to this because my professor for principles of AM mentioned this exact concept. Surprised it doesn't have more support yet.
How about implementing a random number generator in the slicer, so that the thickness of a small section of any print line is different from it's neighbouring lines in all directions.
Potentially introducing a lot of internal asymmetric patterns creating additional area for adhesion.
The secondary cracking looks to me like you are approaching isotropy. This is definitely a positive development. I’m excited to see what can be done with ideas like this. Varying deposition width and thickness to direct forces to more advantageous areas.
Would it be possible to print a part with small-diameter z-channels, then, when the print head gets to the top of the part, instruct the nozzle to pause over each channel and inject material down into it? I’m curious if you could increase tensile strength by adding z-oriented structural members in this way.
This! I just left a comment about this haha.
Dang little injection molding channels would be incredible!
Good idea. Bah, in reality might be difficult to achieve. To inject some molten plastic for some length in such a channel you must create a pressure all along already filled region of this channel. Thus, you need to seal somehow the channel entrance and a nozzle. Nozzle is hot. It will melt the walls and plastic will go all around the nozzle but not in the channel.
However, maybe it depends on the channel length, and having such a stitches between 3-10 top layers is achievable.
@@bogdan1543 consider the possible use of a second material which has a lower melt temperature. It likely wouldn’t bond as well to the surrounding “jacket” but it might offer additional shear resistance none-the-less.
This just popped in my RUclips so I'm a bit late... I've tried this by creating a part with z-axis holes and modifying gcode to do the 'extrusion'.
1) Superheated nozzle (260 degrees for PLA instead of 230)
2) Pressed nozzle tight onto the hole (half a layer into the surface)
3) Quick extrusion with some tuning to get timing right (hot PLA was dripping from the nozzle)
Managed to extrude about 10 layers deep, which is promising, but I thought same as @robblincoln2152: One should use a second nozzle with lower melting point & viscosity material. Even something like cyanoacrylate if one could prevent it blocking the nozzle.
5:33 I would love to see an impact test along the layer lines like this as well. I fee like that's where the biggest improvement might be with this method. Also, what if you could over-extrude all the inner layers of the wall to fill in those gaps? Could lead to even stronger layer adhesion and maybe more water resistant parts.
Like this idea. 7:56 - certainly introduces other unwanted results but to learn about this we need more testing. PETg should never explode like this.
It’s just so difficult to quantify FDM part strength because there are soooo many variables - and these variables become clear from one part to the next.
When I want strength - I print my parts one at a time, rather then layer by layer. Your brittle parts were probably the ones that cooled the most between layers. The heat of your previous layer is really important for strength.
Printing one at a time not only allows for reduction in complexity- but also allows you to think more critically about the part you are printing - especially in achieving strength.
Printing one at a time also reduces the chance of inconsistency in extrusion.
I think we need to start breaking down the variables one by one and addressing them. An ideal job for machine learning. But to start with it we figured out…
1. Optimal previous layer temp vs nozzle temp.
2. Extrusion multiplier vs cooling
Obviously the above points interact with one another. But if we start breaking it down we will learn. Then we can decide what technology should be added to make it work together.
Just to add - keep going! You can find solutions if u keep going! Ur very good - keep going
They have alot of videos on similar tests. The point here wasn't maximum strength, it was to show the difference with this singular change. That being said, I would love to see what this change does when trying to achieve maximum strength of a specific part. Enclosures, reduced fan speed, hotter temps, slower print speed.
@@radioactivesdesigns3554 - yes I agree. I have followed for a good while. But the over-arching goal is to increase strength of FDM parts - which is a great goal because it’s tricky.
I really liked the clear printing results. I learned from that myself - helped me print stronger parts.
I feel like the clear printing parameter would be a really good place to start on achieving the holy grail of layer bonding. I would love to see more deep diving into this.
PETG is actually known for flash crystallisation under stress, which causes it to fracture and explode in all directions, forming sharp edges. I have no clue what causes it, it's a weird material. It's also quite capable of producing almost isotropic prints with no clear weak plane even under normal printing algorithm, with just the outer perimeter bead corners and model geometry causing stress risers.
@@SianaGearz - do you mean 3D printed PETg? - PETg sheet / stock is pretty consistent and ductile.
But I have seen all kinds of weird results with PETg printed - sometimes really ductile - sometimes really brittle.
All of my petg parts I've made that underwent prolonged cyclic loads exploded just like that, multiple mounts and adapters I made to attach stuff to my motorcycle (nothing safety critical!) Eventually failed spontaneously and catastrophically just like that. Different filaments and geometries, never failed along a layer line either since I printed to not have significant loads in those directions.
Reprinted with CF nylon and they've lastest more than 2 times as long without a single sign of wear so far.
I was just thinking about this approach a few weeks ago, thanks so much for doing this! I'd love to see a comparison across a range of temperatures, maybe with & without being annealed.
There seems to be a lot of air in the part at the 5:40 mark, would it not be beneficial to try and close those gaps? Maybe by creating perimeters 1 and 3 first and then squishing perimeter 2 in between them?
It would be super interesting to see how the brick method affects shear and bending strength. A lot of as printed lugs and bosses have that failure mode and I think the brick method could provide a big improvement there.
Came here to say the same. I missed some shear loading testing
A true innovation. I also appreciate how extensive your testing is.
I've tried to do something this before but never got it to work properly. My attempts never got past the first 10-20 layers when the print was irrecoverable. I'm happy to see you actually did it!!
If some of the inner layers' height was intermittently doubled, such that the cross-section of some lines is lenticular in shape, it may produce more strength along the diagonal, reducing the catastrophic failure incidents. An example might be at 7:53, the exact frame where it breaks. It breaks nearly symmetrically along the diagonal in the middle, and the two pieces fly away in opposite directions. However, the fractures occurred at nearly the same spots mirrored vertically along the item, which happened to be diagonally along a fault line. The breaks occurred at nearly the same angle which appears proportional to the slope of a slanted failure. My thinking is that failures like that might be caused by the near homogeneity in the size of the material, just like how concrete with near homogenous aggregates can tend to be weaker due to the lack of anisotropic strength due to the accumulation of variances of the aggregates, similar to the fundamental problem of laying your bricks on top of each other. Could be wrong though, but this is just an idea.
Great video! I think a lot of people thought about this because it makes so much sense, but you actually did it. I hope in future this becomes a standard setting for slicers.
To get the most out of this, I think it would be a good idea to slightly increase the pressure so the gaps get filled better, creating basically a hexagonal pattern in the section.
A potentially even better solution could be a system of channel and fill. but require 5 layer at least, so it creates something like this:
nomeclature:
layer x : sequence of the height of the extruded plastic
layer 1: 10111
layer 2: 10101
layer 3: 13001
layer 4: 10031
layer 5: 10301
goto layer 3
this way you can create more planar decoupling, but it is a lot harder to implement. the other thing here is that the channels where the multi line is filled could suffer thermal failure an suffer some bucking (or eventually, re melting and welding better the vertical interlayers).
Also, this solution only works in close-to-vertical walls. and would reduce printing time as some of the load (machine limiting factor? idk) would move from the kinematics to the extruding system.
I think you just invented a whole new and important step in 3d printing's history. Incredible work Stefan!
Way to keep come up with interesting questions to examine and test! I had never considered something like this.
Nice experiement! How do the results change when the samples are printed at 90º (i.e., z axis becomes x axis)? So obviously, while you'd have increased vertical strength, horizontal or sheer strength is reduced, but still an interesting idea to explore IMHO. Then there is the 45º modification that you mentioned also. For comparison, would've been interesting to see tensile tests for this as well.
Furthermore, I could imagine tests with mixed filaments or part metal part PLA filaments.
Bro the concept is so brilliant. I'm only 23 seconds into the video but I'm so excited to see how it works out
Interesting, thanks for the video. I just don't understand why you test the "pure" layer-adhesion and not a shear load. The benefits should be much better there.
i feel like increasing extrusion rate in the central layer(s) may fill in the cracks way more and increase homogeneity while maintaining external layer accuracy
As a welder : what about FILLING these gaps ? what about pushing the material down between the previous layers instead of gently droping plastic and hoping for a weld ?
You can increase the flow of material in your printer settings, that way it spills out more from between the previous layer and nozzle, potentially filling gaps. But you can't really push down since the filament is liquid when printed. What would create a stronger bond is printing at a higher temperature (or reducing fan speed to like 50%) and printing slower for better temperature control.
What I think this method improves is adhesion between two rows on the same layer as well as an increased surface area between consecutive layers.
Вude, it's because of people like you our civilization is moving forward. Thanks especially for showing the visual graphs!
Is it possible to drop the filament from a height slightly greater than the filament width? So that you can lay a filament over an uneven surface without needing to move the print head up and down over the surface as you go? If you can do that then you can weave the filament into the third dimension to effectively knit the layers together.
For example, consider a spool of string which is wound with that familiar herringbone pattern. The spool turns continuously while the string winds onto it by moving back and forth across the length of the spool, following the ridge created by the previous time the string went in that direction. But the spool completes maybe four to six full revolutions in the time it takes the string to reach the other end, and so there are several points where the string crosses windings which were travelling the other way. Where it crosses a the string going in the other direction it locks it down, and it will itself by locked down when the string comes back the other way. I _believe_ this means that overall all the layers are tied together at regular intervals.
So maybe like that, but rather than going around a spool, you would lay one continuous thread zig-zagging across the width of the wall, and continuing indefinitely around that wall.
This idea needs to be tested with z “wobble” this would drastically increase the surface area and should result in stronger layers.
Ps a series on maximizing layer strength would be awesome.
My mind immediately goes to sinusoidal z-oscillation, but I can imagine there are at least as many variations to try as exist in infill patterns.
Yeah that would be a good idea too!
Check out additive manufacturing of non-planar layers with variable layer height (Pelzer, hopmann 2001), they attempted something very close to this with fairly promising results.
How much time did it add to the print?
Practically none.
If this gets implemented into slicers, it's something I would definitely use on occasion. Sometimes part strength is more important than printing speed.
Have you considered the option of perpendicular pouring of layers? If the first layer is filled with a spiral, and the next one with a snake. The threads between the layers will be perpendicular to each other. This can help to increase the strength of the adhesion between the layers.
Very cool. I hope this gets more traction. Could you use ironing to fix the inconsistent top layer? And slightly overextruding in the mid layers also seems like it would be helpful, maybe especially if it could be alternated off and on between alternating layers.
I really want a longer and more detailed video on this. I'm eagerly awaiting
In the US, we typically call the proper way of laying overlapping bricks a _running bond._
Also posting ruclips.net/video/9IdNA_hWiyE/видео.htmlsi=tlcxna8Lnicwjfj3 that replies to the "what about patents?" question in this video.
In trying to see if I could do this in Orca Slicer, I found that they do something similar by default. The outer wall is printed 0.42mm while the inner wall is printed with a 0.45mm thickness. This is the 0.2mm standard layer height. It also allows the setting to use percentages. There is also an extra wall every other layer option that is supposed to enhance the fusion between the wall and the infill. I know it's not the same as the brick method, but it is a step in that direction.
I'm actively doing my CS senior research about this. I finish my preliminary bibliography of existing research last week, and then just saw this.
Found this video via this one which goes into the patterns and that they expired years ago:
ruclips.net/video/9IdNA_hWiyE/видео.htmlsi=TufSmKKeFKtUUQcs
Would love to see this happen
i'm no expert by any means, but pulling those test pieces up and down is like pulling a brick wall up and down. a wall isn't meant to do that.
Im new to this 3d printing and thought of this idea as well, with regards to the interlocking lines of print. I may have a little suggestion or maybe a question as well.
Is there a way for the parts at 150% at @4:50 time stamp, can the temperature be modified instead and have that part make the filament material hot to make it "more gooey" that it acts like a mortar that it seeps in with the gaps of the other side of walls? maybe make the 150% parts have a higher temperature and leave the rest at a hot but lower temperature to retain the shape and not make a hot mess/gooey outcome overall? or its not worth it because of time of printing when changing said temperature? (heating and cooling, etc).
Its just a thought though, I'm an architecture student and currently been researching about 3d printing. I have been watching a lot form your channel and it inspires me to continue with this journey.
I have tried a similar approach printing relatively basic, flat designs in vase mode. For example a flat profile piece of say 2mm thickness, I use a 0.6mm nozzle and set my perimeter width to 1mm. The ramp effect essentially criss crosses the layers and provided your perimeter width is half the wall thickness both sides of the "vase" bond well. I've used it to produce som nice strong parts with the added bonus of no seam lines..
couple opinions of mine:
1. great fricking video!
2. great fucking idea
3. In metal-based material science the idea of a crack path is used to analyze the short and long-term strengths of a material. There, one wants to "force a crack to change direction as often as possible". Instead of extrusion lines, metals are composed of tiny little crystals that can deflect the crack in just the same way. I think in English they are called "grains"...Long story short: The idea of increasing part strength like this is already investigated, so research could be built on what has already been figured out for metal grains.
4. Apart from anisotropy, porosity is also a rather significant mark of quality for not only FDM but all of additive manufacturing (except for SLA i guess). The volume the molten polymer has to fill is "framed to a lesser degree" in your approach. This should impact solidification and therefore porosity. IMO these tests should include density measurements + cross section images of the actual part to give some insight into this. That might explain the PETG vertical crack.
5. one last thing...: to truly give a transparent representation of this approach, you would need to disclose the number of filament lines in your test parts + the number of additional interface areas due to the high offset. I would have loved to see that. One might even set the raw crack area of both versions in relation to the part strength... sorry for my incoherent sentences. i love the idea.
I love it - your statistical variance showed on the first results showed that the weakest of your brick layer samples was stronger than the strongest of the "normal" layer samples.
As a bricklayer, block layer, stone mason; we also add steel horizontally and vertically. As concrete has great compression strength and not such great pulling strength. By adding steel to our work; ie. rebars and or wire lock; we transfer the pulling strength of steel to our masonry, and the compression strength of concrete to the steel. 🇨🇦🙂👍
Totally awesome project and proof of concept! The Prusa Slicer dev team should take notes.
I love this idea. I am also curious on temp increases for the middle layers if that would also help as well since it being in the middle shouldn't drastically effect stringing or part shape. That, or going slower for that layer so more radiant heat can help.
"Jack of all trades, Master of none is always better than a Master of one."
My work around current is place the parts in 45 degrees so the plane is not parallel to the build plate. It often results in low buildplate holding force, so I design often breakable support in the file. FDM is similar handling to wood. One direction is much stronger than the other. Be aware when you’re designing your part. You can also not change the behavior of wood.
If we would heat up a box (air-tight) to certain degree and the draw the air out, would we possibly get the gaps between the layers to get closer together and make it even stronger? 5:50
When I first got into 3D printing, I had that very thought, alternating layers for better interlocking of layers. Seeing your microscopic view show what is lacking, that being the filling of the empty areas.
I see that simplify 3D does not permit individual wall flow changes. That being the case, the voids can be calculated as a percentage of theoretical solidly printed parts, then the flow can be increased as well as printing temp, to get the plastic to flow into those areas.
Very interesting result for the interlocking PETG sample. I wonder if the same sample could somehow be produced via injection molding if it would have the same secondary cracks and at what point would it fail.
Great work! Having 10% stronger parts with different slicing setup (without increasing fill) seems like a great result and I hope open source slicers implement this trick.
@cnckitchen, what about combining this offset method describe in the video to slow printing speed to increase the heat transfer to neighboring layers, hence layer adhesion?
For structural components, I don't see slower printing speed being too much of a problem, but not sure if it would worth the extra printing time..
Stefan, when you analyze data, do you apply Analysis of Variance (ANOVA)? I learnt this recently in Engineering school, but if you don't know, it's one of the most important tools in statistics ever, and it's used for quantitatively determining whether the variance of your data is insignificant enough to draw a conclusion for your tests, or in other words, whether there is too much overlap between the error bars of a bar chart for the analyst to draw a conclusion. ANOVA's tedious to do by hand but I'll bet there is an excel spreadsheet out there that does it all for you already.
Around 6:40 you mentioned finite element analysis. Do you have a video explaining that?
Correct me if I'm wrong, but I noticed that the test you conducted was tensile stress test. As an analogy you used bricks wall pattern, I'm no engineer so I might be wrong, but I think brick wall pattern is meant to increase resistance to other type of stress like compression and maybe bending. I believe you would see interesting results by testing for other types of stress. Anyway 10% increase is no joke tho, nice video.
I have an idea to maybe also help with layer adhesion...but no time or skills to implement it. How would turning off part cooling for ONLY interior layers work? Still cool the outside perimeters, but let the interior layers have more heat and time to bond.
In Simplify3D, is there an option to set a horizontal overlap of perimeters ? The main strongness is still in vertical connections and the brick connections add the observed improvement. Bigger overlap of perimeters would distribute the stress more to shear stress of the brick connections.
Did you also test UCS, or tried to pull test in another direction? For example, instead of printing them "standing", have you tried printing them "laying" from l to _, and then put in the same machine?
Nice video, congrats!
Those gaps between the extrusions layers could surely be reinforced with some kind of resin and a vacuum chamber, right?
I know those gaps are really tiny but I’ve seen resin impregnated into lignin using a vacuum chamber. Lignin gaps are surely smaller.
would it be possible to force more material through the nozzle in addition to this method ? This would probably result in more rough surfaces, but less air inside the sections so more adhesion and wway better contact/surface area for it to adhere.
@4:00 used to use them then they changed to subscription and the products I bought were no longer usable ....